if(WITH_NVCUVID)
find_cuda_helper_libs(nvcuvid)
+ if(WIN32)
+ find_cuda_helper_libs(nvcuvenc)
+ endif()
set(HAVE_NVCUVID 1)
endif()
return (x >= 0 && x < width) ? saturate_cast<D>(data[x]) : val;
}
- const int width;
- const D val;
+ int width;
+ D val;
};
template <typename D> struct BrdColConstant
return (y >= 0 && y < height) ? saturate_cast<D>(*(const T*)((const char*)data + y * step)) : val;
}
- const int height;
- const D val;
+ int height;
+ D val;
};
template <typename D> struct BrdConstant
return (x >= 0 && x < width && y >= 0 && y < height) ? saturate_cast<D>(src(y, x)) : val;
}
- const int height;
- const int width;
- const D val;
+ int height;
+ int width;
+ D val;
};
//////////////////////////////////////////////////////////////
return saturate_cast<D>(data[idx_col(x)]);
}
- const int last_col;
+ int last_col;
};
template <typename D> struct BrdColReplicate
return saturate_cast<D>(*(const T*)((const char*)data + idx_row(y) * step));
}
- const int last_row;
+ int last_row;
};
template <typename D> struct BrdReplicate
return saturate_cast<D>(src(idx_row(y), idx_col(x)));
}
- const int last_row;
- const int last_col;
+ int last_row;
+ int last_col;
};
//////////////////////////////////////////////////////////////
return saturate_cast<D>(data[idx_col(x)]);
}
- const int last_col;
+ int last_col;
};
template <typename D> struct BrdColReflect101
return saturate_cast<D>(*(const D*)((const char*)data + idx_row(y) * step));
}
- const int last_row;
+ int last_row;
};
template <typename D> struct BrdReflect101
return saturate_cast<D>(src(idx_row(y), idx_col(x)));
}
- const int last_row;
- const int last_col;
+ int last_row;
+ int last_col;
};
//////////////////////////////////////////////////////////////
return saturate_cast<D>(data[idx_col(x)]);
}
- const int last_col;
+ int last_col;
};
template <typename D> struct BrdColReflect
return saturate_cast<D>(*(const D*)((const char*)data + idx_row(y) * step));
}
- const int last_row;
+ int last_row;
};
template <typename D> struct BrdReflect
return saturate_cast<D>(src(idx_row(y), idx_col(x)));
}
- const int last_row;
- const int last_col;
+ int last_row;
+ int last_col;
};
//////////////////////////////////////////////////////////////
return saturate_cast<D>(data[idx_col(x)]);
}
- const int width;
+ int width;
};
template <typename D> struct BrdColWrap
return saturate_cast<D>(*(const D*)((const char*)data + idx_row(y) * step));
}
- const int height;
+ int height;
};
template <typename D> struct BrdWrap
return saturate_cast<D>(src(idx_row(y), idx_col(x)));
}
- const int height;
- const int width;
+ int height;
+ int width;
};
//////////////////////////////////////////////////////////////
return b.at(y, x, ptr);
}
- const Ptr2D ptr;
- const B b;
+ Ptr2D ptr;
+ B b;
};
// under win32 there is some bug with templated types that passed as kernel parameters
return (x >= 0 && x < width && y >= 0 && y < height) ? saturate_cast<D>(src(y, x)) : val;
}
- const Ptr2D src;
- const int height;
- const int width;
- const D val;
+ Ptr2D src;
+ int height;
+ int width;
+ D val;
};
}}} // namespace cv { namespace gpu { namespace cudev
namespace cv { namespace gpu
{
- enum
- {
- BORDER_REFLECT101_GPU = 0,
- BORDER_REPLICATE_GPU,
- BORDER_CONSTANT_GPU,
- BORDER_REFLECT_GPU,
- BORDER_WRAP_GPU
- };
-
namespace cudev
{
__host__ __device__ __forceinline__ int divUp(int total, int grain)
#ifndef OPENCV_GPU_EMULATION_HPP_
#define OPENCV_GPU_EMULATION_HPP_
+#include "common.hpp"
#include "warp_reduce.hpp"
namespace cv { namespace gpu { namespace cudev
return ::atomicMin(address, val);
#endif
}
+ }; // struct cmem
+
+ struct glob
+ {
+ static __device__ __forceinline__ int atomicAdd(int* address, int val)
+ {
+ return ::atomicAdd(address, val);
+ }
+ static __device__ __forceinline__ unsigned int atomicAdd(unsigned int* address, unsigned int val)
+ {
+ return ::atomicAdd(address, val);
+ }
+ static __device__ __forceinline__ float atomicAdd(float* address, float val)
+ {
+ #if __CUDA_ARCH__ >= 200
+ return ::atomicAdd(address, val);
+ #else
+ int* address_as_i = (int*) address;
+ int old = *address_as_i, assumed;
+ do {
+ assumed = old;
+ old = ::atomicCAS(address_as_i, assumed,
+ __float_as_int(val + __int_as_float(assumed)));
+ } while (assumed != old);
+ return __int_as_float(old);
+ #endif
+ }
+ static __device__ __forceinline__ double atomicAdd(double* address, double val)
+ {
+ #if __CUDA_ARCH__ >= 130
+ unsigned long long int* address_as_ull = (unsigned long long int*) address;
+ unsigned long long int old = *address_as_ull, assumed;
+ do {
+ assumed = old;
+ old = ::atomicCAS(address_as_ull, assumed,
+ __double_as_longlong(val + __longlong_as_double(assumed)));
+ } while (assumed != old);
+ return __longlong_as_double(old);
+ #else
+ (void) address;
+ (void) val;
+ return 0.0;
+ #endif
+ }
+
+ static __device__ __forceinline__ int atomicMin(int* address, int val)
+ {
+ return ::atomicMin(address, val);
+ }
+ static __device__ __forceinline__ float atomicMin(float* address, float val)
+ {
+ #if __CUDA_ARCH__ >= 120
+ int* address_as_i = (int*) address;
+ int old = *address_as_i, assumed;
+ do {
+ assumed = old;
+ old = ::atomicCAS(address_as_i, assumed,
+ __float_as_int(::fminf(val, __int_as_float(assumed))));
+ } while (assumed != old);
+ return __int_as_float(old);
+ #else
+ (void) address;
+ (void) val;
+ return 0.0f;
+ #endif
+ }
+ static __device__ __forceinline__ double atomicMin(double* address, double val)
+ {
+ #if __CUDA_ARCH__ >= 130
+ unsigned long long int* address_as_ull = (unsigned long long int*) address;
+ unsigned long long int old = *address_as_ull, assumed;
+ do {
+ assumed = old;
+ old = ::atomicCAS(address_as_ull, assumed,
+ __double_as_longlong(::fmin(val, __longlong_as_double(assumed))));
+ } while (assumed != old);
+ return __longlong_as_double(old);
+ #else
+ (void) address;
+ (void) val;
+ return 0.0;
+ #endif
+ }
+
+ static __device__ __forceinline__ int atomicMax(int* address, int val)
+ {
+ return ::atomicMax(address, val);
+ }
+ static __device__ __forceinline__ float atomicMax(float* address, float val)
+ {
+ #if __CUDA_ARCH__ >= 120
+ int* address_as_i = (int*) address;
+ int old = *address_as_i, assumed;
+ do {
+ assumed = old;
+ old = ::atomicCAS(address_as_i, assumed,
+ __float_as_int(::fmaxf(val, __int_as_float(assumed))));
+ } while (assumed != old);
+ return __int_as_float(old);
+ #else
+ (void) address;
+ (void) val;
+ return 0.0f;
+ #endif
+ }
+ static __device__ __forceinline__ double atomicMax(double* address, double val)
+ {
+ #if __CUDA_ARCH__ >= 130
+ unsigned long long int* address_as_ull = (unsigned long long int*) address;
+ unsigned long long int old = *address_as_ull, assumed;
+ do {
+ assumed = old;
+ old = ::atomicCAS(address_as_ull, assumed,
+ __double_as_longlong(::fmax(val, __longlong_as_double(assumed))));
+ } while (assumed != old);
+ return __longlong_as_double(old);
+ #else
+ (void) address;
+ (void) val;
+ return 0.0;
+ #endif
+ }
};
- };
+ }; //struct Emulation
}}} // namespace cv { namespace gpu { namespace cudev
#endif /* OPENCV_GPU_EMULATION_HPP_ */
return src(__float2int_rz(y), __float2int_rz(x));
}
- const Ptr2D src;
+ Ptr2D src;
};
template <typename Ptr2D> struct LinearFilter
return saturate_cast<elem_type>(out);
}
- const Ptr2D src;
+ Ptr2D src;
};
template <typename Ptr2D> struct CubicFilter
return saturate_cast<elem_type>(res);
}
- const Ptr2D src;
+ Ptr2D src;
};
// for integer scaling
template <typename Ptr2D> struct IntegerAreaFilter
return saturate_cast<elem_type>(out);
}
- const Ptr2D src;
+ Ptr2D src;
float scale_x, scale_y ,scale;
};
return saturate_cast<elem_type>(out);
}
- const Ptr2D src;
+ Ptr2D src;
float scale_x, scale_y;
int width, haight;
};
__host__ __device__ __forceinline__ thresh_binary_func(const thresh_binary_func& other)
: thresh(other.thresh), maxVal(other.maxVal) {}
- const T thresh;
- const T maxVal;
+ T thresh;
+ T maxVal;
};
template <typename T> struct thresh_binary_inv_func : unary_function<T, T>
__host__ __device__ __forceinline__ thresh_binary_inv_func(const thresh_binary_inv_func& other)
: thresh(other.thresh), maxVal(other.maxVal) {}
- const T thresh;
- const T maxVal;
+ T thresh;
+ T maxVal;
};
template <typename T> struct thresh_trunc_func : unary_function<T, T>
__host__ __device__ __forceinline__ thresh_trunc_func(const thresh_trunc_func& other)
: thresh(other.thresh) {}
- const T thresh;
+ T thresh;
};
template <typename T> struct thresh_to_zero_func : unary_function<T, T>
__host__ __device__ __forceinline__ thresh_to_zero_func(const thresh_to_zero_func& other)
: thresh(other.thresh) {}
- const T thresh;
+ T thresh;
};
template <typename T> struct thresh_to_zero_inv_func : unary_function<T, T>
__host__ __device__ __forceinline__ thresh_to_zero_inv_func(const thresh_to_zero_inv_func& other)
: thresh(other.thresh) {}
- const T thresh;
+ T thresh;
};
// Function Object Adaptors
__host__ __device__ __forceinline__ unary_negate() {}
__host__ __device__ __forceinline__ unary_negate(const unary_negate& other) : pred(other.pred) {}
- const Predicate pred;
+ Predicate pred;
};
template <typename Predicate> __host__ __device__ __forceinline__ unary_negate<Predicate> not1(const Predicate& pred)
__host__ __device__ __forceinline__ binary_negate() {}
__host__ __device__ __forceinline__ binary_negate(const binary_negate& other) : pred(other.pred) {}
- const Predicate pred;
+ Predicate pred;
};
template <typename BinaryPredicate> __host__ __device__ __forceinline__ binary_negate<BinaryPredicate> not2(const BinaryPredicate& pred)
__host__ __device__ __forceinline__ binder1st() {}
__host__ __device__ __forceinline__ binder1st(const binder1st& other) : op(other.op), arg1(other.arg1) {}
- const Op op;
- const typename Op::first_argument_type arg1;
+ Op op;
+ typename Op::first_argument_type arg1;
};
template <typename Op, typename T> __host__ __device__ __forceinline__ binder1st<Op> bind1st(const Op& op, const T& x)
__host__ __device__ __forceinline__ binder2nd() {}
__host__ __device__ __forceinline__ binder2nd(const binder2nd& other) : op(other.op), arg2(other.arg2) {}
- const Op op;
- const typename Op::second_argument_type arg2;
+ Op op;
+ typename Op::second_argument_type arg2;
};
template <typename Op, typename T> __host__ __device__ __forceinline__ binder2nd<Op> bind2nd(const Op& op, const T& x)
namespace cv { namespace gpu {
CV_EXPORTS cv::String getNppErrorMessage(int code);
CV_EXPORTS cv::String getCudaDriverApiErrorMessage(int code);
-
- // Converts CPU border extrapolation mode into GPU internal analogue.
- // Returns true if the GPU analogue exists, false otherwise.
- CV_EXPORTS bool tryConvertToGpuBorderType(int cpuBorderType, int& gpuBorderType);
}}
#ifndef HAVE_CUDA
return getErrorString(code, cu_errors, cu_errors_num);
#endif
}
-
-bool cv::gpu::tryConvertToGpuBorderType(int cpuBorderType, int& gpuBorderType)
-{
-#ifndef HAVE_CUDA
- (void) cpuBorderType;
- (void) gpuBorderType;
- return false;
-#else
- switch (cpuBorderType)
- {
- case IPL_BORDER_REFLECT_101:
- gpuBorderType = cv::gpu::BORDER_REFLECT101_GPU;
- return true;
- case IPL_BORDER_REPLICATE:
- gpuBorderType = cv::gpu::BORDER_REPLICATE_GPU;
- return true;
- case IPL_BORDER_CONSTANT:
- gpuBorderType = cv::gpu::BORDER_CONSTANT_GPU;
- return true;
- case IPL_BORDER_REFLECT:
- gpuBorderType = cv::gpu::BORDER_REFLECT_GPU;
- return true;
- case IPL_BORDER_WRAP:
- gpuBorderType = cv::gpu::BORDER_WRAP_GPU;
- return true;
- default:
- return false;
- };
-#endif
-}
endif()
set(the_description "GPU-accelerated Computer Vision")
-ocv_add_module(gpu opencv_imgproc opencv_calib3d opencv_objdetect opencv_video opencv_photo opencv_legacy)
-ocv_module_include_directories("${CMAKE_CURRENT_SOURCE_DIR}/src/cuda")
+ocv_warnings_disable(CMAKE_CXX_FLAGS /wd4127 /wd4100 /wd4324 /wd4512 /wd4515 -Wundef -Wmissing-declarations -Wshadow -Wunused-parameter)
-file(GLOB lib_hdrs "include/opencv2/*.hpp" "include/opencv2/${name}/*.hpp" "include/opencv2/${name}/*.h")
-file(GLOB lib_int_hdrs "src/*.hpp" "src/*.h")
-file(GLOB lib_cuda_hdrs "src/cuda/*.hpp" "src/cuda/*.h")
-file(GLOB lib_srcs "src/*.cpp")
-file(GLOB lib_cuda "src/cuda/*.cu*")
-
-source_group("Include" FILES ${lib_hdrs})
-source_group("Src\\Host" FILES ${lib_srcs} ${lib_int_hdrs})
-source_group("Src\\Cuda" FILES ${lib_cuda} ${lib_cuda_hdrs})
-
-if(HAVE_CUDA)
- file(GLOB_RECURSE ncv_srcs "src/nvidia/*.cpp" "src/nvidia/*.h*")
- file(GLOB_RECURSE ncv_cuda "src/nvidia/*.cu")
- set(ncv_files ${ncv_srcs} ${ncv_cuda})
-
- source_group("Src\\NVidia" FILES ${ncv_files})
- ocv_include_directories("src/nvidia" "src/nvidia/core" "src/nvidia/NPP_staging" ${CUDA_INCLUDE_DIRS})
- ocv_warnings_disable(CMAKE_CXX_FLAGS -Wundef -Wmissing-declarations -Wshadow -Wunused-parameter /wd4211 /wd4201 /wd4100 /wd4505 /wd4408)
-
- if(MSVC)
- if(NOT ENABLE_NOISY_WARNINGS)
- foreach(var CMAKE_CXX_FLAGS CMAKE_CXX_FLAGS_RELEASE CMAKE_CXX_FLAGS_DEBUG)
- string(REPLACE "/W4" "/W3" ${var} "${${var}}")
- endforeach()
-
- set(CUDA_NVCC_FLAGS ${CUDA_NVCC_FLAGS} -Xcompiler /wd4251)
- endif()
- endif()
-
- ocv_cuda_compile(cuda_objs ${lib_cuda} ${ncv_cuda})
-
- set(cuda_link_libs ${CUDA_LIBRARIES} ${CUDA_CUDA_LIBRARY} ${CUDA_npp_LIBRARY})
-
- if(WITH_NVCUVID)
- set(cuda_link_libs ${cuda_link_libs} ${CUDA_nvcuvid_LIBRARY})
- endif()
-
- if(WIN32)
- find_cuda_helper_libs(nvcuvenc)
- set(cuda_link_libs ${cuda_link_libs} ${CUDA_nvcuvenc_LIBRARY})
- endif()
-
- if(WITH_FFMPEG)
- set(cuda_link_libs ${cuda_link_libs} ${HIGHGUI_LIBRARIES})
- endif()
-else()
- set(lib_cuda "")
- set(cuda_objs "")
- set(cuda_link_libs "")
- set(ncv_files "")
-endif()
-
-ocv_set_module_sources(
- HEADERS ${lib_hdrs}
- SOURCES ${lib_int_hdrs} ${lib_cuda_hdrs} ${lib_srcs} ${lib_cuda} ${ncv_files} ${cuda_objs}
- )
-
-ocv_create_module(${cuda_link_libs})
-
-if(HAVE_CUDA)
- if(HAVE_CUFFT)
- CUDA_ADD_CUFFT_TO_TARGET(${the_module})
- endif()
-
- if(HAVE_CUBLAS)
- CUDA_ADD_CUBLAS_TO_TARGET(${the_module})
- endif()
-
- install(FILES src/nvidia/NPP_staging/NPP_staging.hpp src/nvidia/core/NCV.hpp
- DESTINATION ${OPENCV_INCLUDE_INSTALL_PATH}/opencv2/${name}
- COMPONENT main)
-endif()
-
-ocv_add_precompiled_headers(${the_module})
-
-################################################################################################################
-################################ GPU Module Tests #####################################################
-################################################################################################################
-file(GLOB test_srcs "test/*.cpp")
-file(GLOB test_hdrs "test/*.hpp" "test/*.h")
-
-set(nvidia "")
-if(HAVE_CUDA)
- file(GLOB nvidia "test/nvidia/*.cpp" "test/nvidia/*.hpp" "test/nvidia/*.h")
- set(nvidia FILES "Src\\\\\\\\NVidia" ${nvidia}) # 8 ugly backslashes :'(
-endif()
-
-ocv_add_accuracy_tests(FILES "Include" ${test_hdrs}
- FILES "Src" ${test_srcs}
- ${nvidia})
-ocv_add_perf_tests()
+ocv_define_module(gpu opencv_calib3d opencv_objdetect opencv_gpuarithm opencv_gpuwarping OPTIONAL opencv_gpulegacy)
if(HAVE_CUDA)
add_subdirectory(perf4au)
+++ /dev/null
-cmake_minimum_required(VERSION 2.8.3)
-
-project(nv_perf_test)
-
-find_package(OpenCV REQUIRED)
-include_directories(${OpenCV_INCLUDE_DIR})
-
-add_executable(${PROJECT_NAME} main.cpp)
-
-target_link_libraries(${PROJECT_NAME} ${OpenCV_LIBS})
+++ /dev/null
-#include <cstdio>
-#define HAVE_CUDA 1
-#include <opencv2/core.hpp>
-#include <opencv2/gpu.hpp>
-#include <opencv2/highgui.hpp>
-#include <opencv2/video.hpp>
-#include <opencv2/ts.hpp>
-
-static void printOsInfo()
-{
-#if defined _WIN32
-# if defined _WIN64
- printf("[----------]\n[ GPU INFO ] \tRun on OS Windows x64.\n[----------]\n"); fflush(stdout);
-# else
- printf("[----------]\n[ GPU INFO ] \tRun on OS Windows x32.\n[----------]\n"); fflush(stdout);
-# endif
-#elif defined linux
-# if defined _LP64
- printf("[----------]\n[ GPU INFO ] \tRun on OS Linux x64.\n[----------]\n"); fflush(stdout);
-# else
- printf("[----------]\n[ GPU INFO ] \tRun on OS Linux x32.\n[----------]\n"); fflush(stdout);
-# endif
-#elif defined __APPLE__
-# if defined _LP64
- printf("[----------]\n[ GPU INFO ] \tRun on OS Apple x64.\n[----------]\n"); fflush(stdout);
-# else
- printf("[----------]\n[ GPU INFO ] \tRun on OS Apple x32.\n[----------]\n"); fflush(stdout);
-# endif
-#endif
-}
-
-static void printCudaInfo()
-{
- const int deviceCount = cv::gpu::getCudaEnabledDeviceCount();
-
- printf("[----------]\n"); fflush(stdout);
- printf("[ GPU INFO ] \tCUDA device count:: %d.\n", deviceCount); fflush(stdout);
- printf("[----------]\n"); fflush(stdout);
-
- for (int i = 0; i < deviceCount; ++i)
- {
- cv::gpu::DeviceInfo info(i);
-
- printf("[----------]\n"); fflush(stdout);
- printf("[ DEVICE ] \t# %d %s.\n", i, info.name().c_str()); fflush(stdout);
- printf("[ ] \tCompute capability: %d.%d\n", info.majorVersion(), info.minorVersion()); fflush(stdout);
- printf("[ ] \tMulti Processor Count: %d\n", info.multiProcessorCount()); fflush(stdout);
- printf("[ ] \tTotal memory: %d Mb\n", static_cast<int>(static_cast<int>(info.totalMemory() / 1024.0) / 1024.0)); fflush(stdout);
- printf("[ ] \tFree memory: %d Mb\n", static_cast<int>(static_cast<int>(info.freeMemory() / 1024.0) / 1024.0)); fflush(stdout);
- if (!info.isCompatible())
- printf("[ GPU INFO ] \tThis device is NOT compatible with current GPU module build\n");
- printf("[----------]\n"); fflush(stdout);
- }
-}
-
-int main(int argc, char* argv[])
-{
- printOsInfo();
- printCudaInfo();
-
- perf::Regression::Init("nv_perf_test");
- perf::TestBase::Init(argc, argv);
- testing::InitGoogleTest(&argc, argv);
-
- return RUN_ALL_TESTS();
-}
-
-#define DEF_PARAM_TEST(name, ...) typedef ::perf::TestBaseWithParam< std::tr1::tuple< __VA_ARGS__ > > name
-#define DEF_PARAM_TEST_1(name, param_type) typedef ::perf::TestBaseWithParam< param_type > name
-
-//////////////////////////////////////////////////////////
-// HoughLinesP
-
-DEF_PARAM_TEST_1(Image, std::string);
-
-GPU_PERF_TEST_P(Image, HoughLinesP, testing::Values(std::string("im1_1280x800.jpg")))
-{
- declare.time(30.0);
-
- std::string fileName = GetParam();
-
- const float rho = 1.f;
- const float theta = 1.f;
- const int threshold = 40;
- const int minLineLenght = 20;
- const int maxLineGap = 5;
-
- cv::Mat image = cv::imread(fileName, cv::IMREAD_GRAYSCALE);
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::GpuMat d_image(image);
- cv::gpu::GpuMat d_lines;
- cv::gpu::HoughLinesBuf d_buf;
-
- cv::gpu::HoughLinesP(d_image, d_lines, d_buf, rho, theta, minLineLenght, maxLineGap);
-
- TEST_CYCLE()
- {
- cv::gpu::HoughLinesP(d_image, d_lines, d_buf, rho, theta, minLineLenght, maxLineGap);
- }
- }
- else
- {
- cv::Mat mask;
- cv::Canny(image, mask, 50, 100);
-
- std::vector<cv::Vec4i> lines;
- cv::HoughLinesP(mask, lines, rho, theta, threshold, minLineLenght, maxLineGap);
-
- TEST_CYCLE()
- {
- cv::HoughLinesP(mask, lines, rho, theta, threshold, minLineLenght, maxLineGap);
- }
- }
-
- SANITY_CHECK(0);
-}
-
-//////////////////////////////////////////////////////////
-// GoodFeaturesToTrack
-
-DEF_PARAM_TEST(Image_Depth, std::string, perf::MatDepth);
-
-GPU_PERF_TEST_P(Image_Depth, GoodFeaturesToTrack,
- testing::Combine(
- testing::Values(std::string("im1_1280x800.jpg")),
- testing::Values(CV_8U, CV_16U)
- ))
-{
- declare.time(60);
-
- const std::string fileName = std::tr1::get<0>(GetParam());
- const int depth = std::tr1::get<1>(GetParam());
-
- const int maxCorners = 5000;
- const double qualityLevel = 0.05;
- const int minDistance = 5;
- const int blockSize = 3;
- const bool useHarrisDetector = true;
- const double k = 0.05;
-
- cv::Mat src = cv::imread(fileName, cv::IMREAD_GRAYSCALE);
- if (src.empty())
- FAIL() << "Unable to load source image [" << fileName << "]";
-
- if (depth != CV_8U)
- src.convertTo(src, depth);
-
- cv::Mat mask(src.size(), CV_8UC1, cv::Scalar::all(1));
- mask(cv::Rect(0, 0, 100, 100)).setTo(cv::Scalar::all(0));
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::GoodFeaturesToTrackDetector_GPU d_detector(maxCorners, qualityLevel, minDistance, blockSize, useHarrisDetector, k);
-
- cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat d_mask(mask);
- cv::gpu::GpuMat d_pts;
-
- d_detector(d_src, d_pts, d_mask);
-
- TEST_CYCLE()
- {
- d_detector(d_src, d_pts, d_mask);
- }
- }
- else
- {
- if (depth != CV_8U)
- FAIL() << "Unsupported depth";
-
- cv::Mat pts;
-
- cv::goodFeaturesToTrack(src, pts, maxCorners, qualityLevel, minDistance, mask, blockSize, useHarrisDetector, k);
-
- TEST_CYCLE()
- {
- cv::goodFeaturesToTrack(src, pts, maxCorners, qualityLevel, minDistance, mask, blockSize, useHarrisDetector, k);
- }
- }
-
- SANITY_CHECK(0);
-}
-
-//////////////////////////////////////////////////////////
-// OpticalFlowPyrLKSparse
-
-typedef std::pair<std::string, std::string> string_pair;
-
-DEF_PARAM_TEST(ImagePair_Depth_GraySource, string_pair, perf::MatDepth, bool);
-
-GPU_PERF_TEST_P(ImagePair_Depth_GraySource, OpticalFlowPyrLKSparse,
- testing::Combine(
- testing::Values(string_pair("im1_1280x800.jpg", "im2_1280x800.jpg")),
- testing::Values(CV_8U, CV_16U),
- testing::Bool()
- ))
-{
- declare.time(60);
-
- const string_pair fileNames = std::tr1::get<0>(GetParam());
- const int depth = std::tr1::get<1>(GetParam());
- const bool graySource = std::tr1::get<2>(GetParam());
-
- // PyrLK params
- const cv::Size winSize(15, 15);
- const int maxLevel = 5;
- const cv::TermCriteria criteria(cv::TermCriteria::COUNT + cv::TermCriteria::EPS, 30, 0.01);
-
- // GoodFeaturesToTrack params
- const int maxCorners = 5000;
- const double qualityLevel = 0.05;
- const int minDistance = 5;
- const int blockSize = 3;
- const bool useHarrisDetector = true;
- const double k = 0.05;
-
- cv::Mat src1 = cv::imread(fileNames.first, graySource ? cv::IMREAD_GRAYSCALE : cv::IMREAD_COLOR);
- if (src1.empty())
- FAIL() << "Unable to load source image [" << fileNames.first << "]";
-
- cv::Mat src2 = cv::imread(fileNames.second, graySource ? cv::IMREAD_GRAYSCALE : cv::IMREAD_COLOR);
- if (src2.empty())
- FAIL() << "Unable to load source image [" << fileNames.second << "]";
-
- cv::Mat gray_src;
- if (graySource)
- gray_src = src1;
- else
- cv::cvtColor(src1, gray_src, cv::COLOR_BGR2GRAY);
-
- cv::Mat pts;
- cv::goodFeaturesToTrack(gray_src, pts, maxCorners, qualityLevel, minDistance, cv::noArray(), blockSize, useHarrisDetector, k);
-
- if (depth != CV_8U)
- {
- src1.convertTo(src1, depth);
- src2.convertTo(src2, depth);
- }
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::GpuMat d_src1(src1);
- cv::gpu::GpuMat d_src2(src2);
- cv::gpu::GpuMat d_pts(pts.reshape(2, 1));
- cv::gpu::GpuMat d_nextPts;
- cv::gpu::GpuMat d_status;
-
- cv::gpu::PyrLKOpticalFlow d_pyrLK;
- d_pyrLK.winSize = winSize;
- d_pyrLK.maxLevel = maxLevel;
- d_pyrLK.iters = criteria.maxCount;
- d_pyrLK.useInitialFlow = false;
-
- d_pyrLK.sparse(d_src1, d_src2, d_pts, d_nextPts, d_status);
-
- TEST_CYCLE()
- {
- d_pyrLK.sparse(d_src1, d_src2, d_pts, d_nextPts, d_status);
- }
- }
- else
- {
- if (depth != CV_8U)
- FAIL() << "Unsupported depth";
-
- cv::Mat nextPts;
- cv::Mat status;
-
- cv::calcOpticalFlowPyrLK(src1, src2, pts, nextPts, status, cv::noArray(), winSize, maxLevel, criteria);
-
- TEST_CYCLE()
- {
- cv::calcOpticalFlowPyrLK(src1, src2, pts, nextPts, status, cv::noArray(), winSize, maxLevel, criteria);
- }
- }
-
- SANITY_CHECK(0);
-}
-
-//////////////////////////////////////////////////////////
-// OpticalFlowFarneback
-
-DEF_PARAM_TEST(ImagePair_Depth, string_pair, perf::MatDepth);
-
-GPU_PERF_TEST_P(ImagePair_Depth, OpticalFlowFarneback,
- testing::Combine(
- testing::Values(string_pair("im1_1280x800.jpg", "im2_1280x800.jpg")),
- testing::Values(CV_8U, CV_16U)
- ))
-{
- declare.time(500);
-
- const string_pair fileNames = std::tr1::get<0>(GetParam());
- const int depth = std::tr1::get<1>(GetParam());
-
- const double pyrScale = 0.5;
- const int numLevels = 6;
- const int winSize = 7;
- const int numIters = 15;
- const int polyN = 7;
- const double polySigma = 1.5;
- const int flags = cv::OPTFLOW_USE_INITIAL_FLOW;
-
- cv::Mat src1 = cv::imread(fileNames.first, cv::IMREAD_GRAYSCALE);
- if (src1.empty())
- FAIL() << "Unable to load source image [" << fileNames.first << "]";
-
- cv::Mat src2 = cv::imread(fileNames.second, cv::IMREAD_GRAYSCALE);
- if (src2.empty())
- FAIL() << "Unable to load source image [" << fileNames.second << "]";
-
- if (depth != CV_8U)
- {
- src1.convertTo(src1, depth);
- src2.convertTo(src2, depth);
- }
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::GpuMat d_src1(src1);
- cv::gpu::GpuMat d_src2(src2);
- cv::gpu::GpuMat d_u(src1.size(), CV_32FC1, cv::Scalar::all(0));
- cv::gpu::GpuMat d_v(src1.size(), CV_32FC1, cv::Scalar::all(0));
-
- cv::gpu::FarnebackOpticalFlow d_farneback;
- d_farneback.pyrScale = pyrScale;
- d_farneback.numLevels = numLevels;
- d_farneback.winSize = winSize;
- d_farneback.numIters = numIters;
- d_farneback.polyN = polyN;
- d_farneback.polySigma = polySigma;
- d_farneback.flags = flags;
-
- d_farneback(d_src1, d_src2, d_u, d_v);
-
- TEST_CYCLE_N(10)
- {
- d_farneback(d_src1, d_src2, d_u, d_v);
- }
- }
- else
- {
- if (depth != CV_8U)
- FAIL() << "Unsupported depth";
-
- cv::Mat flow(src1.size(), CV_32FC2, cv::Scalar::all(0));
-
- cv::calcOpticalFlowFarneback(src1, src2, flow, pyrScale, numLevels, winSize, numIters, polyN, polySigma, flags);
-
- TEST_CYCLE_N(10)
- {
- cv::calcOpticalFlowFarneback(src1, src2, flow, pyrScale, numLevels, winSize, numIters, polyN, polySigma, flags);
- }
- }
-
- SANITY_CHECK(0);
-}
-
-//////////////////////////////////////////////////////////
-// OpticalFlowBM
-
-void calcOpticalFlowBM(const cv::Mat& prev, const cv::Mat& curr,
- cv::Size bSize, cv::Size shiftSize, cv::Size maxRange, int usePrevious,
- cv::Mat& velx, cv::Mat& vely)
-{
- cv::Size sz((curr.cols - bSize.width + shiftSize.width)/shiftSize.width, (curr.rows - bSize.height + shiftSize.height)/shiftSize.height);
-
- velx.create(sz, CV_32FC1);
- vely.create(sz, CV_32FC1);
-
- CvMat cvprev = prev;
- CvMat cvcurr = curr;
-
- CvMat cvvelx = velx;
- CvMat cvvely = vely;
-
- cvCalcOpticalFlowBM(&cvprev, &cvcurr, bSize, shiftSize, maxRange, usePrevious, &cvvelx, &cvvely);
-}
-
-DEF_PARAM_TEST(ImagePair_BlockSize_ShiftSize_MaxRange, string_pair, cv::Size, cv::Size, cv::Size);
-
-GPU_PERF_TEST_P(ImagePair_BlockSize_ShiftSize_MaxRange, OpticalFlowBM,
- testing::Combine(
- testing::Values(string_pair("im1_1280x800.jpg", "im2_1280x800.jpg")),
- testing::Values(cv::Size(16, 16)),
- testing::Values(cv::Size(2, 2)),
- testing::Values(cv::Size(16, 16))
- ))
-{
- declare.time(3000);
-
- const string_pair fileNames = std::tr1::get<0>(GetParam());
- const cv::Size block_size = std::tr1::get<1>(GetParam());
- const cv::Size shift_size = std::tr1::get<2>(GetParam());
- const cv::Size max_range = std::tr1::get<3>(GetParam());
-
- cv::Mat src1 = cv::imread(fileNames.first, cv::IMREAD_GRAYSCALE);
- if (src1.empty())
- FAIL() << "Unable to load source image [" << fileNames.first << "]";
-
- cv::Mat src2 = cv::imread(fileNames.second, cv::IMREAD_GRAYSCALE);
- if (src2.empty())
- FAIL() << "Unable to load source image [" << fileNames.second << "]";
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::GpuMat d_src1(src1);
- cv::gpu::GpuMat d_src2(src2);
- cv::gpu::GpuMat d_velx, d_vely, buf;
-
- cv::gpu::calcOpticalFlowBM(d_src1, d_src2, block_size, shift_size, max_range, false, d_velx, d_vely, buf);
-
- TEST_CYCLE_N(10)
- {
- cv::gpu::calcOpticalFlowBM(d_src1, d_src2, block_size, shift_size, max_range, false, d_velx, d_vely, buf);
- }
- }
- else
- {
- cv::Mat velx, vely;
-
- calcOpticalFlowBM(src1, src2, block_size, shift_size, max_range, false, velx, vely);
-
- TEST_CYCLE_N(10)
- {
- calcOpticalFlowBM(src1, src2, block_size, shift_size, max_range, false, velx, vely);
- }
- }
-
- SANITY_CHECK(0);
-}
-
-GPU_PERF_TEST_P(ImagePair_BlockSize_ShiftSize_MaxRange, FastOpticalFlowBM,
- testing::Combine(
- testing::Values(string_pair("im1_1280x800.jpg", "im2_1280x800.jpg")),
- testing::Values(cv::Size(16, 16)),
- testing::Values(cv::Size(1, 1)),
- testing::Values(cv::Size(16, 16))
- ))
-{
- declare.time(3000);
-
- const string_pair fileNames = std::tr1::get<0>(GetParam());
- const cv::Size block_size = std::tr1::get<1>(GetParam());
- const cv::Size shift_size = std::tr1::get<2>(GetParam());
- const cv::Size max_range = std::tr1::get<3>(GetParam());
-
- cv::Mat src1 = cv::imread(fileNames.first, cv::IMREAD_GRAYSCALE);
- if (src1.empty())
- FAIL() << "Unable to load source image [" << fileNames.first << "]";
-
- cv::Mat src2 = cv::imread(fileNames.second, cv::IMREAD_GRAYSCALE);
- if (src2.empty())
- FAIL() << "Unable to load source image [" << fileNames.second << "]";
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::GpuMat d_src1(src1);
- cv::gpu::GpuMat d_src2(src2);
- cv::gpu::GpuMat d_velx, d_vely;
-
- cv::gpu::FastOpticalFlowBM fastBM;
-
- fastBM(d_src1, d_src2, d_velx, d_vely, max_range.width, block_size.width);
-
- TEST_CYCLE_N(10)
- {
- fastBM(d_src1, d_src2, d_velx, d_vely, max_range.width, block_size.width);
- }
- }
- else
- {
- cv::Mat velx, vely;
-
- calcOpticalFlowBM(src1, src2, block_size, shift_size, max_range, false, velx, vely);
-
- TEST_CYCLE_N(10)
- {
- calcOpticalFlowBM(src1, src2, block_size, shift_size, max_range, false, velx, vely);
- }
- }
-
- SANITY_CHECK(0);
-}
--- /dev/null
+Camera Calibration and 3D Reconstruction
+========================================
+
+.. highlight:: cpp
+
+
+
+gpu::solvePnPRansac
+-------------------
+Finds the object pose from 3D-2D point correspondences.
+
+.. ocv:function:: void gpu::solvePnPRansac(const Mat& object, const Mat& image, const Mat& camera_mat, const Mat& dist_coef, Mat& rvec, Mat& tvec, bool use_extrinsic_guess=false, int num_iters=100, float max_dist=8.0, int min_inlier_count=100, vector<int>* inliers=NULL)
+
+ :param object: Single-row matrix of object points.
+
+ :param image: Single-row matrix of image points.
+
+ :param camera_mat: 3x3 matrix of intrinsic camera parameters.
+
+ :param dist_coef: Distortion coefficients. See :ocv:func:`undistortPoints` for details.
+
+ :param rvec: Output 3D rotation vector.
+
+ :param tvec: Output 3D translation vector.
+
+ :param use_extrinsic_guess: Flag to indicate that the function must use ``rvec`` and ``tvec`` as an initial transformation guess. It is not supported for now.
+
+ :param num_iters: Maximum number of RANSAC iterations.
+
+ :param max_dist: Euclidean distance threshold to detect whether point is inlier or not.
+
+ :param min_inlier_count: Flag to indicate that the function must stop if greater or equal number of inliers is achieved. It is not supported for now.
+
+ :param inliers: Output vector of inlier indices.
+
+.. seealso:: :ocv:func:`solvePnPRansac`
introduction
initalization_and_information
data_structures
- operations_on_matrices
- per_element_operations
- image_processing
- matrix_reductions
object_detection
- feature_detection_and_description
- image_filtering
- camera_calibration_and_3d_reconstruction
- video
+ calib3d
+++ /dev/null
-Image Processing
-================
-
-.. highlight:: cpp
-
-
-
-gpu::meanShiftFiltering
----------------------------
-Performs mean-shift filtering for each point of the source image.
-
-.. ocv:function:: void gpu::meanShiftFiltering( const GpuMat& src, GpuMat& dst, int sp, int sr, TermCriteria criteria=TermCriteria(TermCriteria::MAX_ITER + TermCriteria::EPS, 5, 1), Stream& stream=Stream::Null() )
-
- :param src: Source image. Only ``CV_8UC4`` images are supported for now.
-
- :param dst: Destination image containing the color of mapped points. It has the same size and type as ``src`` .
-
- :param sp: Spatial window radius.
-
- :param sr: Color window radius.
-
- :param criteria: Termination criteria. See :ocv:class:`TermCriteria`.
-
-It maps each point of the source image into another point. As a result, you have a new color and new position of each point.
-
-
-
-gpu::meanShiftProc
-----------------------
-Performs a mean-shift procedure and stores information about processed points (their colors and positions) in two images.
-
-.. ocv:function:: void gpu::meanShiftProc( const GpuMat& src, GpuMat& dstr, GpuMat& dstsp, int sp, int sr, TermCriteria criteria=TermCriteria(TermCriteria::MAX_ITER + TermCriteria::EPS, 5, 1), Stream& stream=Stream::Null() )
-
- :param src: Source image. Only ``CV_8UC4`` images are supported for now.
-
- :param dstr: Destination image containing the color of mapped points. The size and type is the same as ``src`` .
-
- :param dstsp: Destination image containing the position of mapped points. The size is the same as ``src`` size. The type is ``CV_16SC2`` .
-
- :param sp: Spatial window radius.
-
- :param sr: Color window radius.
-
- :param criteria: Termination criteria. See :ocv:class:`TermCriteria`.
-
-.. seealso:: :ocv:func:`gpu::meanShiftFiltering`
-
-
-
-gpu::meanShiftSegmentation
-------------------------------
-Performs a mean-shift segmentation of the source image and eliminates small segments.
-
-.. ocv:function:: void gpu::meanShiftSegmentation(const GpuMat& src, Mat& dst, int sp, int sr, int minsize, TermCriteria criteria = TermCriteria(TermCriteria::MAX_ITER + TermCriteria::EPS, 5, 1))
-
- :param src: Source image. Only ``CV_8UC4`` images are supported for now.
-
- :param dst: Segmented image with the same size and type as ``src`` .
-
- :param sp: Spatial window radius.
-
- :param sr: Color window radius.
-
- :param minsize: Minimum segment size. Smaller segments are merged.
-
- :param criteria: Termination criteria. See :ocv:class:`TermCriteria`.
-
-
-
-gpu::integral
------------------
-Computes an integral image.
-
-.. ocv:function:: void gpu::integral(const GpuMat& src, GpuMat& sum, Stream& stream = Stream::Null())
-
- :param src: Source image. Only ``CV_8UC1`` images are supported for now.
-
- :param sum: Integral image containing 32-bit unsigned integer values packed into ``CV_32SC1`` .
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`integral`
-
-
-
-gpu::sqrIntegral
---------------------
-Computes a squared integral image.
-
-.. ocv:function:: void gpu::sqrIntegral(const GpuMat& src, GpuMat& sqsum, Stream& stream = Stream::Null())
-
- :param src: Source image. Only ``CV_8UC1`` images are supported for now.
-
- :param sqsum: Squared integral image containing 64-bit unsigned integer values packed into ``CV_64FC1`` .
-
- :param stream: Stream for the asynchronous version.
-
-
-
-gpu::columnSum
-------------------
-Computes a vertical (column) sum.
-
-.. ocv:function:: void gpu::columnSum(const GpuMat& src, GpuMat& sum)
-
- :param src: Source image. Only ``CV_32FC1`` images are supported for now.
-
- :param sum: Destination image of the ``CV_32FC1`` type.
-
-
-
-gpu::cornerHarris
----------------------
-Computes the Harris cornerness criteria at each image pixel.
-
-.. ocv:function:: void gpu::cornerHarris(const GpuMat& src, GpuMat& dst, int blockSize, int ksize, double k, int borderType=BORDER_REFLECT101)
-
- :param src: Source image. Only ``CV_8UC1`` and ``CV_32FC1`` images are supported for now.
-
- :param dst: Destination image containing cornerness values. It has the same size as ``src`` and ``CV_32FC1`` type.
-
- :param blockSize: Neighborhood size.
-
- :param ksize: Aperture parameter for the Sobel operator.
-
- :param k: Harris detector free parameter.
-
- :param borderType: Pixel extrapolation method. Only ``BORDER_REFLECT101`` and ``BORDER_REPLICATE`` are supported for now.
-
-.. seealso:: :ocv:func:`cornerHarris`
-
-
-
-gpu::cornerMinEigenVal
---------------------------
-Computes the minimum eigen value of a 2x2 derivative covariation matrix at each pixel (the cornerness criteria).
-
-.. ocv:function:: void gpu::cornerMinEigenVal(const GpuMat& src, GpuMat& dst, int blockSize, int ksize, int borderType=BORDER_REFLECT101)
-
-.. ocv:function:: void gpu::cornerMinEigenVal(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, int blockSize, int ksize, int borderType=BORDER_REFLECT101)
-
-.. ocv:function:: void gpu::cornerMinEigenVal(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, GpuMat& buf, int blockSize, int ksize, int borderType=BORDER_REFLECT101, Stream& stream = Stream::Null())
-
- :param src: Source image. Only ``CV_8UC1`` and ``CV_32FC1`` images are supported for now.
-
- :param dst: Destination image containing cornerness values. The size is the same. The type is ``CV_32FC1`` .
-
- :param blockSize: Neighborhood size.
-
- :param ksize: Aperture parameter for the Sobel operator.
-
- :param borderType: Pixel extrapolation method. Only ``BORDER_REFLECT101`` and ``BORDER_REPLICATE`` are supported for now.
-
-.. seealso:: :ocv:func:`cornerMinEigenVal`
-
-
-
-gpu::mulSpectrums
----------------------
-Performs a per-element multiplication of two Fourier spectrums.
-
-.. ocv:function:: void gpu::mulSpectrums( const GpuMat& a, const GpuMat& b, GpuMat& c, int flags, bool conjB=false, Stream& stream=Stream::Null() )
-
- :param a: First spectrum.
-
- :param b: Second spectrum with the same size and type as ``a`` .
-
- :param c: Destination spectrum.
-
- :param flags: Mock parameter used for CPU/GPU interfaces similarity.
-
- :param conjB: Optional flag to specify if the second spectrum needs to be conjugated before the multiplication.
-
- Only full (not packed) ``CV_32FC2`` complex spectrums in the interleaved format are supported for now.
-
-.. seealso:: :ocv:func:`mulSpectrums`
-
-
-
-gpu::mulAndScaleSpectrums
------------------------------
-Performs a per-element multiplication of two Fourier spectrums and scales the result.
-
-.. ocv:function:: void gpu::mulAndScaleSpectrums( const GpuMat& a, const GpuMat& b, GpuMat& c, int flags, float scale, bool conjB=false, Stream& stream=Stream::Null() )
-
- :param a: First spectrum.
-
- :param b: Second spectrum with the same size and type as ``a`` .
-
- :param c: Destination spectrum.
-
- :param flags: Mock parameter used for CPU/GPU interfaces similarity.
-
- :param scale: Scale constant.
-
- :param conjB: Optional flag to specify if the second spectrum needs to be conjugated before the multiplication.
-
- Only full (not packed) ``CV_32FC2`` complex spectrums in the interleaved format are supported for now.
-
-.. seealso:: :ocv:func:`mulSpectrums`
-
-
-
-gpu::dft
-------------
-Performs a forward or inverse discrete Fourier transform (1D or 2D) of the floating point matrix.
-
-.. ocv:function:: void gpu::dft( const GpuMat& src, GpuMat& dst, Size dft_size, int flags=0, Stream& stream=Stream::Null() )
-
- :param src: Source matrix (real or complex).
-
- :param dst: Destination matrix (real or complex).
-
- :param dft_size: Size of a discrete Fourier transform.
-
- :param flags: Optional flags:
-
- * **DFT_ROWS** transforms each individual row of the source matrix.
-
- * **DFT_SCALE** scales the result: divide it by the number of elements in the transform (obtained from ``dft_size`` ).
-
- * **DFT_INVERSE** inverts DFT. Use for complex-complex cases (real-complex and complex-real cases are always forward and inverse, respectively).
-
- * **DFT_REAL_OUTPUT** specifies the output as real. The source matrix is the result of real-complex transform, so the destination matrix must be real.
-
-Use to handle real matrices ( ``CV32FC1`` ) and complex matrices in the interleaved format ( ``CV32FC2`` ).
-
-The source matrix should be continuous, otherwise reallocation and data copying is performed. The function chooses an operation mode depending on the flags, size, and channel count of the source matrix:
-
- * If the source matrix is complex and the output is not specified as real, the destination matrix is complex and has the ``dft_size`` size and ``CV_32FC2`` type. The destination matrix contains a full result of the DFT (forward or inverse).
-
- * If the source matrix is complex and the output is specified as real, the function assumes that its input is the result of the forward transform (see the next item). The destination matrix has the ``dft_size`` size and ``CV_32FC1`` type. It contains the result of the inverse DFT.
-
- * If the source matrix is real (its type is ``CV_32FC1`` ), forward DFT is performed. The result of the DFT is packed into complex ( ``CV_32FC2`` ) matrix. So, the width of the destination matrix is ``dft_size.width / 2 + 1`` . But if the source is a single column, the height is reduced instead of the width.
-
-.. seealso:: :ocv:func:`dft`
-
-
-gpu::ConvolveBuf
-----------------
-.. ocv:struct:: gpu::ConvolveBuf
-
-Class providing a memory buffer for :ocv:func:`gpu::convolve` function, plus it allows to adjust some specific parameters. ::
-
- struct CV_EXPORTS ConvolveBuf
- {
- Size result_size;
- Size block_size;
- Size user_block_size;
- Size dft_size;
- int spect_len;
-
- GpuMat image_spect, templ_spect, result_spect;
- GpuMat image_block, templ_block, result_data;
-
- void create(Size image_size, Size templ_size);
- static Size estimateBlockSize(Size result_size, Size templ_size);
- };
-
-You can use field `user_block_size` to set specific block size for :ocv:func:`gpu::convolve` function. If you leave its default value `Size(0,0)` then automatic estimation of block size will be used (which is optimized for speed). By varying `user_block_size` you can reduce memory requirements at the cost of speed.
-
-gpu::ConvolveBuf::create
-------------------------
-.. ocv:function:: gpu::ConvolveBuf::create(Size image_size, Size templ_size)
-
-Constructs a buffer for :ocv:func:`gpu::convolve` function with respective arguments.
-
-
-gpu::convolve
------------------
-Computes a convolution (or cross-correlation) of two images.
-
-.. ocv:function:: void gpu::convolve(const GpuMat& image, const GpuMat& templ, GpuMat& result, bool ccorr=false)
-
-.. ocv:function:: void gpu::convolve( const GpuMat& image, const GpuMat& templ, GpuMat& result, bool ccorr, ConvolveBuf& buf, Stream& stream=Stream::Null() )
-
- :param image: Source image. Only ``CV_32FC1`` images are supported for now.
-
- :param templ: Template image. The size is not greater than the ``image`` size. The type is the same as ``image`` .
-
- :param result: Result image. If ``image`` is *W x H* and ``templ`` is *w x h*, then ``result`` must be *W-w+1 x H-h+1*.
-
- :param ccorr: Flags to evaluate cross-correlation instead of convolution.
-
- :param buf: Optional buffer to avoid extra memory allocations and to adjust some specific parameters. See :ocv:struct:`gpu::ConvolveBuf`.
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`gpu::filter2D`
-
-gpu::MatchTemplateBuf
----------------------
-.. ocv:struct:: gpu::MatchTemplateBuf
-
-Class providing memory buffers for :ocv:func:`gpu::matchTemplate` function, plus it allows to adjust some specific parameters. ::
-
- struct CV_EXPORTS MatchTemplateBuf
- {
- Size user_block_size;
- GpuMat imagef, templf;
- std::vector<GpuMat> images;
- std::vector<GpuMat> image_sums;
- std::vector<GpuMat> image_sqsums;
- };
-
-You can use field `user_block_size` to set specific block size for :ocv:func:`gpu::matchTemplate` function. If you leave its default value `Size(0,0)` then automatic estimation of block size will be used (which is optimized for speed). By varying `user_block_size` you can reduce memory requirements at the cost of speed.
-
-gpu::matchTemplate
-----------------------
-Computes a proximity map for a raster template and an image where the template is searched for.
-
-.. ocv:function:: void gpu::matchTemplate(const GpuMat& image, const GpuMat& templ, GpuMat& result, int method, Stream &stream = Stream::Null())
-
-.. ocv:function:: void gpu::matchTemplate(const GpuMat& image, const GpuMat& templ, GpuMat& result, int method, MatchTemplateBuf &buf, Stream& stream = Stream::Null())
-
- :param image: Source image. ``CV_32F`` and ``CV_8U`` depth images (1..4 channels) are supported for now.
-
- :param templ: Template image with the size and type the same as ``image`` .
-
- :param result: Map containing comparison results ( ``CV_32FC1`` ). If ``image`` is *W x H* and ``templ`` is *w x h*, then ``result`` must be *W-w+1 x H-h+1*.
-
- :param method: Specifies the way to compare the template with the image.
-
- :param buf: Optional buffer to avoid extra memory allocations and to adjust some specific parameters. See :ocv:struct:`gpu::MatchTemplateBuf`.
-
- :param stream: Stream for the asynchronous version.
-
- The following methods are supported for the ``CV_8U`` depth images for now:
-
- * ``CV_TM_SQDIFF``
- * ``CV_TM_SQDIFF_NORMED``
- * ``CV_TM_CCORR``
- * ``CV_TM_CCORR_NORMED``
- * ``CV_TM_CCOEFF``
- * ``CV_TM_CCOEFF_NORMED``
-
- The following methods are supported for the ``CV_32F`` images for now:
-
- * ``CV_TM_SQDIFF``
- * ``CV_TM_CCORR``
-
-.. seealso:: :ocv:func:`matchTemplate`
-
-
-gpu::remap
---------------
-Applies a generic geometrical transformation to an image.
-
-.. ocv:function:: void gpu::remap( const GpuMat& src, GpuMat& dst, const GpuMat& xmap, const GpuMat& ymap, int interpolation, int borderMode=BORDER_CONSTANT, Scalar borderValue=Scalar(), Stream& stream=Stream::Null() )
-
- :param src: Source image.
-
- :param dst: Destination image with the size the same as ``xmap`` and the type the same as ``src`` .
-
- :param xmap: X values. Only ``CV_32FC1`` type is supported.
-
- :param ymap: Y values. Only ``CV_32FC1`` type is supported.
-
- :param interpolation: Interpolation method (see :ocv:func:`resize` ). ``INTER_NEAREST`` , ``INTER_LINEAR`` and ``INTER_CUBIC`` are supported for now.
-
- :param borderMode: Pixel extrapolation method (see :ocv:func:`borderInterpolate` ). ``BORDER_REFLECT101`` , ``BORDER_REPLICATE`` , ``BORDER_CONSTANT`` , ``BORDER_REFLECT`` and ``BORDER_WRAP`` are supported for now.
-
- :param borderValue: Value used in case of a constant border. By default, it is 0.
-
- :param stream: Stream for the asynchronous version.
-
-The function transforms the source image using the specified map:
-
-.. math::
-
- \texttt{dst} (x,y) = \texttt{src} (xmap(x,y), ymap(x,y))
-
-Values of pixels with non-integer coordinates are computed using the bilinear interpolation.
-
-.. seealso:: :ocv:func:`remap`
-
-
-
-gpu::cvtColor
------------------
-Converts an image from one color space to another.
-
-.. ocv:function:: void gpu::cvtColor(const GpuMat& src, GpuMat& dst, int code, int dcn = 0, Stream& stream = Stream::Null())
-
- :param src: Source image with ``CV_8U`` , ``CV_16U`` , or ``CV_32F`` depth and 1, 3, or 4 channels.
-
- :param dst: Destination image with the same size and depth as ``src`` .
-
- :param code: Color space conversion code. For details, see :ocv:func:`cvtColor` . Conversion to/from Luv and Bayer color spaces is not supported.
-
- :param dcn: Number of channels in the destination image. If the parameter is 0, the number of the channels is derived automatically from ``src`` and the ``code`` .
-
- :param stream: Stream for the asynchronous version.
-
-3-channel color spaces (like ``HSV``, ``XYZ``, and so on) can be stored in a 4-channel image for better performance.
-
-.. seealso:: :ocv:func:`cvtColor`
-
-
-
-gpu::swapChannels
------------------
-Exchanges the color channels of an image in-place.
-
-.. ocv:function:: void gpu::swapChannels(GpuMat& image, const int dstOrder[4], Stream& stream = Stream::Null())
-
- :param image: Source image. Supports only ``CV_8UC4`` type.
-
- :param dstOrder: Integer array describing how channel values are permutated. The n-th entry of the array contains the number of the channel that is stored in the n-th channel of the output image. E.g. Given an RGBA image, aDstOrder = [3,2,1,0] converts this to ABGR channel order.
-
- :param stream: Stream for the asynchronous version.
-
-The methods support arbitrary permutations of the original channels, including replication.
-
-
-
-gpu::threshold
-------------------
-Applies a fixed-level threshold to each array element.
-
-.. ocv:function:: double gpu::threshold(const GpuMat& src, GpuMat& dst, double thresh, double maxval, int type, Stream& stream = Stream::Null())
-
- :param src: Source array (single-channel).
-
- :param dst: Destination array with the same size and type as ``src`` .
-
- :param thresh: Threshold value.
-
- :param maxval: Maximum value to use with ``THRESH_BINARY`` and ``THRESH_BINARY_INV`` threshold types.
-
- :param type: Threshold type. For details, see :ocv:func:`threshold` . The ``THRESH_OTSU`` threshold type is not supported.
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`threshold`
-
-
-
-gpu::resize
----------------
-Resizes an image.
-
-.. ocv:function:: void gpu::resize(const GpuMat& src, GpuMat& dst, Size dsize, double fx=0, double fy=0, int interpolation = INTER_LINEAR, Stream& stream = Stream::Null())
-
- :param src: Source image.
-
- :param dst: Destination image with the same type as ``src`` . The size is ``dsize`` (when it is non-zero) or the size is computed from ``src.size()`` , ``fx`` , and ``fy`` .
-
- :param dsize: Destination image size. If it is zero, it is computed as:
-
- .. math::
- \texttt{dsize = Size(round(fx*src.cols), round(fy*src.rows))}
-
- Either ``dsize`` or both ``fx`` and ``fy`` must be non-zero.
-
- :param fx: Scale factor along the horizontal axis. If it is zero, it is computed as:
-
- .. math::
-
- \texttt{(double)dsize.width/src.cols}
-
- :param fy: Scale factor along the vertical axis. If it is zero, it is computed as:
-
- .. math::
-
- \texttt{(double)dsize.height/src.rows}
-
- :param interpolation: Interpolation method. ``INTER_NEAREST`` , ``INTER_LINEAR`` and ``INTER_CUBIC`` are supported for now.
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`resize`
-
-
-
-gpu::warpAffine
--------------------
-Applies an affine transformation to an image.
-
-.. ocv:function:: void gpu::warpAffine( const GpuMat& src, GpuMat& dst, const Mat& M, Size dsize, int flags=INTER_LINEAR, int borderMode=BORDER_CONSTANT, Scalar borderValue=Scalar(), Stream& stream=Stream::Null() )
-
- :param src: Source image. ``CV_8U`` , ``CV_16U`` , ``CV_32S`` , or ``CV_32F`` depth and 1, 3, or 4 channels are supported.
-
- :param dst: Destination image with the same type as ``src`` . The size is ``dsize`` .
-
- :param M: *2x3* transformation matrix.
-
- :param dsize: Size of the destination image.
-
- :param flags: Combination of interpolation methods (see :ocv:func:`resize`) and the optional flag ``WARP_INVERSE_MAP`` specifying that ``M`` is an inverse transformation ( ``dst=>src`` ). Only ``INTER_NEAREST`` , ``INTER_LINEAR`` , and ``INTER_CUBIC`` interpolation methods are supported.
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`warpAffine`
-
-
-
-gpu::buildWarpAffineMaps
-------------------------
-Builds transformation maps for affine transformation.
-
-.. ocv:function:: void gpu::buildWarpAffineMaps(const Mat& M, bool inverse, Size dsize, GpuMat& xmap, GpuMat& ymap, Stream& stream = Stream::Null())
-
- :param M: *2x3* transformation matrix.
-
- :param inverse: Flag specifying that ``M`` is an inverse transformation ( ``dst=>src`` ).
-
- :param dsize: Size of the destination image.
-
- :param xmap: X values with ``CV_32FC1`` type.
-
- :param ymap: Y values with ``CV_32FC1`` type.
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`gpu::warpAffine` , :ocv:func:`gpu::remap`
-
-
-
-gpu::warpPerspective
-------------------------
-Applies a perspective transformation to an image.
-
-.. ocv:function:: void gpu::warpPerspective( const GpuMat& src, GpuMat& dst, const Mat& M, Size dsize, int flags=INTER_LINEAR, int borderMode=BORDER_CONSTANT, Scalar borderValue=Scalar(), Stream& stream=Stream::Null() )
-
- :param src: Source image. ``CV_8U`` , ``CV_16U`` , ``CV_32S`` , or ``CV_32F`` depth and 1, 3, or 4 channels are supported.
-
- :param dst: Destination image with the same type as ``src`` . The size is ``dsize`` .
-
- :param M: *3x3* transformation matrix.
-
- :param dsize: Size of the destination image.
-
- :param flags: Combination of interpolation methods (see :ocv:func:`resize` ) and the optional flag ``WARP_INVERSE_MAP`` specifying that ``M`` is the inverse transformation ( ``dst => src`` ). Only ``INTER_NEAREST`` , ``INTER_LINEAR`` , and ``INTER_CUBIC`` interpolation methods are supported.
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`warpPerspective`
-
-
-
-gpu::buildWarpPerspectiveMaps
------------------------------
-Builds transformation maps for perspective transformation.
-
-.. ocv:function:: void gpu::buildWarpAffineMaps(const Mat& M, bool inverse, Size dsize, GpuMat& xmap, GpuMat& ymap, Stream& stream = Stream::Null())
-
- :param M: *3x3* transformation matrix.
-
- :param inverse: Flag specifying that ``M`` is an inverse transformation ( ``dst=>src`` ).
-
- :param dsize: Size of the destination image.
-
- :param xmap: X values with ``CV_32FC1`` type.
-
- :param ymap: Y values with ``CV_32FC1`` type.
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`gpu::warpPerspective` , :ocv:func:`gpu::remap`
-
-
-
-gpu::rotate
----------------
-Rotates an image around the origin (0,0) and then shifts it.
-
-.. ocv:function:: void gpu::rotate(const GpuMat& src, GpuMat& dst, Size dsize, double angle, double xShift = 0, double yShift = 0, int interpolation = INTER_LINEAR, Stream& stream = Stream::Null())
-
- :param src: Source image. Supports 1, 3 or 4 channels images with ``CV_8U`` , ``CV_16U`` or ``CV_32F`` depth.
-
- :param dst: Destination image with the same type as ``src`` . The size is ``dsize`` .
-
- :param dsize: Size of the destination image.
-
- :param angle: Angle of rotation in degrees.
-
- :param xShift: Shift along the horizontal axis.
-
- :param yShift: Shift along the vertical axis.
-
- :param interpolation: Interpolation method. Only ``INTER_NEAREST`` , ``INTER_LINEAR`` , and ``INTER_CUBIC`` are supported.
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`gpu::warpAffine`
-
-
-
-gpu::copyMakeBorder
------------------------
-Forms a border around an image.
-
-.. ocv:function:: void gpu::copyMakeBorder(const GpuMat& src, GpuMat& dst, int top, int bottom, int left, int right, int borderType, const Scalar& value = Scalar(), Stream& stream = Stream::Null())
-
- :param src: Source image. ``CV_8UC1`` , ``CV_8UC4`` , ``CV_32SC1`` , and ``CV_32FC1`` types are supported.
-
- :param dst: Destination image with the same type as ``src``. The size is ``Size(src.cols+left+right, src.rows+top+bottom)`` .
-
- :param top:
-
- :param bottom:
-
- :param left:
-
- :param right: Number of pixels in each direction from the source image rectangle to extrapolate. For example: ``top=1, bottom=1, left=1, right=1`` mean that 1 pixel-wide border needs to be built.
-
- :param borderType: Border type. See :ocv:func:`borderInterpolate` for details. ``BORDER_REFLECT101`` , ``BORDER_REPLICATE`` , ``BORDER_CONSTANT`` , ``BORDER_REFLECT`` and ``BORDER_WRAP`` are supported for now.
-
- :param value: Border value.
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`copyMakeBorder`
-
-
-
-gpu::rectStdDev
--------------------
-Computes a standard deviation of integral images.
-
-.. ocv:function:: void gpu::rectStdDev(const GpuMat& src, const GpuMat& sqr, GpuMat& dst, const Rect& rect, Stream& stream = Stream::Null())
-
- :param src: Source image. Only the ``CV_32SC1`` type is supported.
-
- :param sqr: Squared source image. Only the ``CV_32FC1`` type is supported.
-
- :param dst: Destination image with the same type and size as ``src`` .
-
- :param rect: Rectangular window.
-
- :param stream: Stream for the asynchronous version.
-
-
-
-gpu::evenLevels
--------------------
-Computes levels with even distribution.
-
-.. ocv:function:: void gpu::evenLevels(GpuMat& levels, int nLevels, int lowerLevel, int upperLevel)
-
- :param levels: Destination array. ``levels`` has 1 row, ``nLevels`` columns, and the ``CV_32SC1`` type.
-
- :param nLevels: Number of computed levels. ``nLevels`` must be at least 2.
-
- :param lowerLevel: Lower boundary value of the lowest level.
-
- :param upperLevel: Upper boundary value of the greatest level.
-
-
-
-gpu::histEven
------------------
-Calculates a histogram with evenly distributed bins.
-
-.. ocv:function:: void gpu::histEven(const GpuMat& src, GpuMat& hist, int histSize, int lowerLevel, int upperLevel, Stream& stream = Stream::Null())
-
-.. ocv:function:: void gpu::histEven(const GpuMat& src, GpuMat& hist, GpuMat& buf, int histSize, int lowerLevel, int upperLevel, Stream& stream = Stream::Null())
-
-.. ocv:function:: void gpu::histEven( const GpuMat& src, GpuMat hist[4], int histSize[4], int lowerLevel[4], int upperLevel[4], Stream& stream=Stream::Null() )
-
-.. ocv:function:: void gpu::histEven( const GpuMat& src, GpuMat hist[4], GpuMat& buf, int histSize[4], int lowerLevel[4], int upperLevel[4], Stream& stream=Stream::Null() )
-
- :param src: Source image. ``CV_8U``, ``CV_16U``, or ``CV_16S`` depth and 1 or 4 channels are supported. For a four-channel image, all channels are processed separately.
-
- :param hist: Destination histogram with one row, ``histSize`` columns, and the ``CV_32S`` type.
-
- :param histSize: Size of the histogram.
-
- :param lowerLevel: Lower boundary of lowest-level bin.
-
- :param upperLevel: Upper boundary of highest-level bin.
-
- :param buf: Optional buffer to avoid extra memory allocations (for many calls with the same sizes).
-
- :param stream: Stream for the asynchronous version.
-
-
-
-gpu::histRange
-------------------
-Calculates a histogram with bins determined by the ``levels`` array.
-
-.. ocv:function:: void gpu::histRange(const GpuMat& src, GpuMat& hist, const GpuMat& levels, Stream& stream = Stream::Null())
-
-.. ocv:function:: void gpu::histRange(const GpuMat& src, GpuMat& hist, const GpuMat& levels, GpuMat& buf, Stream& stream = Stream::Null())
-
- :param src: Source image. ``CV_8U`` , ``CV_16U`` , or ``CV_16S`` depth and 1 or 4 channels are supported. For a four-channel image, all channels are processed separately.
-
- :param hist: Destination histogram with one row, ``(levels.cols-1)`` columns, and the ``CV_32SC1`` type.
-
- :param levels: Number of levels in the histogram.
-
- :param buf: Optional buffer to avoid extra memory allocations (for many calls with the same sizes).
-
- :param stream: Stream for the asynchronous version.
-
-
-
-gpu::calcHist
-------------------
-Calculates histogram for one channel 8-bit image.
-
-.. ocv:function:: void gpu::calcHist(const GpuMat& src, GpuMat& hist, Stream& stream = Stream::Null())
-
- :param src: Source image.
-
- :param hist: Destination histogram with one row, 256 columns, and the ``CV_32SC1`` type.
-
- :param stream: Stream for the asynchronous version.
-
-
-
-gpu::equalizeHist
-------------------
-Equalizes the histogram of a grayscale image.
-
-.. ocv:function:: void gpu::equalizeHist(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null())
-
-.. ocv:function:: void gpu::equalizeHist(const GpuMat& src, GpuMat& dst, GpuMat& hist, GpuMat& buf, Stream& stream = Stream::Null())
-
- :param src: Source image.
-
- :param dst: Destination image.
-
- :param hist: Destination histogram with one row, 256 columns, and the ``CV_32SC1`` type.
-
- :param buf: Optional buffer to avoid extra memory allocations (for many calls with the same sizes).
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`equalizeHist`
-
-
-
-gpu::buildWarpPlaneMaps
------------------------
-Builds plane warping maps.
-
-.. ocv:function:: void gpu::buildWarpPlaneMaps( Size src_size, Rect dst_roi, const Mat & K, const Mat& R, const Mat & T, float scale, GpuMat& map_x, GpuMat& map_y, Stream& stream=Stream::Null() )
-
- :param stream: Stream for the asynchronous version.
-
-
-
-gpu::buildWarpCylindricalMaps
------------------------------
-Builds cylindrical warping maps.
-
-.. ocv:function:: void gpu::buildWarpCylindricalMaps( Size src_size, Rect dst_roi, const Mat & K, const Mat& R, float scale, GpuMat& map_x, GpuMat& map_y, Stream& stream=Stream::Null() )
-
- :param stream: Stream for the asynchronous version.
-
-
-
-gpu::buildWarpSphericalMaps
----------------------------
-Builds spherical warping maps.
-
-.. ocv:function:: void gpu::buildWarpSphericalMaps( Size src_size, Rect dst_roi, const Mat & K, const Mat& R, float scale, GpuMat& map_x, GpuMat& map_y, Stream& stream=Stream::Null() )
-
- :param stream: Stream for the asynchronous version.
-
-
-
-gpu::pyrDown
--------------------
-Smoothes an image and downsamples it.
-
-.. ocv:function:: void gpu::pyrDown(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null())
-
- :param src: Source image.
-
- :param dst: Destination image. Will have ``Size((src.cols+1)/2, (src.rows+1)/2)`` size and the same type as ``src`` .
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`pyrDown`
-
-
-
-gpu::pyrUp
--------------------
-Upsamples an image and then smoothes it.
-
-.. ocv:function:: void gpu::pyrUp(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null())
-
- :param src: Source image.
-
- :param dst: Destination image. Will have ``Size(src.cols*2, src.rows*2)`` size and the same type as ``src`` .
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`pyrUp`
-
-
-
-gpu::blendLinear
--------------------
-Performs linear blending of two images.
-
-.. ocv:function:: void gpu::blendLinear(const GpuMat& img1, const GpuMat& img2, const GpuMat& weights1, const GpuMat& weights2, GpuMat& result, Stream& stream = Stream::Null())
-
- :param img1: First image. Supports only ``CV_8U`` and ``CV_32F`` depth.
-
- :param img2: Second image. Must have the same size and the same type as ``img1`` .
-
- :param weights1: Weights for first image. Must have tha same size as ``img1`` . Supports only ``CV_32F`` type.
-
- :param weights2: Weights for second image. Must have tha same size as ``img2`` . Supports only ``CV_32F`` type.
-
- :param result: Destination image.
-
- :param stream: Stream for the asynchronous version.
-
-
-gpu::bilateralFilter
---------------------
-Performs bilateral filtering of passed image
-
-.. ocv:function:: void gpu::bilateralFilter( const GpuMat& src, GpuMat& dst, int kernel_size, float sigma_color, float sigma_spatial, int borderMode=BORDER_DEFAULT, Stream& stream=Stream::Null() )
-
- :param src: Source image. Supports only (channles != 2 && depth() != CV_8S && depth() != CV_32S && depth() != CV_64F).
-
- :param dst: Destination imagwe.
-
- :param kernel_size: Kernel window size.
-
- :param sigma_color: Filter sigma in the color space.
-
- :param sigma_spatial: Filter sigma in the coordinate space.
-
- :param borderMode: Border type. See :ocv:func:`borderInterpolate` for details. ``BORDER_REFLECT101`` , ``BORDER_REPLICATE`` , ``BORDER_CONSTANT`` , ``BORDER_REFLECT`` and ``BORDER_WRAP`` are supported for now.
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso::
-
- :ocv:func:`bilateralFilter`,
-
-
-gpu::nonLocalMeans
--------------------
-Performs pure non local means denoising without any simplification, and thus it is not fast.
-
-.. ocv:function:: void gpu::nonLocalMeans(const GpuMat& src, GpuMat& dst, float h, int search_window = 21, int block_size = 7, int borderMode = BORDER_DEFAULT, Stream& s = Stream::Null())
-
- :param src: Source image. Supports only CV_8UC1, CV_8UC2 and CV_8UC3.
-
- :param dst: Destination image.
-
- :param h: Filter sigma regulating filter strength for color.
-
- :param search_window: Size of search window.
-
- :param block_size: Size of block used for computing weights.
-
- :param borderMode: Border type. See :ocv:func:`borderInterpolate` for details. ``BORDER_REFLECT101`` , ``BORDER_REPLICATE`` , ``BORDER_CONSTANT`` , ``BORDER_REFLECT`` and ``BORDER_WRAP`` are supported for now.
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso::
-
- :ocv:func:`fastNlMeansDenoising`
-
-gpu::FastNonLocalMeansDenoising
--------------------------------
-.. ocv:class:: gpu::FastNonLocalMeansDenoising
-
- ::
-
- class FastNonLocalMeansDenoising
- {
- public:
- //! Simple method, recommended for grayscale images (though it supports multichannel images)
- void simpleMethod(const GpuMat& src, GpuMat& dst, float h, int search_window = 21, int block_size = 7, Stream& s = Stream::Null())
- //! Processes luminance and color components separatelly
- void labMethod(const GpuMat& src, GpuMat& dst, float h_luminance, float h_color, int search_window = 21, int block_size = 7, Stream& s = Stream::Null())
- };
-
-The class implements fast approximate Non Local Means Denoising algorithm.
-
-gpu::FastNonLocalMeansDenoising::simpleMethod()
------------------------------------------------
-Perform image denoising using Non-local Means Denoising algorithm http://www.ipol.im/pub/algo/bcm_non_local_means_denoising with several computational optimizations. Noise expected to be a gaussian white noise
-
-.. ocv:function:: void gpu::FastNonLocalMeansDenoising::simpleMethod(const GpuMat& src, GpuMat& dst, float h, int search_window = 21, int block_size = 7, Stream& s = Stream::Null())
-
- :param src: Input 8-bit 1-channel, 2-channel or 3-channel image.
-
- :param dst: Output image with the same size and type as ``src`` .
-
- :param h: Parameter regulating filter strength. Big h value perfectly removes noise but also removes image details, smaller h value preserves details but also preserves some noise
-
- :param search_window: Size in pixels of the window that is used to compute weighted average for given pixel. Should be odd. Affect performance linearly: greater search_window - greater denoising time. Recommended value 21 pixels
-
- :param block_size: Size in pixels of the template patch that is used to compute weights. Should be odd. Recommended value 7 pixels
-
- :param stream: Stream for the asynchronous invocations.
-
-This function expected to be applied to grayscale images. For colored images look at ``FastNonLocalMeansDenoising::labMethod``.
-
-.. seealso::
-
- :ocv:func:`fastNlMeansDenoising`
-
-gpu::FastNonLocalMeansDenoising::labMethod()
---------------------------------------------
-Modification of ``FastNonLocalMeansDenoising::simpleMethod`` for color images
-
-.. ocv:function:: void gpu::FastNonLocalMeansDenoising::labMethod(const GpuMat& src, GpuMat& dst, float h_luminance, float h_color, int search_window = 21, int block_size = 7, Stream& s = Stream::Null())
-
- :param src: Input 8-bit 3-channel image.
-
- :param dst: Output image with the same size and type as ``src`` .
-
- :param h_luminance: Parameter regulating filter strength. Big h value perfectly removes noise but also removes image details, smaller h value preserves details but also preserves some noise
-
- :param float: The same as h but for color components. For most images value equals 10 will be enought to remove colored noise and do not distort colors
-
- :param search_window: Size in pixels of the window that is used to compute weighted average for given pixel. Should be odd. Affect performance linearly: greater search_window - greater denoising time. Recommended value 21 pixels
-
- :param block_size: Size in pixels of the template patch that is used to compute weights. Should be odd. Recommended value 7 pixels
-
- :param stream: Stream for the asynchronous invocations.
-
-The function converts image to CIELAB colorspace and then separately denoise L and AB components with given h parameters using ``FastNonLocalMeansDenoising::simpleMethod`` function.
-
-.. seealso::
-
- :ocv:func:`fastNlMeansDenoisingColored`
-
-gpu::alphaComp
--------------------
-Composites two images using alpha opacity values contained in each image.
-
-.. ocv:function:: void gpu::alphaComp(const GpuMat& img1, const GpuMat& img2, GpuMat& dst, int alpha_op, Stream& stream = Stream::Null())
-
- :param img1: First image. Supports ``CV_8UC4`` , ``CV_16UC4`` , ``CV_32SC4`` and ``CV_32FC4`` types.
-
- :param img2: Second image. Must have the same size and the same type as ``img1`` .
-
- :param dst: Destination image.
-
- :param alpha_op: Flag specifying the alpha-blending operation:
-
- * **ALPHA_OVER**
- * **ALPHA_IN**
- * **ALPHA_OUT**
- * **ALPHA_ATOP**
- * **ALPHA_XOR**
- * **ALPHA_PLUS**
- * **ALPHA_OVER_PREMUL**
- * **ALPHA_IN_PREMUL**
- * **ALPHA_OUT_PREMUL**
- * **ALPHA_ATOP_PREMUL**
- * **ALPHA_XOR_PREMUL**
- * **ALPHA_PLUS_PREMUL**
- * **ALPHA_PREMUL**
-
- :param stream: Stream for the asynchronous version.
-
-
-
-gpu::Canny
--------------------
-Finds edges in an image using the [Canny86]_ algorithm.
-
-.. ocv:function:: void gpu::Canny(const GpuMat& image, GpuMat& edges, double low_thresh, double high_thresh, int apperture_size = 3, bool L2gradient = false)
-
-.. ocv:function:: void gpu::Canny(const GpuMat& image, CannyBuf& buf, GpuMat& edges, double low_thresh, double high_thresh, int apperture_size = 3, bool L2gradient = false)
-
-.. ocv:function:: void gpu::Canny(const GpuMat& dx, const GpuMat& dy, GpuMat& edges, double low_thresh, double high_thresh, bool L2gradient = false)
-
-.. ocv:function:: void gpu::Canny(const GpuMat& dx, const GpuMat& dy, CannyBuf& buf, GpuMat& edges, double low_thresh, double high_thresh, bool L2gradient = false)
-
- :param image: Single-channel 8-bit input image.
-
- :param dx: First derivative of image in the vertical direction. Support only ``CV_32S`` type.
-
- :param dy: First derivative of image in the horizontal direction. Support only ``CV_32S`` type.
-
- :param edges: Output edge map. It has the same size and type as ``image`` .
-
- :param low_thresh: First threshold for the hysteresis procedure.
-
- :param high_thresh: Second threshold for the hysteresis procedure.
-
- :param apperture_size: Aperture size for the :ocv:func:`Sobel` operator.
-
- :param L2gradient: Flag indicating whether a more accurate :math:`L_2` norm :math:`=\sqrt{(dI/dx)^2 + (dI/dy)^2}` should be used to compute the image gradient magnitude ( ``L2gradient=true`` ), or a faster default :math:`L_1` norm :math:`=|dI/dx|+|dI/dy|` is enough ( ``L2gradient=false`` ).
-
- :param buf: Optional buffer to avoid extra memory allocations (for many calls with the same sizes).
-
-.. seealso:: :ocv:func:`Canny`
-
-
-
-gpu::HoughLines
----------------
-Finds lines in a binary image using the classical Hough transform.
-
-.. ocv:function:: void gpu::HoughLines(const GpuMat& src, GpuMat& lines, float rho, float theta, int threshold, bool doSort = false, int maxLines = 4096)
-
-.. ocv:function:: void gpu::HoughLines(const GpuMat& src, GpuMat& lines, HoughLinesBuf& buf, float rho, float theta, int threshold, bool doSort = false, int maxLines = 4096)
-
- :param src: 8-bit, single-channel binary source image.
-
- :param lines: Output vector of lines. Each line is represented by a two-element vector :math:`(\rho, \theta)` . :math:`\rho` is the distance from the coordinate origin :math:`(0,0)` (top-left corner of the image). :math:`\theta` is the line rotation angle in radians ( :math:`0 \sim \textrm{vertical line}, \pi/2 \sim \textrm{horizontal line}` ).
-
- :param rho: Distance resolution of the accumulator in pixels.
-
- :param theta: Angle resolution of the accumulator in radians.
-
- :param threshold: Accumulator threshold parameter. Only those lines are returned that get enough votes ( :math:`>\texttt{threshold}` ).
-
- :param doSort: Performs lines sort by votes.
-
- :param maxLines: Maximum number of output lines.
-
- :param buf: Optional buffer to avoid extra memory allocations (for many calls with the same sizes).
-
-.. seealso:: :ocv:func:`HoughLines`
-
-
-
-gpu::HoughLinesDownload
------------------------
-Downloads results from :ocv:func:`gpu::HoughLines` to host memory.
-
-.. ocv:function:: void gpu::HoughLinesDownload(const GpuMat& d_lines, OutputArray h_lines, OutputArray h_votes = noArray())
-
- :param d_lines: Result of :ocv:func:`gpu::HoughLines` .
-
- :param h_lines: Output host array.
-
- :param h_votes: Optional output array for line's votes.
-
-.. seealso:: :ocv:func:`gpu::HoughLines`
-
-
-
-gpu::HoughCircles
------------------
-Finds circles in a grayscale image using the Hough transform.
-
-.. ocv:function:: void gpu::HoughCircles(const GpuMat& src, GpuMat& circles, int method, float dp, float minDist, int cannyThreshold, int votesThreshold, int minRadius, int maxRadius, int maxCircles = 4096)
-
-.. ocv:function:: void gpu::HoughCircles(const GpuMat& src, GpuMat& circles, HoughCirclesBuf& buf, int method, float dp, float minDist, int cannyThreshold, int votesThreshold, int minRadius, int maxRadius, int maxCircles = 4096)
-
- :param src: 8-bit, single-channel grayscale input image.
-
- :param circles: Output vector of found circles. Each vector is encoded as a 3-element floating-point vector :math:`(x, y, radius)` .
-
- :param method: Detection method to use. Currently, the only implemented method is ``CV_HOUGH_GRADIENT`` , which is basically *21HT* , described in [Yuen90]_.
-
- :param dp: Inverse ratio of the accumulator resolution to the image resolution. For example, if ``dp=1`` , the accumulator has the same resolution as the input image. If ``dp=2`` , the accumulator has half as big width and height.
-
- :param minDist: Minimum distance between the centers of the detected circles. If the parameter is too small, multiple neighbor circles may be falsely detected in addition to a true one. If it is too large, some circles may be missed.
-
- :param cannyThreshold: The higher threshold of the two passed to the :ocv:func:`gpu::Canny` edge detector (the lower one is twice smaller).
-
- :param votesThreshold: The accumulator threshold for the circle centers at the detection stage. The smaller it is, the more false circles may be detected.
-
- :param minRadius: Minimum circle radius.
-
- :param maxRadius: Maximum circle radius.
-
- :param maxCircles: Maximum number of output circles.
-
- :param buf: Optional buffer to avoid extra memory allocations (for many calls with the same sizes).
-
-.. seealso:: :ocv:func:`HoughCircles`
-
-
-
-gpu::HoughCirclesDownload
--------------------------
-Downloads results from :ocv:func:`gpu::HoughCircles` to host memory.
-
-.. ocv:function:: void gpu::HoughCirclesDownload(const GpuMat& d_circles, OutputArray h_circles)
-
- :param d_circles: Result of :ocv:func:`gpu::HoughCircles` .
-
- :param h_circles: Output host array.
-
-.. seealso:: :ocv:func:`gpu::HoughCircles`
+++ /dev/null
-Operations on Matrices
-======================
-
-.. highlight:: cpp
-
-
-
-gpu::gemm
-------------------
-Performs generalized matrix multiplication.
-
-.. ocv:function:: void gpu::gemm(const GpuMat& src1, const GpuMat& src2, double alpha, const GpuMat& src3, double beta, GpuMat& dst, int flags = 0, Stream& stream = Stream::Null())
-
- :param src1: First multiplied input matrix that should have ``CV_32FC1`` , ``CV_64FC1`` , ``CV_32FC2`` , or ``CV_64FC2`` type.
-
- :param src2: Second multiplied input matrix of the same type as ``src1`` .
-
- :param alpha: Weight of the matrix product.
-
- :param src3: Third optional delta matrix added to the matrix product. It should have the same type as ``src1`` and ``src2`` .
-
- :param beta: Weight of ``src3`` .
-
- :param dst: Destination matrix. It has the proper size and the same type as input matrices.
-
- :param flags: Operation flags:
-
- * **GEMM_1_T** transpose ``src1``
- * **GEMM_2_T** transpose ``src2``
- * **GEMM_3_T** transpose ``src3``
-
- :param stream: Stream for the asynchronous version.
-
-The function performs generalized matrix multiplication similar to the ``gemm`` functions in BLAS level 3. For example, ``gemm(src1, src2, alpha, src3, beta, dst, GEMM_1_T + GEMM_3_T)`` corresponds to
-
-.. math::
-
- \texttt{dst} = \texttt{alpha} \cdot \texttt{src1} ^T \cdot \texttt{src2} + \texttt{beta} \cdot \texttt{src3} ^T
-
-.. note:: Transposition operation doesn't support ``CV_64FC2`` input type.
-
-.. seealso:: :ocv:func:`gemm`
-
-
-
-gpu::transpose
-------------------
-Transposes a matrix.
-
-.. ocv:function:: void gpu::transpose( const GpuMat& src1, GpuMat& dst, Stream& stream=Stream::Null() )
-
- :param src1: Source matrix. 1-, 4-, 8-byte element sizes are supported for now (CV_8UC1, CV_8UC4, CV_16UC2, CV_32FC1, etc).
-
- :param dst: Destination matrix.
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`transpose`
-
-
-
-gpu::flip
--------------
-Flips a 2D matrix around vertical, horizontal, or both axes.
-
-.. ocv:function:: void gpu::flip( const GpuMat& a, GpuMat& b, int flipCode, Stream& stream=Stream::Null() )
-
- :param a: Source matrix. Supports 1, 3 and 4 channels images with ``CV_8U``, ``CV_16U``, ``CV_32S`` or ``CV_32F`` depth.
-
- :param b: Destination matrix.
-
- :param flipCode: Flip mode for the source:
-
- * ``0`` Flips around x-axis.
-
- * ``>0`` Flips around y-axis.
-
- * ``<0`` Flips around both axes.
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`flip`
-
-
-
-gpu::LUT
-------------
-Transforms the source matrix into the destination matrix using the given look-up table: ``dst(I) = lut(src(I))``
-
-.. ocv:function:: void gpu::LUT(const GpuMat& src, const Mat& lut, GpuMat& dst, Stream& stream = Stream::Null())
-
- :param src: Source matrix. ``CV_8UC1`` and ``CV_8UC3`` matrices are supported for now.
-
- :param lut: Look-up table of 256 elements. It is a continuous ``CV_8U`` matrix.
-
- :param dst: Destination matrix with the same depth as ``lut`` and the same number of channels as ``src`` .
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`LUT`
-
-
-
-gpu::merge
---------------
-Makes a multi-channel matrix out of several single-channel matrices.
-
-.. ocv:function:: void gpu::merge(const GpuMat* src, size_t n, GpuMat& dst, Stream& stream = Stream::Null())
-
-.. ocv:function:: void gpu::merge(const vector<GpuMat>& src, GpuMat& dst, Stream& stream = Stream::Null())
-
- :param src: Array/vector of source matrices.
-
- :param n: Number of source matrices.
-
- :param dst: Destination matrix.
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`merge`
-
-
-
-gpu::split
---------------
-Copies each plane of a multi-channel matrix into an array.
-
-.. ocv:function:: void gpu::split(const GpuMat& src, GpuMat* dst, Stream& stream = Stream::Null())
-
-.. ocv:function:: void gpu::split(const GpuMat& src, vector<GpuMat>& dst, Stream& stream = Stream::Null())
-
- :param src: Source matrix.
-
- :param dst: Destination array/vector of single-channel matrices.
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`split`
-
-
-
-gpu::magnitude
-------------------
-Computes magnitudes of complex matrix elements.
-
-.. ocv:function:: void gpu::magnitude( const GpuMat& xy, GpuMat& magnitude, Stream& stream=Stream::Null() )
-
-.. ocv:function:: void gpu::magnitude(const GpuMat& x, const GpuMat& y, GpuMat& magnitude, Stream& stream = Stream::Null())
-
- :param xy: Source complex matrix in the interleaved format ( ``CV_32FC2`` ).
-
- :param x: Source matrix containing real components ( ``CV_32FC1`` ).
-
- :param y: Source matrix containing imaginary components ( ``CV_32FC1`` ).
-
- :param magnitude: Destination matrix of float magnitudes ( ``CV_32FC1`` ).
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`magnitude`
-
-
-
-gpu::magnitudeSqr
----------------------
-Computes squared magnitudes of complex matrix elements.
-
-.. ocv:function:: void gpu::magnitudeSqr( const GpuMat& xy, GpuMat& magnitude, Stream& stream=Stream::Null() )
-
-.. ocv:function:: void gpu::magnitudeSqr(const GpuMat& x, const GpuMat& y, GpuMat& magnitude, Stream& stream = Stream::Null())
-
- :param xy: Source complex matrix in the interleaved format ( ``CV_32FC2`` ).
-
- :param x: Source matrix containing real components ( ``CV_32FC1`` ).
-
- :param y: Source matrix containing imaginary components ( ``CV_32FC1`` ).
-
- :param magnitude: Destination matrix of float magnitude squares ( ``CV_32FC1`` ).
-
- :param stream: Stream for the asynchronous version.
-
-
-
-gpu::phase
---------------
-Computes polar angles of complex matrix elements.
-
-.. ocv:function:: void gpu::phase(const GpuMat& x, const GpuMat& y, GpuMat& angle, bool angleInDegrees=false, Stream& stream = Stream::Null())
-
- :param x: Source matrix containing real components ( ``CV_32FC1`` ).
-
- :param y: Source matrix containing imaginary components ( ``CV_32FC1`` ).
-
- :param angle: Destination matrix of angles ( ``CV_32FC1`` ).
-
- :param angleInDegrees: Flag for angles that must be evaluated in degrees.
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`phase`
-
-
-
-gpu::cartToPolar
---------------------
-Converts Cartesian coordinates into polar.
-
-.. ocv:function:: void gpu::cartToPolar(const GpuMat& x, const GpuMat& y, GpuMat& magnitude, GpuMat& angle, bool angleInDegrees=false, Stream& stream = Stream::Null())
-
- :param x: Source matrix containing real components ( ``CV_32FC1`` ).
-
- :param y: Source matrix containing imaginary components ( ``CV_32FC1`` ).
-
- :param magnitude: Destination matrix of float magnitudes ( ``CV_32FC1`` ).
-
- :param angle: Destination matrix of angles ( ``CV_32FC1`` ).
-
- :param angleInDegrees: Flag for angles that must be evaluated in degrees.
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`cartToPolar`
-
-
-
-gpu::polarToCart
---------------------
-Converts polar coordinates into Cartesian.
-
-.. ocv:function:: void gpu::polarToCart(const GpuMat& magnitude, const GpuMat& angle, GpuMat& x, GpuMat& y, bool angleInDegrees=false, Stream& stream = Stream::Null())
-
- :param magnitude: Source matrix containing magnitudes ( ``CV_32FC1`` ).
-
- :param angle: Source matrix containing angles ( ``CV_32FC1`` ).
-
- :param x: Destination matrix of real components ( ``CV_32FC1`` ).
-
- :param y: Destination matrix of imaginary components ( ``CV_32FC1`` ).
-
- :param angleInDegrees: Flag that indicates angles in degrees.
-
- :param stream: Stream for the asynchronous version.
-
-.. seealso:: :ocv:func:`polarToCart`
-
-
-
-gpu::normalize
---------------
-Normalizes the norm or value range of an array.
-
-.. ocv:function:: void gpu::normalize(const GpuMat& src, GpuMat& dst, double alpha = 1, double beta = 0, int norm_type = NORM_L2, int dtype = -1, const GpuMat& mask = GpuMat())
-
-.. ocv:function:: void gpu::normalize(const GpuMat& src, GpuMat& dst, double a, double b, int norm_type, int dtype, const GpuMat& mask, GpuMat& norm_buf, GpuMat& cvt_buf)
-
- :param src: input array.
-
- :param dst: output array of the same size as ``src`` .
-
- :param alpha: norm value to normalize to or the lower range boundary in case of the range normalization.
-
- :param beta: upper range boundary in case of the range normalization; it is not used for the norm normalization.
-
- :param normType: normalization type (see the details below).
-
- :param dtype: when negative, the output array has the same type as ``src``; otherwise, it has the same number of channels as ``src`` and the depth ``=CV_MAT_DEPTH(dtype)``.
-
- :param mask: optional operation mask.
-
- :param norm_buf: Optional buffer to avoid extra memory allocations. It is resized automatically.
-
- :param cvt_buf: Optional buffer to avoid extra memory allocations. It is resized automatically.
-
-.. seealso:: :ocv:func:`normalize`
+++ /dev/null
-Video Analysis
-==============
-
-.. highlight:: cpp
-
-
-
-gpu::BroxOpticalFlow
---------------------
-.. ocv:class:: gpu::BroxOpticalFlow
-
-Class computing the optical flow for two images using Brox et al Optical Flow algorithm ([Brox2004]_). ::
-
- class BroxOpticalFlow
- {
- public:
- BroxOpticalFlow(float alpha_, float gamma_, float scale_factor_, int inner_iterations_, int outer_iterations_, int solver_iterations_);
-
- //! Compute optical flow
- //! frame0 - source frame (supports only CV_32FC1 type)
- //! frame1 - frame to track (with the same size and type as frame0)
- //! u - flow horizontal component (along x axis)
- //! v - flow vertical component (along y axis)
- void operator ()(const GpuMat& frame0, const GpuMat& frame1, GpuMat& u, GpuMat& v, Stream& stream = Stream::Null());
-
- //! flow smoothness
- float alpha;
-
- //! gradient constancy importance
- float gamma;
-
- //! pyramid scale factor
- float scale_factor;
-
- //! number of lagged non-linearity iterations (inner loop)
- int inner_iterations;
-
- //! number of warping iterations (number of pyramid levels)
- int outer_iterations;
-
- //! number of linear system solver iterations
- int solver_iterations;
-
- GpuMat buf;
- };
-
-
-
-gpu::GoodFeaturesToTrackDetector_GPU
-------------------------------------
-.. ocv:class:: gpu::GoodFeaturesToTrackDetector_GPU
-
-Class used for strong corners detection on an image. ::
-
- class GoodFeaturesToTrackDetector_GPU
- {
- public:
- explicit GoodFeaturesToTrackDetector_GPU(int maxCorners_ = 1000, double qualityLevel_ = 0.01, double minDistance_ = 0.0,
- int blockSize_ = 3, bool useHarrisDetector_ = false, double harrisK_ = 0.04);
-
- void operator ()(const GpuMat& image, GpuMat& corners, const GpuMat& mask = GpuMat());
-
- int maxCorners;
- double qualityLevel;
- double minDistance;
-
- int blockSize;
- bool useHarrisDetector;
- double harrisK;
-
- void releaseMemory();
- };
-
-The class finds the most prominent corners in the image.
-
-.. seealso:: :ocv:func:`goodFeaturesToTrack`
-
-
-
-gpu::GoodFeaturesToTrackDetector_GPU::GoodFeaturesToTrackDetector_GPU
----------------------------------------------------------------------
-Constructor.
-
-.. ocv:function:: gpu::GoodFeaturesToTrackDetector_GPU::GoodFeaturesToTrackDetector_GPU(int maxCorners = 1000, double qualityLevel = 0.01, double minDistance = 0.0, int blockSize = 3, bool useHarrisDetector = false, double harrisK = 0.04)
-
- :param maxCorners: Maximum number of corners to return. If there are more corners than are found, the strongest of them is returned.
-
- :param qualityLevel: Parameter characterizing the minimal accepted quality of image corners. The parameter value is multiplied by the best corner quality measure, which is the minimal eigenvalue (see :ocv:func:`gpu::cornerMinEigenVal` ) or the Harris function response (see :ocv:func:`gpu::cornerHarris` ). The corners with the quality measure less than the product are rejected. For example, if the best corner has the quality measure = 1500, and the ``qualityLevel=0.01`` , then all the corners with the quality measure less than 15 are rejected.
-
- :param minDistance: Minimum possible Euclidean distance between the returned corners.
-
- :param blockSize: Size of an average block for computing a derivative covariation matrix over each pixel neighborhood. See :ocv:func:`cornerEigenValsAndVecs` .
-
- :param useHarrisDetector: Parameter indicating whether to use a Harris detector (see :ocv:func:`gpu::cornerHarris`) or :ocv:func:`gpu::cornerMinEigenVal`.
-
- :param harrisK: Free parameter of the Harris detector.
-
-
-
-gpu::GoodFeaturesToTrackDetector_GPU::operator ()
--------------------------------------------------
-Finds the most prominent corners in the image.
-
-.. ocv:function:: void gpu::GoodFeaturesToTrackDetector_GPU::operator ()(const GpuMat& image, GpuMat& corners, const GpuMat& mask = GpuMat())
-
- :param image: Input 8-bit, single-channel image.
-
- :param corners: Output vector of detected corners (it will be one row matrix with CV_32FC2 type).
-
- :param mask: Optional region of interest. If the image is not empty (it needs to have the type ``CV_8UC1`` and the same size as ``image`` ), it specifies the region in which the corners are detected.
-
-.. seealso:: :ocv:func:`goodFeaturesToTrack`
-
-
-
-gpu::GoodFeaturesToTrackDetector_GPU::releaseMemory
----------------------------------------------------
-Releases inner buffers memory.
-
-.. ocv:function:: void gpu::GoodFeaturesToTrackDetector_GPU::releaseMemory()
-
-
-
-gpu::FarnebackOpticalFlow
--------------------------
-.. ocv:class:: gpu::FarnebackOpticalFlow
-
-Class computing a dense optical flow using the Gunnar Farneback’s algorithm. ::
-
- class CV_EXPORTS FarnebackOpticalFlow
- {
- public:
- FarnebackOpticalFlow()
- {
- numLevels = 5;
- pyrScale = 0.5;
- fastPyramids = false;
- winSize = 13;
- numIters = 10;
- polyN = 5;
- polySigma = 1.1;
- flags = 0;
- }
-
- int numLevels;
- double pyrScale;
- bool fastPyramids;
- int winSize;
- int numIters;
- int polyN;
- double polySigma;
- int flags;
-
- void operator ()(const GpuMat &frame0, const GpuMat &frame1, GpuMat &flowx, GpuMat &flowy, Stream &s = Stream::Null());
-
- void releaseMemory();
-
- private:
- /* hidden */
- };
-
-
-
-gpu::FarnebackOpticalFlow::operator ()
---------------------------------------
-Computes a dense optical flow using the Gunnar Farneback’s algorithm.
-
-.. ocv:function:: void gpu::FarnebackOpticalFlow::operator ()(const GpuMat &frame0, const GpuMat &frame1, GpuMat &flowx, GpuMat &flowy, Stream &s = Stream::Null())
-
- :param frame0: First 8-bit gray-scale input image
- :param frame1: Second 8-bit gray-scale input image
- :param flowx: Flow horizontal component
- :param flowy: Flow vertical component
- :param s: Stream
-
-.. seealso:: :ocv:func:`calcOpticalFlowFarneback`
-
-
-
-gpu::FarnebackOpticalFlow::releaseMemory
-----------------------------------------
-Releases unused auxiliary memory buffers.
-
-.. ocv:function:: void gpu::FarnebackOpticalFlow::releaseMemory()
-
-
-
-gpu::PyrLKOpticalFlow
----------------------
-.. ocv:class:: gpu::PyrLKOpticalFlow
-
-Class used for calculating an optical flow. ::
-
- class PyrLKOpticalFlow
- {
- public:
- PyrLKOpticalFlow();
-
- void sparse(const GpuMat& prevImg, const GpuMat& nextImg, const GpuMat& prevPts, GpuMat& nextPts,
- GpuMat& status, GpuMat* err = 0);
-
- void dense(const GpuMat& prevImg, const GpuMat& nextImg, GpuMat& u, GpuMat& v, GpuMat* err = 0);
-
- Size winSize;
- int maxLevel;
- int iters;
- bool useInitialFlow;
-
- void releaseMemory();
- };
-
-The class can calculate an optical flow for a sparse feature set or dense optical flow using the iterative Lucas-Kanade method with pyramids.
-
-.. seealso:: :ocv:func:`calcOpticalFlowPyrLK`
-
-
-
-gpu::PyrLKOpticalFlow::sparse
------------------------------
-Calculate an optical flow for a sparse feature set.
-
-.. ocv:function:: void gpu::PyrLKOpticalFlow::sparse(const GpuMat& prevImg, const GpuMat& nextImg, const GpuMat& prevPts, GpuMat& nextPts, GpuMat& status, GpuMat* err = 0)
-
- :param prevImg: First 8-bit input image (supports both grayscale and color images).
-
- :param nextImg: Second input image of the same size and the same type as ``prevImg`` .
-
- :param prevPts: Vector of 2D points for which the flow needs to be found. It must be one row matrix with CV_32FC2 type.
-
- :param nextPts: Output vector of 2D points (with single-precision floating-point coordinates) containing the calculated new positions of input features in the second image. When ``useInitialFlow`` is true, the vector must have the same size as in the input.
-
- :param status: Output status vector (CV_8UC1 type). Each element of the vector is set to 1 if the flow for the corresponding features has been found. Otherwise, it is set to 0.
-
- :param err: Output vector (CV_32FC1 type) that contains the difference between patches around the original and moved points or min eigen value if ``getMinEigenVals`` is checked. It can be NULL, if not needed.
-
-.. seealso:: :ocv:func:`calcOpticalFlowPyrLK`
-
-
-
-gpu::PyrLKOpticalFlow::dense
------------------------------
-Calculate dense optical flow.
-
-.. ocv:function:: void gpu::PyrLKOpticalFlow::dense(const GpuMat& prevImg, const GpuMat& nextImg, GpuMat& u, GpuMat& v, GpuMat* err = 0)
-
- :param prevImg: First 8-bit grayscale input image.
-
- :param nextImg: Second input image of the same size and the same type as ``prevImg`` .
-
- :param u: Horizontal component of the optical flow of the same size as input images, 32-bit floating-point, single-channel
-
- :param v: Vertical component of the optical flow of the same size as input images, 32-bit floating-point, single-channel
-
- :param err: Output vector (CV_32FC1 type) that contains the difference between patches around the original and moved points or min eigen value if ``getMinEigenVals`` is checked. It can be NULL, if not needed.
-
-
-
-gpu::PyrLKOpticalFlow::releaseMemory
-------------------------------------
-Releases inner buffers memory.
-
-.. ocv:function:: void gpu::PyrLKOpticalFlow::releaseMemory()
-
-
-
-gpu::interpolateFrames
-----------------------
-Interpolates frames (images) using provided optical flow (displacement field).
-
-.. ocv:function:: void gpu::interpolateFrames(const GpuMat& frame0, const GpuMat& frame1, const GpuMat& fu, const GpuMat& fv, const GpuMat& bu, const GpuMat& bv, float pos, GpuMat& newFrame, GpuMat& buf, Stream& stream = Stream::Null())
-
- :param frame0: First frame (32-bit floating point images, single channel).
-
- :param frame1: Second frame. Must have the same type and size as ``frame0`` .
-
- :param fu: Forward horizontal displacement.
-
- :param fv: Forward vertical displacement.
-
- :param bu: Backward horizontal displacement.
-
- :param bv: Backward vertical displacement.
-
- :param pos: New frame position.
-
- :param newFrame: Output image.
-
- :param buf: Temporary buffer, will have width x 6*height size, CV_32FC1 type and contain 6 GpuMat: occlusion masks for first frame, occlusion masks for second, interpolated forward horizontal flow, interpolated forward vertical flow, interpolated backward horizontal flow, interpolated backward vertical flow.
-
- :param stream: Stream for the asynchronous version.
-
-
-
-gpu::FGDStatModel
------------------
-.. ocv:class:: gpu::FGDStatModel
-
-Class used for background/foreground segmentation. ::
-
- class FGDStatModel
- {
- public:
- struct Params
- {
- ...
- };
-
- explicit FGDStatModel(int out_cn = 3);
- explicit FGDStatModel(const cv::gpu::GpuMat& firstFrame, const Params& params = Params(), int out_cn = 3);
-
- ~FGDStatModel();
-
- void create(const cv::gpu::GpuMat& firstFrame, const Params& params = Params());
- void release();
-
- int update(const cv::gpu::GpuMat& curFrame);
-
- //8UC3 or 8UC4 reference background image
- cv::gpu::GpuMat background;
-
- //8UC1 foreground image
- cv::gpu::GpuMat foreground;
-
- std::vector< std::vector<cv::Point> > foreground_regions;
- };
-
- The class discriminates between foreground and background pixels by building and maintaining a model of the background. Any pixel which does not fit this model is then deemed to be foreground. The class implements algorithm described in [FGD2003]_.
-
- The results are available through the class fields:
-
- .. ocv:member:: cv::gpu::GpuMat background
-
- The output background image.
-
- .. ocv:member:: cv::gpu::GpuMat foreground
-
- The output foreground mask as an 8-bit binary image.
-
- .. ocv:member:: cv::gpu::GpuMat foreground_regions
-
- The output foreground regions calculated by :ocv:func:`findContours`.
-
-
-
-gpu::FGDStatModel::FGDStatModel
--------------------------------
-Constructors.
-
-.. ocv:function:: gpu::FGDStatModel::FGDStatModel(int out_cn = 3)
-.. ocv:function:: gpu::FGDStatModel::FGDStatModel(const cv::gpu::GpuMat& firstFrame, const Params& params = Params(), int out_cn = 3)
-
- :param firstFrame: First frame from video stream. Supports 3- and 4-channels input ( ``CV_8UC3`` and ``CV_8UC4`` ).
-
- :param params: Algorithm's parameters. See [FGD2003]_ for explanation.
-
- :param out_cn: Channels count in output result and inner buffers. Can be 3 or 4. 4-channels version requires more memory, but works a bit faster.
-
-.. seealso:: :ocv:func:`gpu::FGDStatModel::create`
-
-
-
-gpu::FGDStatModel::create
--------------------------
-Initializes background model.
-
-.. ocv:function:: void gpu::FGDStatModel::create(const cv::gpu::GpuMat& firstFrame, const Params& params = Params())
-
- :param firstFrame: First frame from video stream. Supports 3- and 4-channels input ( ``CV_8UC3`` and ``CV_8UC4`` ).
-
- :param params: Algorithm's parameters. See [FGD2003]_ for explanation.
-
-
-
-gpu::FGDStatModel::release
---------------------------
-Releases all inner buffer's memory.
-
-.. ocv:function:: void gpu::FGDStatModel::release()
-
-
-
-gpu::FGDStatModel::update
---------------------------
-Updates the background model and returns foreground regions count.
-
-.. ocv:function:: int gpu::FGDStatModel::update(const cv::gpu::GpuMat& curFrame)
-
- :param curFrame: Next video frame.
-
-
-
-gpu::MOG_GPU
-------------
-.. ocv:class:: gpu::MOG_GPU
-
-Gaussian Mixture-based Backbround/Foreground Segmentation Algorithm. ::
-
- class MOG_GPU
- {
- public:
- MOG_GPU(int nmixtures = -1);
-
- void initialize(Size frameSize, int frameType);
-
- void operator()(const GpuMat& frame, GpuMat& fgmask, float learningRate = 0.0f, Stream& stream = Stream::Null());
-
- void getBackgroundImage(GpuMat& backgroundImage, Stream& stream = Stream::Null()) const;
-
- void release();
-
- int history;
- float varThreshold;
- float backgroundRatio;
- float noiseSigma;
- };
-
-The class discriminates between foreground and background pixels by building and maintaining a model of the background. Any pixel which does not fit this model is then deemed to be foreground. The class implements algorithm described in [MOG2001]_.
-
-.. seealso:: :ocv:class:`BackgroundSubtractorMOG`
-
-
-
-gpu::MOG_GPU::MOG_GPU
----------------------
-The constructor.
-
-.. ocv:function:: gpu::MOG_GPU::MOG_GPU(int nmixtures = -1)
-
- :param nmixtures: Number of Gaussian mixtures.
-
-Default constructor sets all parameters to default values.
-
-
-
-gpu::MOG_GPU::operator()
-------------------------
-Updates the background model and returns the foreground mask.
-
-.. ocv:function:: void gpu::MOG_GPU::operator()(const GpuMat& frame, GpuMat& fgmask, float learningRate = 0.0f, Stream& stream = Stream::Null())
-
- :param frame: Next video frame.
-
- :param fgmask: The output foreground mask as an 8-bit binary image.
-
- :param stream: Stream for the asynchronous version.
-
-
-
-gpu::MOG_GPU::getBackgroundImage
---------------------------------
-Computes a background image.
-
-.. ocv:function:: void gpu::MOG_GPU::getBackgroundImage(GpuMat& backgroundImage, Stream& stream = Stream::Null()) const
-
- :param backgroundImage: The output background image.
-
- :param stream: Stream for the asynchronous version.
-
-
-
-gpu::MOG_GPU::release
----------------------
-Releases all inner buffer's memory.
-
-.. ocv:function:: void gpu::MOG_GPU::release()
-
-
-
-gpu::MOG2_GPU
--------------
-.. ocv:class:: gpu::MOG2_GPU
-
-Gaussian Mixture-based Background/Foreground Segmentation Algorithm. ::
-
- class MOG2_GPU
- {
- public:
- MOG2_GPU(int nmixtures = -1);
-
- void initialize(Size frameSize, int frameType);
-
- void operator()(const GpuMat& frame, GpuMat& fgmask, float learningRate = 0.0f, Stream& stream = Stream::Null());
-
- void getBackgroundImage(GpuMat& backgroundImage, Stream& stream = Stream::Null()) const;
-
- void release();
-
- // parameters
- ...
- };
-
- The class discriminates between foreground and background pixels by building and maintaining a model of the background. Any pixel which does not fit this model is then deemed to be foreground. The class implements algorithm described in [MOG2004]_.
-
- Here are important members of the class that control the algorithm, which you can set after constructing the class instance:
-
- .. ocv:member:: float backgroundRatio
-
- Threshold defining whether the component is significant enough to be included into the background model ( corresponds to ``TB=1-cf`` from the paper??which paper??). ``cf=0.1 => TB=0.9`` is default. For ``alpha=0.001``, it means that the mode should exist for approximately 105 frames before it is considered foreground.
-
- .. ocv:member:: float varThreshold
-
- Threshold for the squared Mahalanobis distance that helps decide when a sample is close to the existing components (corresponds to ``Tg``). If it is not close to any component, a new component is generated. ``3 sigma => Tg=3*3=9`` is default. A smaller ``Tg`` value generates more components. A higher ``Tg`` value may result in a small number of components but they can grow too large.
-
- .. ocv:member:: float fVarInit
-
- Initial variance for the newly generated components. It affects the speed of adaptation. The parameter value is based on your estimate of the typical standard deviation from the images. OpenCV uses 15 as a reasonable value.
-
- .. ocv:member:: float fVarMin
-
- Parameter used to further control the variance.
-
- .. ocv:member:: float fVarMax
-
- Parameter used to further control the variance.
-
- .. ocv:member:: float fCT
-
- Complexity reduction parameter. This parameter defines the number of samples needed to accept to prove the component exists. ``CT=0.05`` is a default value for all the samples. By setting ``CT=0`` you get an algorithm very similar to the standard Stauffer&Grimson algorithm.
-
- .. ocv:member:: uchar nShadowDetection
-
- The value for marking shadow pixels in the output foreground mask. Default value is 127.
-
- .. ocv:member:: float fTau
-
- Shadow threshold. The shadow is detected if the pixel is a darker version of the background. ``Tau`` is a threshold defining how much darker the shadow can be. ``Tau= 0.5`` means that if a pixel is more than twice darker then it is not shadow. See [ShadowDetect2003]_.
-
- .. ocv:member:: bool bShadowDetection
-
- Parameter defining whether shadow detection should be enabled.
-
-.. seealso:: :ocv:class:`BackgroundSubtractorMOG2`
-
-
-
-gpu::MOG2_GPU::MOG2_GPU
------------------------
-The constructor.
-
-.. ocv:function:: gpu::MOG2_GPU::MOG2_GPU(int nmixtures = -1)
-
- :param nmixtures: Number of Gaussian mixtures.
-
-Default constructor sets all parameters to default values.
-
-
-
-gpu::MOG2_GPU::operator()
--------------------------
-Updates the background model and returns the foreground mask.
-
-.. ocv:function:: void gpu::MOG2_GPU::operator()( const GpuMat& frame, GpuMat& fgmask, float learningRate=-1.0f, Stream& stream=Stream::Null() )
-
- :param frame: Next video frame.
-
- :param fgmask: The output foreground mask as an 8-bit binary image.
-
- :param stream: Stream for the asynchronous version.
-
-
-
-gpu::MOG2_GPU::getBackgroundImage
----------------------------------
-Computes a background image.
-
-.. ocv:function:: void gpu::MOG2_GPU::getBackgroundImage(GpuMat& backgroundImage, Stream& stream = Stream::Null()) const
-
- :param backgroundImage: The output background image.
-
- :param stream: Stream for the asynchronous version.
-
-
-
-gpu::MOG2_GPU::release
-----------------------
-Releases all inner buffer's memory.
-
-.. ocv:function:: void gpu::MOG2_GPU::release()
-
-
-
-gpu::GMG_GPU
-------------
-.. ocv:class:: gpu::GMG_GPU
-
- Class used for background/foreground segmentation. ::
-
- class GMG_GPU_GPU
- {
- public:
- GMG_GPU();
-
- void initialize(Size frameSize, float min = 0.0f, float max = 255.0f);
-
- void operator ()(const GpuMat& frame, GpuMat& fgmask, float learningRate = -1.0f, Stream& stream = Stream::Null());
-
- void release();
-
- int maxFeatures;
- float learningRate;
- int numInitializationFrames;
- int quantizationLevels;
- float backgroundPrior;
- float decisionThreshold;
- int smoothingRadius;
-
- ...
- };
-
- The class discriminates between foreground and background pixels by building and maintaining a model of the background. Any pixel which does not fit this model is then deemed to be foreground. The class implements algorithm described in [GMG2012]_.
-
- Here are important members of the class that control the algorithm, which you can set after constructing the class instance:
-
- .. ocv:member:: int maxFeatures
-
- Total number of distinct colors to maintain in histogram.
-
- .. ocv:member:: float learningRate
-
- Set between 0.0 and 1.0, determines how quickly features are "forgotten" from histograms.
-
- .. ocv:member:: int numInitializationFrames
-
- Number of frames of video to use to initialize histograms.
-
- .. ocv:member:: int quantizationLevels
-
- Number of discrete levels in each channel to be used in histograms.
-
- .. ocv:member:: float backgroundPrior
-
- Prior probability that any given pixel is a background pixel. A sensitivity parameter.
-
- .. ocv:member:: float decisionThreshold
-
- Value above which pixel is determined to be FG.
-
- .. ocv:member:: float smoothingRadius
-
- Smoothing radius, in pixels, for cleaning up FG image.
-
-
-
-gpu::GMG_GPU::GMG_GPU
----------------------
-The default constructor.
-
-.. ocv:function:: gpu::GMG_GPU::GMG_GPU()
-
-Default constructor sets all parameters to default values.
-
-
-
-gpu::GMG_GPU::initialize
-------------------------
-Initialize background model and allocates all inner buffers.
-
-.. ocv:function:: void gpu::GMG_GPU::initialize(Size frameSize, float min = 0.0f, float max = 255.0f)
-
- :param frameSize: Input frame size.
-
- :param min: Minimum value taken on by pixels in image sequence. Usually 0.
-
- :param max: Maximum value taken on by pixels in image sequence, e.g. 1.0 or 255.
-
-
-
-gpu::GMG_GPU::operator()
-------------------------
-Updates the background model and returns the foreground mask
-
-.. ocv:function:: void gpu::GMG_GPU::operator ()( const GpuMat& frame, GpuMat& fgmask, float learningRate=-1.0f, Stream& stream=Stream::Null() )
-
- :param frame: Next video frame.
-
- :param fgmask: The output foreground mask as an 8-bit binary image.
-
- :param stream: Stream for the asynchronous version.
-
-
-
-gpu::GMG_GPU::release
----------------------
-Releases all inner buffer's memory.
-
-.. ocv:function:: void gpu::GMG_GPU::release()
-
-
-
-gpu::VideoWriter_GPU
----------------------
-Video writer class.
-
-.. ocv:class:: gpu::VideoWriter_GPU
-
-The class uses H264 video codec.
-
-.. note:: Currently only Windows platform is supported.
-
-
-
-gpu::VideoWriter_GPU::VideoWriter_GPU
--------------------------------------
-Constructors.
-
-.. ocv:function:: gpu::VideoWriter_GPU::VideoWriter_GPU()
-.. ocv:function:: gpu::VideoWriter_GPU::VideoWriter_GPU(const String& fileName, cv::Size frameSize, double fps, SurfaceFormat format = SF_BGR)
-.. ocv:function:: gpu::VideoWriter_GPU::VideoWriter_GPU(const String& fileName, cv::Size frameSize, double fps, const EncoderParams& params, SurfaceFormat format = SF_BGR)
-.. ocv:function:: gpu::VideoWriter_GPU::VideoWriter_GPU(const cv::Ptr<EncoderCallBack>& encoderCallback, cv::Size frameSize, double fps, SurfaceFormat format = SF_BGR)
-.. ocv:function:: gpu::VideoWriter_GPU::VideoWriter_GPU(const cv::Ptr<EncoderCallBack>& encoderCallback, cv::Size frameSize, double fps, const EncoderParams& params, SurfaceFormat format = SF_BGR)
-
- :param fileName: Name of the output video file. Only AVI file format is supported.
-
- :param frameSize: Size of the input video frames.
-
- :param fps: Framerate of the created video stream.
-
- :param params: Encoder parameters. See :ocv:struct:`gpu::VideoWriter_GPU::EncoderParams` .
-
- :param format: Surface format of input frames ( ``SF_UYVY`` , ``SF_YUY2`` , ``SF_YV12`` , ``SF_NV12`` , ``SF_IYUV`` , ``SF_BGR`` or ``SF_GRAY``). BGR or gray frames will be converted to YV12 format before encoding, frames with other formats will be used as is.
-
- :param encoderCallback: Callbacks for video encoder. See :ocv:class:`gpu::VideoWriter_GPU::EncoderCallBack` . Use it if you want to work with raw video stream.
-
-The constructors initialize video writer. FFMPEG is used to write videos. User can implement own multiplexing with :ocv:class:`gpu::VideoWriter_GPU::EncoderCallBack` .
-
-
-
-gpu::VideoWriter_GPU::open
---------------------------
-Initializes or reinitializes video writer.
-
-.. ocv:function:: void gpu::VideoWriter_GPU::open(const String& fileName, cv::Size frameSize, double fps, SurfaceFormat format = SF_BGR)
-.. ocv:function:: void gpu::VideoWriter_GPU::open(const String& fileName, cv::Size frameSize, double fps, const EncoderParams& params, SurfaceFormat format = SF_BGR)
-.. ocv:function:: void gpu::VideoWriter_GPU::open(const cv::Ptr<EncoderCallBack>& encoderCallback, cv::Size frameSize, double fps, SurfaceFormat format = SF_BGR)
-.. ocv:function:: void gpu::VideoWriter_GPU::open(const cv::Ptr<EncoderCallBack>& encoderCallback, cv::Size frameSize, double fps, const EncoderParams& params, SurfaceFormat format = SF_BGR)
-
-The method opens video writer. Parameters are the same as in the constructor :ocv:func:`gpu::VideoWriter_GPU::VideoWriter_GPU` . The method throws :ocv:class:`Exception` if error occurs.
-
-
-
-gpu::VideoWriter_GPU::isOpened
-------------------------------
-Returns true if video writer has been successfully initialized.
-
-.. ocv:function:: bool gpu::VideoWriter_GPU::isOpened() const
-
-
-
-gpu::VideoWriter_GPU::close
----------------------------
-Releases the video writer.
-
-.. ocv:function:: void gpu::VideoWriter_GPU::close()
-
-
-
-gpu::VideoWriter_GPU::write
----------------------------
-Writes the next video frame.
-
-.. ocv:function:: void gpu::VideoWriter_GPU::write(const cv::gpu::GpuMat& image, bool lastFrame = false)
-
- :param image: The written frame.
-
- :param lastFrame: Indicates that it is end of stream. The parameter can be ignored.
-
-The method write the specified image to video file. The image must have the same size and the same surface format as has been specified when opening the video writer.
-
-
-
-gpu::VideoWriter_GPU::EncoderParams
------------------------------------
-.. ocv:struct:: gpu::VideoWriter_GPU::EncoderParams
-
-Different parameters for CUDA video encoder. ::
-
- struct EncoderParams
- {
- int P_Interval; // NVVE_P_INTERVAL,
- int IDR_Period; // NVVE_IDR_PERIOD,
- int DynamicGOP; // NVVE_DYNAMIC_GOP,
- int RCType; // NVVE_RC_TYPE,
- int AvgBitrate; // NVVE_AVG_BITRATE,
- int PeakBitrate; // NVVE_PEAK_BITRATE,
- int QP_Level_Intra; // NVVE_QP_LEVEL_INTRA,
- int QP_Level_InterP; // NVVE_QP_LEVEL_INTER_P,
- int QP_Level_InterB; // NVVE_QP_LEVEL_INTER_B,
- int DeblockMode; // NVVE_DEBLOCK_MODE,
- int ProfileLevel; // NVVE_PROFILE_LEVEL,
- int ForceIntra; // NVVE_FORCE_INTRA,
- int ForceIDR; // NVVE_FORCE_IDR,
- int ClearStat; // NVVE_CLEAR_STAT,
- int DIMode; // NVVE_SET_DEINTERLACE,
- int Presets; // NVVE_PRESETS,
- int DisableCabac; // NVVE_DISABLE_CABAC,
- int NaluFramingType; // NVVE_CONFIGURE_NALU_FRAMING_TYPE
- int DisableSPSPPS; // NVVE_DISABLE_SPS_PPS
-
- EncoderParams();
- explicit EncoderParams(const String& configFile);
-
- void load(const String& configFile);
- void save(const String& configFile) const;
- };
-
-
-
-gpu::VideoWriter_GPU::EncoderParams::EncoderParams
---------------------------------------------------
-Constructors.
-
-.. ocv:function:: gpu::VideoWriter_GPU::EncoderParams::EncoderParams()
-.. ocv:function:: gpu::VideoWriter_GPU::EncoderParams::EncoderParams(const String& configFile)
-
- :param configFile: Config file name.
-
-Creates default parameters or reads parameters from config file.
-
-
-
-gpu::VideoWriter_GPU::EncoderParams::load
------------------------------------------
-Reads parameters from config file.
-
-.. ocv:function:: void gpu::VideoWriter_GPU::EncoderParams::load(const String& configFile)
-
- :param configFile: Config file name.
-
-
-
-gpu::VideoWriter_GPU::EncoderParams::save
------------------------------------------
-Saves parameters to config file.
-
-.. ocv:function:: void gpu::VideoWriter_GPU::EncoderParams::save(const String& configFile) const
-
- :param configFile: Config file name.
-
-
-
-gpu::VideoWriter_GPU::EncoderCallBack
--------------------------------------
-.. ocv:class:: gpu::VideoWriter_GPU::EncoderCallBack
-
-Callbacks for CUDA video encoder. ::
-
- class EncoderCallBack
- {
- public:
- enum PicType
- {
- IFRAME = 1,
- PFRAME = 2,
- BFRAME = 3
- };
-
- virtual ~EncoderCallBack() {}
-
- virtual unsigned char* acquireBitStream(int* bufferSize) = 0;
- virtual void releaseBitStream(unsigned char* data, int size) = 0;
- virtual void onBeginFrame(int frameNumber, PicType picType) = 0;
- virtual void onEndFrame(int frameNumber, PicType picType) = 0;
- };
-
-
-
-gpu::VideoWriter_GPU::EncoderCallBack::acquireBitStream
--------------------------------------------------------
-Callback function to signal the start of bitstream that is to be encoded.
-
-.. ocv:function:: virtual uchar* gpu::VideoWriter_GPU::EncoderCallBack::acquireBitStream(int* bufferSize) = 0
-
-Callback must allocate buffer for CUDA encoder and return pointer to it and it's size.
-
-
-
-gpu::VideoWriter_GPU::EncoderCallBack::releaseBitStream
--------------------------------------------------------
-Callback function to signal that the encoded bitstream is ready to be written to file.
-
-.. ocv:function:: virtual void gpu::VideoWriter_GPU::EncoderCallBack::releaseBitStream(unsigned char* data, int size) = 0
-
-
-
-gpu::VideoWriter_GPU::EncoderCallBack::onBeginFrame
----------------------------------------------------
-Callback function to signal that the encoding operation on the frame has started.
-
-.. ocv:function:: virtual void gpu::VideoWriter_GPU::EncoderCallBack::onBeginFrame(int frameNumber, PicType picType) = 0
-
- :param picType: Specify frame type (I-Frame, P-Frame or B-Frame).
-
-
-
-gpu::VideoWriter_GPU::EncoderCallBack::onEndFrame
--------------------------------------------------
-Callback function signals that the encoding operation on the frame has finished.
-
-.. ocv:function:: virtual void gpu::VideoWriter_GPU::EncoderCallBack::onEndFrame(int frameNumber, PicType picType) = 0
-
- :param picType: Specify frame type (I-Frame, P-Frame or B-Frame).
-
-
-
-gpu::VideoReader_GPU
---------------------
-Class for reading video from files.
-
-.. ocv:class:: gpu::VideoReader_GPU
-
-.. note:: Currently only Windows and Linux platforms are supported.
-
-
-
-gpu::VideoReader_GPU::Codec
----------------------------
-
-Video codecs supported by :ocv:class:`gpu::VideoReader_GPU` .
-
-.. ocv:enum:: gpu::VideoReader_GPU::Codec
-
- .. ocv:emember:: MPEG1 = 0
- .. ocv:emember:: MPEG2
- .. ocv:emember:: MPEG4
- .. ocv:emember:: VC1
- .. ocv:emember:: H264
- .. ocv:emember:: JPEG
- .. ocv:emember:: H264_SVC
- .. ocv:emember:: H264_MVC
-
- .. ocv:emember:: Uncompressed_YUV420 = (('I'<<24)|('Y'<<16)|('U'<<8)|('V'))
-
- Y,U,V (4:2:0)
-
- .. ocv:emember:: Uncompressed_YV12 = (('Y'<<24)|('V'<<16)|('1'<<8)|('2'))
-
- Y,V,U (4:2:0)
-
- .. ocv:emember:: Uncompressed_NV12 = (('N'<<24)|('V'<<16)|('1'<<8)|('2'))
-
- Y,UV (4:2:0)
-
- .. ocv:emember:: Uncompressed_YUYV = (('Y'<<24)|('U'<<16)|('Y'<<8)|('V'))
-
- YUYV/YUY2 (4:2:2)
-
- .. ocv:emember:: Uncompressed_UYVY = (('U'<<24)|('Y'<<16)|('V'<<8)|('Y'))
-
- UYVY (4:2:2)
-
-
-gpu::VideoReader_GPU::ChromaFormat
-----------------------------------
-
-Chroma formats supported by :ocv:class:`gpu::VideoReader_GPU` .
-
-.. ocv:enum:: gpu::VideoReader_GPU::ChromaFormat
-
- .. ocv:emember:: Monochrome = 0
- .. ocv:emember:: YUV420
- .. ocv:emember:: YUV422
- .. ocv:emember:: YUV444
-
-
-gpu::VideoReader_GPU::FormatInfo
---------------------------------
-.. ocv:struct:: gpu::VideoReader_GPU::FormatInfo
-
-Struct providing information about video file format. ::
-
- struct FormatInfo
- {
- Codec codec;
- ChromaFormat chromaFormat;
- int width;
- int height;
- };
-
-
-gpu::VideoReader_GPU::VideoReader_GPU
--------------------------------------
-Constructors.
-
-.. ocv:function:: gpu::VideoReader_GPU::VideoReader_GPU()
-.. ocv:function:: gpu::VideoReader_GPU::VideoReader_GPU(const String& filename)
-.. ocv:function:: gpu::VideoReader_GPU::VideoReader_GPU(const cv::Ptr<VideoSource>& source)
-
- :param filename: Name of the input video file.
-
- :param source: Video file parser implemented by user.
-
-The constructors initialize video reader. FFMPEG is used to read videos. User can implement own demultiplexing with :ocv:class:`gpu::VideoReader_GPU::VideoSource` .
-
-
-
-gpu::VideoReader_GPU::open
---------------------------
-Initializes or reinitializes video reader.
-
-.. ocv:function:: void gpu::VideoReader_GPU::open(const String& filename)
-.. ocv:function:: void gpu::VideoReader_GPU::open(const cv::Ptr<VideoSource>& source)
-
-The method opens video reader. Parameters are the same as in the constructor :ocv:func:`gpu::VideoReader_GPU::VideoReader_GPU` . The method throws :ocv:class:`Exception` if error occurs.
-
-
-
-gpu::VideoReader_GPU::isOpened
-------------------------------
-Returns true if video reader has been successfully initialized.
-
-.. ocv:function:: bool gpu::VideoReader_GPU::isOpened() const
-
-
-
-gpu::VideoReader_GPU::close
----------------------------
-Releases the video reader.
-
-.. ocv:function:: void gpu::VideoReader_GPU::close()
-
-
-
-gpu::VideoReader_GPU::read
---------------------------
-Grabs, decodes and returns the next video frame.
-
-.. ocv:function:: bool gpu::VideoReader_GPU::read(GpuMat& image)
-
-If no frames has been grabbed (there are no more frames in video file), the methods return ``false`` . The method throws :ocv:class:`Exception` if error occurs.
-
-
-
-gpu::VideoReader_GPU::format
-----------------------------
-Returns information about video file format.
-
-.. ocv:function:: FormatInfo gpu::VideoReader_GPU::format() const
-
-The method throws :ocv:class:`Exception` if video reader wasn't initialized.
-
-
-
-gpu::VideoReader_GPU::dumpFormat
---------------------------------
-Dump information about video file format to specified stream.
-
-.. ocv:function:: void gpu::VideoReader_GPU::dumpFormat(std::ostream& st)
-
- :param st: Output stream.
-
-The method throws :ocv:class:`Exception` if video reader wasn't initialized.
-
-
-
-gpu::VideoReader_GPU::VideoSource
------------------------------------
-.. ocv:class:: gpu::VideoReader_GPU::VideoSource
-
-Interface for video demultiplexing. ::
-
- class VideoSource
- {
- public:
- VideoSource();
- virtual ~VideoSource() {}
-
- virtual FormatInfo format() const = 0;
- virtual void start() = 0;
- virtual void stop() = 0;
- virtual bool isStarted() const = 0;
- virtual bool hasError() const = 0;
-
- protected:
- bool parseVideoData(const unsigned char* data, size_t size, bool endOfStream = false);
- };
-
-User can implement own demultiplexing by implementing this interface.
-
-
-
-gpu::VideoReader_GPU::VideoSource::format
------------------------------------------
-Returns information about video file format.
-
-.. ocv:function:: virtual FormatInfo gpu::VideoReader_GPU::VideoSource::format() const = 0
-
-
-
-gpu::VideoReader_GPU::VideoSource::start
-----------------------------------------
-Starts processing.
-
-.. ocv:function:: virtual void gpu::VideoReader_GPU::VideoSource::start() = 0
-
-Implementation must create own thread with video processing and call periodic :ocv:func:`gpu::VideoReader_GPU::VideoSource::parseVideoData` .
-
-
-
-gpu::VideoReader_GPU::VideoSource::stop
----------------------------------------
-Stops processing.
-
-.. ocv:function:: virtual void gpu::VideoReader_GPU::VideoSource::stop() = 0
-
-
-
-gpu::VideoReader_GPU::VideoSource::isStarted
---------------------------------------------
-Returns ``true`` if processing was successfully started.
-
-.. ocv:function:: virtual bool gpu::VideoReader_GPU::VideoSource::isStarted() const = 0
-
-
-
-gpu::VideoReader_GPU::VideoSource::hasError
--------------------------------------------
-Returns ``true`` if error occured during processing.
-
-.. ocv:function:: virtual bool gpu::VideoReader_GPU::VideoSource::hasError() const = 0
-
-
-
-gpu::VideoReader_GPU::VideoSource::parseVideoData
--------------------------------------------------
-Parse next video frame. Implementation must call this method after new frame was grabbed.
-
-.. ocv:function:: bool gpu::VideoReader_GPU::VideoSource::parseVideoData(const uchar* data, size_t size, bool endOfStream = false)
-
- :param data: Pointer to frame data. Can be ``NULL`` if ``endOfStream`` if ``true`` .
-
- :param size: Size in bytes of current frame.
-
- :param endOfStream: Indicates that it is end of stream.
-
-
-
-.. [Brox2004] T. Brox, A. Bruhn, N. Papenberg, J. Weickert. *High accuracy optical flow estimation based on a theory for warping*. ECCV 2004.
-.. [FGD2003] Liyuan Li, Weimin Huang, Irene Y.H. Gu, and Qi Tian. *Foreground Object Detection from Videos Containing Complex Background*. ACM MM2003 9p, 2003.
-.. [MOG2001] P. KadewTraKuPong and R. Bowden. *An improved adaptive background mixture model for real-time tracking with shadow detection*. Proc. 2nd European Workshop on Advanced Video-Based Surveillance Systems, 2001
-.. [MOG2004] Z. Zivkovic. *Improved adaptive Gausian mixture model for background subtraction*. International Conference Pattern Recognition, UK, August, 2004
-.. [ShadowDetect2003] Prati, Mikic, Trivedi and Cucchiarra. *Detecting Moving Shadows...*. IEEE PAMI, 2003
-.. [GMG2012] A. Godbehere, A. Matsukawa and K. Goldberg. *Visual Tracking of Human Visitors under Variable-Lighting Conditions for a Responsive Audio Art Installation*. American Control Conference, Montreal, June 2012
#ifndef __OPENCV_GPU_HPP__
#define __OPENCV_GPU_HPP__
-#ifndef SKIP_INCLUDES
-#include <vector>
-#include <memory>
-#include <iosfwd>
+#ifndef __cplusplus
+# error gpu.hpp header must be compiled as C++
#endif
#include "opencv2/core/gpumat.hpp"
-#include "opencv2/imgproc.hpp"
-#include "opencv2/objdetect.hpp"
-#include "opencv2/features2d.hpp"
-namespace cv { namespace gpu {
-//////////////////////////////// Filter Engine ////////////////////////////////
-
-/*!
-The Base Class for 1D or Row-wise Filters
-
-This is the base class for linear or non-linear filters that process 1D data.
-In particular, such filters are used for the "horizontal" filtering parts in separable filters.
-*/
-class CV_EXPORTS BaseRowFilter_GPU
-{
-public:
- BaseRowFilter_GPU(int ksize_, int anchor_) : ksize(ksize_), anchor(anchor_) {}
- virtual ~BaseRowFilter_GPU() {}
- virtual void operator()(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null()) = 0;
- int ksize, anchor;
-};
-
-/*!
-The Base Class for Column-wise Filters
-
-This is the base class for linear or non-linear filters that process columns of 2D arrays.
-Such filters are used for the "vertical" filtering parts in separable filters.
-*/
-class CV_EXPORTS BaseColumnFilter_GPU
-{
-public:
- BaseColumnFilter_GPU(int ksize_, int anchor_) : ksize(ksize_), anchor(anchor_) {}
- virtual ~BaseColumnFilter_GPU() {}
- virtual void operator()(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null()) = 0;
- int ksize, anchor;
-};
-
-/*!
-The Base Class for Non-Separable 2D Filters.
-
-This is the base class for linear or non-linear 2D filters.
-*/
-class CV_EXPORTS BaseFilter_GPU
-{
-public:
- BaseFilter_GPU(const Size& ksize_, const Point& anchor_) : ksize(ksize_), anchor(anchor_) {}
- virtual ~BaseFilter_GPU() {}
- virtual void operator()(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null()) = 0;
- Size ksize;
- Point anchor;
-};
-
-/*!
-The Base Class for Filter Engine.
-
-The class can be used to apply an arbitrary filtering operation to an image.
-It contains all the necessary intermediate buffers.
-*/
-class CV_EXPORTS FilterEngine_GPU
-{
-public:
- virtual ~FilterEngine_GPU() {}
-
- virtual void apply(const GpuMat& src, GpuMat& dst, Rect roi = Rect(0,0,-1,-1), Stream& stream = Stream::Null()) = 0;
-};
-
-//! returns the non-separable filter engine with the specified filter
-CV_EXPORTS Ptr<FilterEngine_GPU> createFilter2D_GPU(const Ptr<BaseFilter_GPU>& filter2D, int srcType, int dstType);
-
-//! returns the separable filter engine with the specified filters
-CV_EXPORTS Ptr<FilterEngine_GPU> createSeparableFilter_GPU(const Ptr<BaseRowFilter_GPU>& rowFilter,
- const Ptr<BaseColumnFilter_GPU>& columnFilter, int srcType, int bufType, int dstType);
-CV_EXPORTS Ptr<FilterEngine_GPU> createSeparableFilter_GPU(const Ptr<BaseRowFilter_GPU>& rowFilter,
- const Ptr<BaseColumnFilter_GPU>& columnFilter, int srcType, int bufType, int dstType, GpuMat& buf);
-
-//! returns horizontal 1D box filter
-//! supports only CV_8UC1 source type and CV_32FC1 sum type
-CV_EXPORTS Ptr<BaseRowFilter_GPU> getRowSumFilter_GPU(int srcType, int sumType, int ksize, int anchor = -1);
-
-//! returns vertical 1D box filter
-//! supports only CV_8UC1 sum type and CV_32FC1 dst type
-CV_EXPORTS Ptr<BaseColumnFilter_GPU> getColumnSumFilter_GPU(int sumType, int dstType, int ksize, int anchor = -1);
-
-//! returns 2D box filter
-//! supports CV_8UC1 and CV_8UC4 source type, dst type must be the same as source type
-CV_EXPORTS Ptr<BaseFilter_GPU> getBoxFilter_GPU(int srcType, int dstType, const Size& ksize, Point anchor = Point(-1, -1));
-
-//! returns box filter engine
-CV_EXPORTS Ptr<FilterEngine_GPU> createBoxFilter_GPU(int srcType, int dstType, const Size& ksize,
- const Point& anchor = Point(-1,-1));
-
-//! returns 2D morphological filter
-//! only MORPH_ERODE and MORPH_DILATE are supported
-//! supports CV_8UC1 and CV_8UC4 types
-//! kernel must have CV_8UC1 type, one rows and cols == ksize.width * ksize.height
-CV_EXPORTS Ptr<BaseFilter_GPU> getMorphologyFilter_GPU(int op, int type, const Mat& kernel, const Size& ksize,
- Point anchor=Point(-1,-1));
-
-//! returns morphological filter engine. Only MORPH_ERODE and MORPH_DILATE are supported.
-CV_EXPORTS Ptr<FilterEngine_GPU> createMorphologyFilter_GPU(int op, int type, const Mat& kernel,
- const Point& anchor = Point(-1,-1), int iterations = 1);
-CV_EXPORTS Ptr<FilterEngine_GPU> createMorphologyFilter_GPU(int op, int type, const Mat& kernel, GpuMat& buf,
- const Point& anchor = Point(-1,-1), int iterations = 1);
-
-//! returns 2D filter with the specified kernel
-//! supports CV_8U, CV_16U and CV_32F one and four channel image
-CV_EXPORTS Ptr<BaseFilter_GPU> getLinearFilter_GPU(int srcType, int dstType, const Mat& kernel, Point anchor = Point(-1, -1), int borderType = BORDER_DEFAULT);
-
-//! returns the non-separable linear filter engine
-CV_EXPORTS Ptr<FilterEngine_GPU> createLinearFilter_GPU(int srcType, int dstType, const Mat& kernel,
- Point anchor = Point(-1,-1), int borderType = BORDER_DEFAULT);
-
-//! returns the primitive row filter with the specified kernel.
-//! supports only CV_8UC1, CV_8UC4, CV_16SC1, CV_16SC2, CV_32SC1, CV_32FC1 source type.
-//! there are two version of algorithm: NPP and OpenCV.
-//! NPP calls when srcType == CV_8UC1 or srcType == CV_8UC4 and bufType == srcType,
-//! otherwise calls OpenCV version.
-//! NPP supports only BORDER_CONSTANT border type.
-//! OpenCV version supports only CV_32F as buffer depth and
-//! BORDER_REFLECT101, BORDER_REPLICATE and BORDER_CONSTANT border types.
-CV_EXPORTS Ptr<BaseRowFilter_GPU> getLinearRowFilter_GPU(int srcType, int bufType, const Mat& rowKernel,
- int anchor = -1, int borderType = BORDER_DEFAULT);
-
-//! returns the primitive column filter with the specified kernel.
-//! supports only CV_8UC1, CV_8UC4, CV_16SC1, CV_16SC2, CV_32SC1, CV_32FC1 dst type.
-//! there are two version of algorithm: NPP and OpenCV.
-//! NPP calls when dstType == CV_8UC1 or dstType == CV_8UC4 and bufType == dstType,
-//! otherwise calls OpenCV version.
-//! NPP supports only BORDER_CONSTANT border type.
-//! OpenCV version supports only CV_32F as buffer depth and
-//! BORDER_REFLECT101, BORDER_REPLICATE and BORDER_CONSTANT border types.
-CV_EXPORTS Ptr<BaseColumnFilter_GPU> getLinearColumnFilter_GPU(int bufType, int dstType, const Mat& columnKernel,
- int anchor = -1, int borderType = BORDER_DEFAULT);
-
-//! returns the separable linear filter engine
-CV_EXPORTS Ptr<FilterEngine_GPU> createSeparableLinearFilter_GPU(int srcType, int dstType, const Mat& rowKernel,
- const Mat& columnKernel, const Point& anchor = Point(-1,-1), int rowBorderType = BORDER_DEFAULT,
- int columnBorderType = -1);
-CV_EXPORTS Ptr<FilterEngine_GPU> createSeparableLinearFilter_GPU(int srcType, int dstType, const Mat& rowKernel,
- const Mat& columnKernel, GpuMat& buf, const Point& anchor = Point(-1,-1), int rowBorderType = BORDER_DEFAULT,
- int columnBorderType = -1);
-
-//! returns filter engine for the generalized Sobel operator
-CV_EXPORTS Ptr<FilterEngine_GPU> createDerivFilter_GPU(int srcType, int dstType, int dx, int dy, int ksize,
- int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
-CV_EXPORTS Ptr<FilterEngine_GPU> createDerivFilter_GPU(int srcType, int dstType, int dx, int dy, int ksize, GpuMat& buf,
- int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
-
-//! returns the Gaussian filter engine
-CV_EXPORTS Ptr<FilterEngine_GPU> createGaussianFilter_GPU(int type, Size ksize, double sigma1, double sigma2 = 0,
- int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
-CV_EXPORTS Ptr<FilterEngine_GPU> createGaussianFilter_GPU(int type, Size ksize, GpuMat& buf, double sigma1, double sigma2 = 0,
- int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
-
-//! returns maximum filter
-CV_EXPORTS Ptr<BaseFilter_GPU> getMaxFilter_GPU(int srcType, int dstType, const Size& ksize, Point anchor = Point(-1,-1));
-
-//! returns minimum filter
-CV_EXPORTS Ptr<BaseFilter_GPU> getMinFilter_GPU(int srcType, int dstType, const Size& ksize, Point anchor = Point(-1,-1));
-
-//! smooths the image using the normalized box filter
-//! supports CV_8UC1, CV_8UC4 types
-CV_EXPORTS void boxFilter(const GpuMat& src, GpuMat& dst, int ddepth, Size ksize, Point anchor = Point(-1,-1), Stream& stream = Stream::Null());
-
-//! a synonym for normalized box filter
-static inline void blur(const GpuMat& src, GpuMat& dst, Size ksize, Point anchor = Point(-1,-1), Stream& stream = Stream::Null())
-{
- boxFilter(src, dst, -1, ksize, anchor, stream);
-}
-
-//! erodes the image (applies the local minimum operator)
-CV_EXPORTS void erode(const GpuMat& src, GpuMat& dst, const Mat& kernel, Point anchor = Point(-1, -1), int iterations = 1);
-CV_EXPORTS void erode(const GpuMat& src, GpuMat& dst, const Mat& kernel, GpuMat& buf,
- Point anchor = Point(-1, -1), int iterations = 1,
- Stream& stream = Stream::Null());
-
-//! dilates the image (applies the local maximum operator)
-CV_EXPORTS void dilate(const GpuMat& src, GpuMat& dst, const Mat& kernel, Point anchor = Point(-1, -1), int iterations = 1);
-CV_EXPORTS void dilate(const GpuMat& src, GpuMat& dst, const Mat& kernel, GpuMat& buf,
- Point anchor = Point(-1, -1), int iterations = 1,
- Stream& stream = Stream::Null());
-
-//! applies an advanced morphological operation to the image
-CV_EXPORTS void morphologyEx(const GpuMat& src, GpuMat& dst, int op, const Mat& kernel, Point anchor = Point(-1, -1), int iterations = 1);
-CV_EXPORTS void morphologyEx(const GpuMat& src, GpuMat& dst, int op, const Mat& kernel, GpuMat& buf1, GpuMat& buf2,
- Point anchor = Point(-1, -1), int iterations = 1, Stream& stream = Stream::Null());
-
-//! applies non-separable 2D linear filter to the image
-CV_EXPORTS void filter2D(const GpuMat& src, GpuMat& dst, int ddepth, const Mat& kernel, Point anchor=Point(-1,-1), int borderType = BORDER_DEFAULT, Stream& stream = Stream::Null());
-
-//! applies separable 2D linear filter to the image
-CV_EXPORTS void sepFilter2D(const GpuMat& src, GpuMat& dst, int ddepth, const Mat& kernelX, const Mat& kernelY,
- Point anchor = Point(-1,-1), int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
-CV_EXPORTS void sepFilter2D(const GpuMat& src, GpuMat& dst, int ddepth, const Mat& kernelX, const Mat& kernelY, GpuMat& buf,
- Point anchor = Point(-1,-1), int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1,
- Stream& stream = Stream::Null());
-
-//! applies generalized Sobel operator to the image
-CV_EXPORTS void Sobel(const GpuMat& src, GpuMat& dst, int ddepth, int dx, int dy, int ksize = 3, double scale = 1,
- int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
-CV_EXPORTS void Sobel(const GpuMat& src, GpuMat& dst, int ddepth, int dx, int dy, GpuMat& buf, int ksize = 3, double scale = 1,
- int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1, Stream& stream = Stream::Null());
-
-//! applies the vertical or horizontal Scharr operator to the image
-CV_EXPORTS void Scharr(const GpuMat& src, GpuMat& dst, int ddepth, int dx, int dy, double scale = 1,
- int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
-CV_EXPORTS void Scharr(const GpuMat& src, GpuMat& dst, int ddepth, int dx, int dy, GpuMat& buf, double scale = 1,
- int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1, Stream& stream = Stream::Null());
-
-//! smooths the image using Gaussian filter.
-CV_EXPORTS void GaussianBlur(const GpuMat& src, GpuMat& dst, Size ksize, double sigma1, double sigma2 = 0,
- int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
-CV_EXPORTS void GaussianBlur(const GpuMat& src, GpuMat& dst, Size ksize, GpuMat& buf, double sigma1, double sigma2 = 0,
- int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1, Stream& stream = Stream::Null());
-
-//! applies Laplacian operator to the image
-//! supports only ksize = 1 and ksize = 3
-CV_EXPORTS void Laplacian(const GpuMat& src, GpuMat& dst, int ddepth, int ksize = 1, double scale = 1, int borderType = BORDER_DEFAULT, Stream& stream = Stream::Null());
-
-
-////////////////////////////// Arithmetics ///////////////////////////////////
-
-//! implements generalized matrix product algorithm GEMM from BLAS
-CV_EXPORTS void gemm(const GpuMat& src1, const GpuMat& src2, double alpha,
- const GpuMat& src3, double beta, GpuMat& dst, int flags = 0, Stream& stream = Stream::Null());
-
-//! transposes the matrix
-//! supports matrix with element size = 1, 4 and 8 bytes (CV_8UC1, CV_8UC4, CV_16UC2, CV_32FC1, etc)
-CV_EXPORTS void transpose(const GpuMat& src1, GpuMat& dst, Stream& stream = Stream::Null());
-
-//! reverses the order of the rows, columns or both in a matrix
-//! supports 1, 3 and 4 channels images with CV_8U, CV_16U, CV_32S or CV_32F depth
-CV_EXPORTS void flip(const GpuMat& a, GpuMat& b, int flipCode, Stream& stream = Stream::Null());
-
-//! transforms 8-bit unsigned integers using lookup table: dst(i)=lut(src(i))
-//! destination array will have the depth type as lut and the same channels number as source
-//! supports CV_8UC1, CV_8UC3 types
-CV_EXPORTS void LUT(const GpuMat& src, const Mat& lut, GpuMat& dst, Stream& stream = Stream::Null());
-
-//! makes multi-channel array out of several single-channel arrays
-CV_EXPORTS void merge(const GpuMat* src, size_t n, GpuMat& dst, Stream& stream = Stream::Null());
-
-//! makes multi-channel array out of several single-channel arrays
-CV_EXPORTS void merge(const std::vector<GpuMat>& src, GpuMat& dst, Stream& stream = Stream::Null());
-
-//! copies each plane of a multi-channel array to a dedicated array
-CV_EXPORTS void split(const GpuMat& src, GpuMat* dst, Stream& stream = Stream::Null());
-
-//! copies each plane of a multi-channel array to a dedicated array
-CV_EXPORTS void split(const GpuMat& src, std::vector<GpuMat>& dst, Stream& stream = Stream::Null());
-
-//! computes magnitude of complex (x(i).re, x(i).im) vector
-//! supports only CV_32FC2 type
-CV_EXPORTS void magnitude(const GpuMat& xy, GpuMat& magnitude, Stream& stream = Stream::Null());
-
-//! computes squared magnitude of complex (x(i).re, x(i).im) vector
-//! supports only CV_32FC2 type
-CV_EXPORTS void magnitudeSqr(const GpuMat& xy, GpuMat& magnitude, Stream& stream = Stream::Null());
-
-//! computes magnitude of each (x(i), y(i)) vector
-//! supports only floating-point source
-CV_EXPORTS void magnitude(const GpuMat& x, const GpuMat& y, GpuMat& magnitude, Stream& stream = Stream::Null());
-
-//! computes squared magnitude of each (x(i), y(i)) vector
-//! supports only floating-point source
-CV_EXPORTS void magnitudeSqr(const GpuMat& x, const GpuMat& y, GpuMat& magnitude, Stream& stream = Stream::Null());
-
-//! computes angle (angle(i)) of each (x(i), y(i)) vector
-//! supports only floating-point source
-CV_EXPORTS void phase(const GpuMat& x, const GpuMat& y, GpuMat& angle, bool angleInDegrees = false, Stream& stream = Stream::Null());
-
-//! converts Cartesian coordinates to polar
-//! supports only floating-point source
-CV_EXPORTS void cartToPolar(const GpuMat& x, const GpuMat& y, GpuMat& magnitude, GpuMat& angle, bool angleInDegrees = false, Stream& stream = Stream::Null());
-
-//! converts polar coordinates to Cartesian
-//! supports only floating-point source
-CV_EXPORTS void polarToCart(const GpuMat& magnitude, const GpuMat& angle, GpuMat& x, GpuMat& y, bool angleInDegrees = false, Stream& stream = Stream::Null());
-
-//! scales and shifts array elements so that either the specified norm (alpha) or the minimum (alpha) and maximum (beta) array values get the specified values
-CV_EXPORTS void normalize(const GpuMat& src, GpuMat& dst, double alpha = 1, double beta = 0,
- int norm_type = NORM_L2, int dtype = -1, const GpuMat& mask = GpuMat());
-CV_EXPORTS void normalize(const GpuMat& src, GpuMat& dst, double a, double b,
- int norm_type, int dtype, const GpuMat& mask, GpuMat& norm_buf, GpuMat& cvt_buf);
-
-
-//////////////////////////// Per-element operations ////////////////////////////////////
-
-//! adds one matrix to another (c = a + b)
-CV_EXPORTS void add(const GpuMat& a, const GpuMat& b, GpuMat& c, const GpuMat& mask = GpuMat(), int dtype = -1, Stream& stream = Stream::Null());
-//! adds scalar to a matrix (c = a + s)
-CV_EXPORTS void add(const GpuMat& a, const Scalar& sc, GpuMat& c, const GpuMat& mask = GpuMat(), int dtype = -1, Stream& stream = Stream::Null());
-
-//! subtracts one matrix from another (c = a - b)
-CV_EXPORTS void subtract(const GpuMat& a, const GpuMat& b, GpuMat& c, const GpuMat& mask = GpuMat(), int dtype = -1, Stream& stream = Stream::Null());
-//! subtracts scalar from a matrix (c = a - s)
-CV_EXPORTS void subtract(const GpuMat& a, const Scalar& sc, GpuMat& c, const GpuMat& mask = GpuMat(), int dtype = -1, Stream& stream = Stream::Null());
-
-//! computes element-wise weighted product of the two arrays (c = scale * a * b)
-CV_EXPORTS void multiply(const GpuMat& a, const GpuMat& b, GpuMat& c, double scale = 1, int dtype = -1, Stream& stream = Stream::Null());
-//! weighted multiplies matrix to a scalar (c = scale * a * s)
-CV_EXPORTS void multiply(const GpuMat& a, const Scalar& sc, GpuMat& c, double scale = 1, int dtype = -1, Stream& stream = Stream::Null());
-
-//! computes element-wise weighted quotient of the two arrays (c = a / b)
-CV_EXPORTS void divide(const GpuMat& a, const GpuMat& b, GpuMat& c, double scale = 1, int dtype = -1, Stream& stream = Stream::Null());
-//! computes element-wise weighted quotient of matrix and scalar (c = a / s)
-CV_EXPORTS void divide(const GpuMat& a, const Scalar& sc, GpuMat& c, double scale = 1, int dtype = -1, Stream& stream = Stream::Null());
-//! computes element-wise weighted reciprocal of an array (dst = scale/src2)
-CV_EXPORTS void divide(double scale, const GpuMat& b, GpuMat& c, int dtype = -1, Stream& stream = Stream::Null());
-
-//! computes the weighted sum of two arrays (dst = alpha*src1 + beta*src2 + gamma)
-CV_EXPORTS void addWeighted(const GpuMat& src1, double alpha, const GpuMat& src2, double beta, double gamma, GpuMat& dst,
- int dtype = -1, Stream& stream = Stream::Null());
-
-//! adds scaled array to another one (dst = alpha*src1 + src2)
-static inline void scaleAdd(const GpuMat& src1, double alpha, const GpuMat& src2, GpuMat& dst, Stream& stream = Stream::Null())
-{
- addWeighted(src1, alpha, src2, 1.0, 0.0, dst, -1, stream);
-}
-
-//! computes element-wise absolute difference of two arrays (c = abs(a - b))
-CV_EXPORTS void absdiff(const GpuMat& a, const GpuMat& b, GpuMat& c, Stream& stream = Stream::Null());
-//! computes element-wise absolute difference of array and scalar (c = abs(a - s))
-CV_EXPORTS void absdiff(const GpuMat& a, const Scalar& s, GpuMat& c, Stream& stream = Stream::Null());
-
-//! computes absolute value of each matrix element
-//! supports CV_16S and CV_32F depth
-CV_EXPORTS void abs(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null());
-
-//! computes square of each pixel in an image
-//! supports CV_8U, CV_16U, CV_16S and CV_32F depth
-CV_EXPORTS void sqr(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null());
-
-//! computes square root of each pixel in an image
-//! supports CV_8U, CV_16U, CV_16S and CV_32F depth
-CV_EXPORTS void sqrt(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null());
-
-//! computes exponent of each matrix element (b = e**a)
-//! supports CV_8U, CV_16U, CV_16S and CV_32F depth
-CV_EXPORTS void exp(const GpuMat& a, GpuMat& b, Stream& stream = Stream::Null());
-
-//! computes natural logarithm of absolute value of each matrix element: b = log(abs(a))
-//! supports CV_8U, CV_16U, CV_16S and CV_32F depth
-CV_EXPORTS void log(const GpuMat& a, GpuMat& b, Stream& stream = Stream::Null());
-
-//! computes power of each matrix element:
-// (dst(i,j) = pow( src(i,j) , power), if src.type() is integer
-// (dst(i,j) = pow(fabs(src(i,j)), power), otherwise
-//! supports all, except depth == CV_64F
-CV_EXPORTS void pow(const GpuMat& src, double power, GpuMat& dst, Stream& stream = Stream::Null());
-
-//! compares elements of two arrays (c = a <cmpop> b)
-CV_EXPORTS void compare(const GpuMat& a, const GpuMat& b, GpuMat& c, int cmpop, Stream& stream = Stream::Null());
-CV_EXPORTS void compare(const GpuMat& a, Scalar sc, GpuMat& c, int cmpop, Stream& stream = Stream::Null());
-
-//! performs per-elements bit-wise inversion
-CV_EXPORTS void bitwise_not(const GpuMat& src, GpuMat& dst, const GpuMat& mask=GpuMat(), Stream& stream = Stream::Null());
-
-//! calculates per-element bit-wise disjunction of two arrays
-CV_EXPORTS void bitwise_or(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, const GpuMat& mask=GpuMat(), Stream& stream = Stream::Null());
-//! calculates per-element bit-wise disjunction of array and scalar
-//! supports 1, 3 and 4 channels images with CV_8U, CV_16U or CV_32S depth
-CV_EXPORTS void bitwise_or(const GpuMat& src1, const Scalar& sc, GpuMat& dst, Stream& stream = Stream::Null());
-
-//! calculates per-element bit-wise conjunction of two arrays
-CV_EXPORTS void bitwise_and(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, const GpuMat& mask=GpuMat(), Stream& stream = Stream::Null());
-//! calculates per-element bit-wise conjunction of array and scalar
-//! supports 1, 3 and 4 channels images with CV_8U, CV_16U or CV_32S depth
-CV_EXPORTS void bitwise_and(const GpuMat& src1, const Scalar& sc, GpuMat& dst, Stream& stream = Stream::Null());
-
-//! calculates per-element bit-wise "exclusive or" operation
-CV_EXPORTS void bitwise_xor(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, const GpuMat& mask=GpuMat(), Stream& stream = Stream::Null());
-//! calculates per-element bit-wise "exclusive or" of array and scalar
-//! supports 1, 3 and 4 channels images with CV_8U, CV_16U or CV_32S depth
-CV_EXPORTS void bitwise_xor(const GpuMat& src1, const Scalar& sc, GpuMat& dst, Stream& stream = Stream::Null());
-
-//! pixel by pixel right shift of an image by a constant value
-//! supports 1, 3 and 4 channels images with integers elements
-CV_EXPORTS void rshift(const GpuMat& src, Scalar_<int> sc, GpuMat& dst, Stream& stream = Stream::Null());
-
-//! pixel by pixel left shift of an image by a constant value
-//! supports 1, 3 and 4 channels images with CV_8U, CV_16U or CV_32S depth
-CV_EXPORTS void lshift(const GpuMat& src, Scalar_<int> sc, GpuMat& dst, Stream& stream = Stream::Null());
-
-//! computes per-element minimum of two arrays (dst = min(src1, src2))
-CV_EXPORTS void min(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, Stream& stream = Stream::Null());
-
-//! computes per-element minimum of array and scalar (dst = min(src1, src2))
-CV_EXPORTS void min(const GpuMat& src1, double src2, GpuMat& dst, Stream& stream = Stream::Null());
-
-//! computes per-element maximum of two arrays (dst = max(src1, src2))
-CV_EXPORTS void max(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, Stream& stream = Stream::Null());
-
-//! computes per-element maximum of array and scalar (dst = max(src1, src2))
-CV_EXPORTS void max(const GpuMat& src1, double src2, GpuMat& dst, Stream& stream = Stream::Null());
-
-enum { ALPHA_OVER, ALPHA_IN, ALPHA_OUT, ALPHA_ATOP, ALPHA_XOR, ALPHA_PLUS, ALPHA_OVER_PREMUL, ALPHA_IN_PREMUL, ALPHA_OUT_PREMUL,
- ALPHA_ATOP_PREMUL, ALPHA_XOR_PREMUL, ALPHA_PLUS_PREMUL, ALPHA_PREMUL};
-
-//! Composite two images using alpha opacity values contained in each image
-//! Supports CV_8UC4, CV_16UC4, CV_32SC4 and CV_32FC4 types
-CV_EXPORTS void alphaComp(const GpuMat& img1, const GpuMat& img2, GpuMat& dst, int alpha_op, Stream& stream = Stream::Null());
-
-
-////////////////////////////// Image processing //////////////////////////////
-
-//! DST[x,y] = SRC[xmap[x,y],ymap[x,y]]
-//! supports only CV_32FC1 map type
-CV_EXPORTS void remap(const GpuMat& src, GpuMat& dst, const GpuMat& xmap, const GpuMat& ymap,
- int interpolation, int borderMode = BORDER_CONSTANT, Scalar borderValue = Scalar(),
- Stream& stream = Stream::Null());
-
-//! Does mean shift filtering on GPU.
-CV_EXPORTS void meanShiftFiltering(const GpuMat& src, GpuMat& dst, int sp, int sr,
- TermCriteria criteria = TermCriteria(TermCriteria::MAX_ITER + TermCriteria::EPS, 5, 1),
- Stream& stream = Stream::Null());
-
-//! Does mean shift procedure on GPU.
-CV_EXPORTS void meanShiftProc(const GpuMat& src, GpuMat& dstr, GpuMat& dstsp, int sp, int sr,
- TermCriteria criteria = TermCriteria(TermCriteria::MAX_ITER + TermCriteria::EPS, 5, 1),
- Stream& stream = Stream::Null());
-
-//! Does mean shift segmentation with elimination of small regions.
-CV_EXPORTS void meanShiftSegmentation(const GpuMat& src, Mat& dst, int sp, int sr, int minsize,
- TermCriteria criteria = TermCriteria(TermCriteria::MAX_ITER + TermCriteria::EPS, 5, 1));
-
-//! Does coloring of disparity image: [0..ndisp) -> [0..240, 1, 1] in HSV.
-//! Supported types of input disparity: CV_8U, CV_16S.
-//! Output disparity has CV_8UC4 type in BGRA format (alpha = 255).
-CV_EXPORTS void drawColorDisp(const GpuMat& src_disp, GpuMat& dst_disp, int ndisp, Stream& stream = Stream::Null());
-
-//! Reprojects disparity image to 3D space.
-//! Supports CV_8U and CV_16S types of input disparity.
-//! The output is a 3- or 4-channel floating-point matrix.
-//! Each element of this matrix will contain the 3D coordinates of the point (x,y,z,1), computed from the disparity map.
-//! Q is the 4x4 perspective transformation matrix that can be obtained with cvStereoRectify.
-CV_EXPORTS void reprojectImageTo3D(const GpuMat& disp, GpuMat& xyzw, const Mat& Q, int dst_cn = 4, Stream& stream = Stream::Null());
-
-//! converts image from one color space to another
-CV_EXPORTS void cvtColor(const GpuMat& src, GpuMat& dst, int code, int dcn = 0, Stream& stream = Stream::Null());
-
-enum
-{
- // Bayer Demosaicing (Malvar, He, and Cutler)
- COLOR_BayerBG2BGR_MHT = 256,
- COLOR_BayerGB2BGR_MHT = 257,
- COLOR_BayerRG2BGR_MHT = 258,
- COLOR_BayerGR2BGR_MHT = 259,
-
- COLOR_BayerBG2RGB_MHT = COLOR_BayerRG2BGR_MHT,
- COLOR_BayerGB2RGB_MHT = COLOR_BayerGR2BGR_MHT,
- COLOR_BayerRG2RGB_MHT = COLOR_BayerBG2BGR_MHT,
- COLOR_BayerGR2RGB_MHT = COLOR_BayerGB2BGR_MHT,
-
- COLOR_BayerBG2GRAY_MHT = 260,
- COLOR_BayerGB2GRAY_MHT = 261,
- COLOR_BayerRG2GRAY_MHT = 262,
- COLOR_BayerGR2GRAY_MHT = 263
-};
-CV_EXPORTS void demosaicing(const GpuMat& src, GpuMat& dst, int code, int dcn = -1, Stream& stream = Stream::Null());
-
-//! swap channels
-//! dstOrder - Integer array describing how channel values are permutated. The n-th entry
-//! of the array contains the number of the channel that is stored in the n-th channel of
-//! the output image. E.g. Given an RGBA image, aDstOrder = [3,2,1,0] converts this to ABGR
-//! channel order.
-CV_EXPORTS void swapChannels(GpuMat& image, const int dstOrder[4], Stream& stream = Stream::Null());
-
-//! Routines for correcting image color gamma
-CV_EXPORTS void gammaCorrection(const GpuMat& src, GpuMat& dst, bool forward = true, Stream& stream = Stream::Null());
-
-//! applies fixed threshold to the image
-CV_EXPORTS double threshold(const GpuMat& src, GpuMat& dst, double thresh, double maxval, int type, Stream& stream = Stream::Null());
-
-//! resizes the image
-//! Supports INTER_NEAREST, INTER_LINEAR, INTER_CUBIC, INTER_AREA
-CV_EXPORTS void resize(const GpuMat& src, GpuMat& dst, Size dsize, double fx=0, double fy=0, int interpolation = INTER_LINEAR, Stream& stream = Stream::Null());
-
-//! warps the image using affine transformation
-//! Supports INTER_NEAREST, INTER_LINEAR, INTER_CUBIC
-CV_EXPORTS void warpAffine(const GpuMat& src, GpuMat& dst, const Mat& M, Size dsize, int flags = INTER_LINEAR,
- int borderMode = BORDER_CONSTANT, Scalar borderValue = Scalar(), Stream& stream = Stream::Null());
-
-CV_EXPORTS void buildWarpAffineMaps(const Mat& M, bool inverse, Size dsize, GpuMat& xmap, GpuMat& ymap, Stream& stream = Stream::Null());
-
-//! warps the image using perspective transformation
-//! Supports INTER_NEAREST, INTER_LINEAR, INTER_CUBIC
-CV_EXPORTS void warpPerspective(const GpuMat& src, GpuMat& dst, const Mat& M, Size dsize, int flags = INTER_LINEAR,
- int borderMode = BORDER_CONSTANT, Scalar borderValue = Scalar(), Stream& stream = Stream::Null());
-
-CV_EXPORTS void buildWarpPerspectiveMaps(const Mat& M, bool inverse, Size dsize, GpuMat& xmap, GpuMat& ymap, Stream& stream = Stream::Null());
-
-//! builds plane warping maps
-CV_EXPORTS void buildWarpPlaneMaps(Size src_size, Rect dst_roi, const Mat &K, const Mat& R, const Mat &T, float scale,
- GpuMat& map_x, GpuMat& map_y, Stream& stream = Stream::Null());
-
-//! builds cylindrical warping maps
-CV_EXPORTS void buildWarpCylindricalMaps(Size src_size, Rect dst_roi, const Mat &K, const Mat& R, float scale,
- GpuMat& map_x, GpuMat& map_y, Stream& stream = Stream::Null());
-
-//! builds spherical warping maps
-CV_EXPORTS void buildWarpSphericalMaps(Size src_size, Rect dst_roi, const Mat &K, const Mat& R, float scale,
- GpuMat& map_x, GpuMat& map_y, Stream& stream = Stream::Null());
-
-//! rotates an image around the origin (0,0) and then shifts it
-//! supports INTER_NEAREST, INTER_LINEAR, INTER_CUBIC
-//! supports 1, 3 or 4 channels images with CV_8U, CV_16U or CV_32F depth
-CV_EXPORTS void rotate(const GpuMat& src, GpuMat& dst, Size dsize, double angle, double xShift = 0, double yShift = 0,
- int interpolation = INTER_LINEAR, Stream& stream = Stream::Null());
-
-//! copies 2D array to a larger destination array and pads borders with user-specifiable constant
-CV_EXPORTS void copyMakeBorder(const GpuMat& src, GpuMat& dst, int top, int bottom, int left, int right, int borderType,
- const Scalar& value = Scalar(), Stream& stream = Stream::Null());
-
-//! computes the integral image
-//! sum will have CV_32S type, but will contain unsigned int values
-//! supports only CV_8UC1 source type
-CV_EXPORTS void integral(const GpuMat& src, GpuMat& sum, Stream& stream = Stream::Null());
-//! buffered version
-CV_EXPORTS void integralBuffered(const GpuMat& src, GpuMat& sum, GpuMat& buffer, Stream& stream = Stream::Null());
-
-//! computes squared integral image
-//! result matrix will have 64F type, but will contain 64U values
-//! supports source images of 8UC1 type only
-CV_EXPORTS void sqrIntegral(const GpuMat& src, GpuMat& sqsum, Stream& stream = Stream::Null());
-
-//! computes vertical sum, supports only CV_32FC1 images
-CV_EXPORTS void columnSum(const GpuMat& src, GpuMat& sum);
-
-//! computes the standard deviation of integral images
-//! supports only CV_32SC1 source type and CV_32FC1 sqr type
-//! output will have CV_32FC1 type
-CV_EXPORTS void rectStdDev(const GpuMat& src, const GpuMat& sqr, GpuMat& dst, const Rect& rect, Stream& stream = Stream::Null());
-
-//! computes Harris cornerness criteria at each image pixel
-CV_EXPORTS void cornerHarris(const GpuMat& src, GpuMat& dst, int blockSize, int ksize, double k, int borderType = BORDER_REFLECT101);
-CV_EXPORTS void cornerHarris(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, int blockSize, int ksize, double k, int borderType = BORDER_REFLECT101);
-CV_EXPORTS void cornerHarris(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, GpuMat& buf, int blockSize, int ksize, double k,
- int borderType = BORDER_REFLECT101, Stream& stream = Stream::Null());
-
-//! computes minimum eigen value of 2x2 derivative covariation matrix at each pixel - the cornerness criteria
-CV_EXPORTS void cornerMinEigenVal(const GpuMat& src, GpuMat& dst, int blockSize, int ksize, int borderType=BORDER_REFLECT101);
-CV_EXPORTS void cornerMinEigenVal(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, int blockSize, int ksize, int borderType=BORDER_REFLECT101);
-CV_EXPORTS void cornerMinEigenVal(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, GpuMat& buf, int blockSize, int ksize,
- int borderType=BORDER_REFLECT101, Stream& stream = Stream::Null());
-
-//! performs per-element multiplication of two full (not packed) Fourier spectrums
-//! supports 32FC2 matrixes only (interleaved format)
-CV_EXPORTS void mulSpectrums(const GpuMat& a, const GpuMat& b, GpuMat& c, int flags, bool conjB=false, Stream& stream = Stream::Null());
-
-//! performs per-element multiplication of two full (not packed) Fourier spectrums
-//! supports 32FC2 matrixes only (interleaved format)
-CV_EXPORTS void mulAndScaleSpectrums(const GpuMat& a, const GpuMat& b, GpuMat& c, int flags, float scale, bool conjB=false, Stream& stream = Stream::Null());
-
-//! Performs a forward or inverse discrete Fourier transform (1D or 2D) of floating point matrix.
-//! Param dft_size is the size of DFT transform.
-//!
-//! If the source matrix is not continous, then additional copy will be done,
-//! so to avoid copying ensure the source matrix is continous one. If you want to use
-//! preallocated output ensure it is continuous too, otherwise it will be reallocated.
-//!
-//! Being implemented via CUFFT real-to-complex transform result contains only non-redundant values
-//! in CUFFT's format. Result as full complex matrix for such kind of transform cannot be retrieved.
-//!
-//! For complex-to-real transform it is assumed that the source matrix is packed in CUFFT's format.
-CV_EXPORTS void dft(const GpuMat& src, GpuMat& dst, Size dft_size, int flags=0, Stream& stream = Stream::Null());
-
-struct CV_EXPORTS ConvolveBuf
-{
- Size result_size;
- Size block_size;
- Size user_block_size;
- Size dft_size;
- int spect_len;
-
- GpuMat image_spect, templ_spect, result_spect;
- GpuMat image_block, templ_block, result_data;
-
- void create(Size image_size, Size templ_size);
- static Size estimateBlockSize(Size result_size, Size templ_size);
-};
-
-
-//! computes convolution (or cross-correlation) of two images using discrete Fourier transform
-//! supports source images of 32FC1 type only
-//! result matrix will have 32FC1 type
-CV_EXPORTS void convolve(const GpuMat& image, const GpuMat& templ, GpuMat& result, bool ccorr = false);
-CV_EXPORTS void convolve(const GpuMat& image, const GpuMat& templ, GpuMat& result, bool ccorr, ConvolveBuf& buf, Stream& stream = Stream::Null());
-
-struct CV_EXPORTS MatchTemplateBuf
-{
- Size user_block_size;
- GpuMat imagef, templf;
- std::vector<GpuMat> images;
- std::vector<GpuMat> image_sums;
- std::vector<GpuMat> image_sqsums;
-};
-
-//! computes the proximity map for the raster template and the image where the template is searched for
-CV_EXPORTS void matchTemplate(const GpuMat& image, const GpuMat& templ, GpuMat& result, int method, Stream &stream = Stream::Null());
-
-//! computes the proximity map for the raster template and the image where the template is searched for
-CV_EXPORTS void matchTemplate(const GpuMat& image, const GpuMat& templ, GpuMat& result, int method, MatchTemplateBuf &buf, Stream& stream = Stream::Null());
-
-//! smoothes the source image and downsamples it
-CV_EXPORTS void pyrDown(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null());
-
-//! upsamples the source image and then smoothes it
-CV_EXPORTS void pyrUp(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null());
-
-//! performs linear blending of two images
-//! to avoid accuracy errors sum of weigths shouldn't be very close to zero
-CV_EXPORTS void blendLinear(const GpuMat& img1, const GpuMat& img2, const GpuMat& weights1, const GpuMat& weights2,
- GpuMat& result, Stream& stream = Stream::Null());
-
-//! Performa bilateral filtering of passsed image
-CV_EXPORTS void bilateralFilter(const GpuMat& src, GpuMat& dst, int kernel_size, float sigma_color, float sigma_spatial,
- int borderMode = BORDER_DEFAULT, Stream& stream = Stream::Null());
-
-//! Brute force non-local means algorith (slow but universal)
-CV_EXPORTS void nonLocalMeans(const GpuMat& src, GpuMat& dst, float h, int search_window = 21, int block_size = 7, int borderMode = BORDER_DEFAULT, Stream& s = Stream::Null());
-
-//! Fast (but approximate)version of non-local means algorith similar to CPU function (running sums technique)
-class CV_EXPORTS FastNonLocalMeansDenoising
-{
-public:
- //! Simple method, recommended for grayscale images (though it supports multichannel images)
- void simpleMethod(const GpuMat& src, GpuMat& dst, float h, int search_window = 21, int block_size = 7, Stream& s = Stream::Null());
-
- //! Processes luminance and color components separatelly
- void labMethod(const GpuMat& src, GpuMat& dst, float h_luminance, float h_color, int search_window = 21, int block_size = 7, Stream& s = Stream::Null());
-
-private:
-
- GpuMat buffer, extended_src_buffer;
- GpuMat lab, l, ab;
-};
-
-struct CV_EXPORTS CannyBuf
-{
- void create(const Size& image_size, int apperture_size = 3);
- void release();
-
- GpuMat dx, dy;
- GpuMat mag;
- GpuMat map;
- GpuMat st1, st2;
- Ptr<FilterEngine_GPU> filterDX, filterDY;
-};
-
-CV_EXPORTS void Canny(const GpuMat& image, GpuMat& edges, double low_thresh, double high_thresh, int apperture_size = 3, bool L2gradient = false);
-CV_EXPORTS void Canny(const GpuMat& image, CannyBuf& buf, GpuMat& edges, double low_thresh, double high_thresh, int apperture_size = 3, bool L2gradient = false);
-CV_EXPORTS void Canny(const GpuMat& dx, const GpuMat& dy, GpuMat& edges, double low_thresh, double high_thresh, bool L2gradient = false);
-CV_EXPORTS void Canny(const GpuMat& dx, const GpuMat& dy, CannyBuf& buf, GpuMat& edges, double low_thresh, double high_thresh, bool L2gradient = false);
-
-class CV_EXPORTS ImagePyramid
-{
-public:
- inline ImagePyramid() : nLayers_(0) {}
- inline ImagePyramid(const GpuMat& img, int nLayers, Stream& stream = Stream::Null())
- {
- build(img, nLayers, stream);
- }
-
- void build(const GpuMat& img, int nLayers, Stream& stream = Stream::Null());
-
- void getLayer(GpuMat& outImg, Size outRoi, Stream& stream = Stream::Null()) const;
-
- inline void release()
- {
- layer0_.release();
- pyramid_.clear();
- nLayers_ = 0;
- }
-
-private:
- GpuMat layer0_;
- std::vector<GpuMat> pyramid_;
- int nLayers_;
-};
-
-//! HoughLines
-
-struct HoughLinesBuf
-{
- GpuMat accum;
- GpuMat list;
-};
-
-CV_EXPORTS void HoughLines(const GpuMat& src, GpuMat& lines, float rho, float theta, int threshold, bool doSort = false, int maxLines = 4096);
-CV_EXPORTS void HoughLines(const GpuMat& src, GpuMat& lines, HoughLinesBuf& buf, float rho, float theta, int threshold, bool doSort = false, int maxLines = 4096);
-CV_EXPORTS void HoughLinesDownload(const GpuMat& d_lines, OutputArray h_lines, OutputArray h_votes = noArray());
-
-//! HoughLinesP
-
-//! finds line segments in the black-n-white image using probabalistic Hough transform
-CV_EXPORTS void HoughLinesP(const GpuMat& image, GpuMat& lines, HoughLinesBuf& buf, float rho, float theta, int minLineLength, int maxLineGap, int maxLines = 4096);
-
-//! HoughCircles
-
-struct HoughCirclesBuf
-{
- GpuMat edges;
- GpuMat accum;
- GpuMat list;
- CannyBuf cannyBuf;
-};
-
-CV_EXPORTS void HoughCircles(const GpuMat& src, GpuMat& circles, int method, float dp, float minDist, int cannyThreshold, int votesThreshold, int minRadius, int maxRadius, int maxCircles = 4096);
-CV_EXPORTS void HoughCircles(const GpuMat& src, GpuMat& circles, HoughCirclesBuf& buf, int method, float dp, float minDist, int cannyThreshold, int votesThreshold, int minRadius, int maxRadius, int maxCircles = 4096);
-CV_EXPORTS void HoughCirclesDownload(const GpuMat& d_circles, OutputArray h_circles);
-
-//! finds arbitrary template in the grayscale image using Generalized Hough Transform
-//! Ballard, D.H. (1981). Generalizing the Hough transform to detect arbitrary shapes. Pattern Recognition 13 (2): 111-122.
-//! Guil, N., González-Linares, J.M. and Zapata, E.L. (1999). Bidimensional shape detection using an invariant approach. Pattern Recognition 32 (6): 1025-1038.
-class CV_EXPORTS GeneralizedHough_GPU : public cv::Algorithm
-{
-public:
- static Ptr<GeneralizedHough_GPU> create(int method);
-
- virtual ~GeneralizedHough_GPU();
-
- //! set template to search
- void setTemplate(const GpuMat& templ, int cannyThreshold = 100, Point templCenter = Point(-1, -1));
- void setTemplate(const GpuMat& edges, const GpuMat& dx, const GpuMat& dy, Point templCenter = Point(-1, -1));
-
- //! find template on image
- void detect(const GpuMat& image, GpuMat& positions, int cannyThreshold = 100);
- void detect(const GpuMat& edges, const GpuMat& dx, const GpuMat& dy, GpuMat& positions);
-
- void download(const GpuMat& d_positions, OutputArray h_positions, OutputArray h_votes = noArray());
-
- void release();
-
-protected:
- virtual void setTemplateImpl(const GpuMat& edges, const GpuMat& dx, const GpuMat& dy, Point templCenter) = 0;
- virtual void detectImpl(const GpuMat& edges, const GpuMat& dx, const GpuMat& dy, GpuMat& positions) = 0;
- virtual void releaseImpl() = 0;
-
-private:
- GpuMat edges_;
- CannyBuf cannyBuf_;
-};
-
-////////////////////////////// Matrix reductions //////////////////////////////
-
-//! computes mean value and standard deviation of all or selected array elements
-//! supports only CV_8UC1 type
-CV_EXPORTS void meanStdDev(const GpuMat& mtx, Scalar& mean, Scalar& stddev);
-//! buffered version
-CV_EXPORTS void meanStdDev(const GpuMat& mtx, Scalar& mean, Scalar& stddev, GpuMat& buf);
-
-//! computes norm of array
-//! supports NORM_INF, NORM_L1, NORM_L2
-//! supports all matrices except 64F
-CV_EXPORTS double norm(const GpuMat& src1, int normType=NORM_L2);
-CV_EXPORTS double norm(const GpuMat& src1, int normType, GpuMat& buf);
-CV_EXPORTS double norm(const GpuMat& src1, int normType, const GpuMat& mask, GpuMat& buf);
-
-//! computes norm of the difference between two arrays
-//! supports NORM_INF, NORM_L1, NORM_L2
-//! supports only CV_8UC1 type
-CV_EXPORTS double norm(const GpuMat& src1, const GpuMat& src2, int normType=NORM_L2);
-
-//! computes sum of array elements
-//! supports only single channel images
-CV_EXPORTS Scalar sum(const GpuMat& src);
-CV_EXPORTS Scalar sum(const GpuMat& src, GpuMat& buf);
-CV_EXPORTS Scalar sum(const GpuMat& src, const GpuMat& mask, GpuMat& buf);
-
-//! computes sum of array elements absolute values
-//! supports only single channel images
-CV_EXPORTS Scalar absSum(const GpuMat& src);
-CV_EXPORTS Scalar absSum(const GpuMat& src, GpuMat& buf);
-CV_EXPORTS Scalar absSum(const GpuMat& src, const GpuMat& mask, GpuMat& buf);
-
-//! computes squared sum of array elements
-//! supports only single channel images
-CV_EXPORTS Scalar sqrSum(const GpuMat& src);
-CV_EXPORTS Scalar sqrSum(const GpuMat& src, GpuMat& buf);
-CV_EXPORTS Scalar sqrSum(const GpuMat& src, const GpuMat& mask, GpuMat& buf);
-
-//! finds global minimum and maximum array elements and returns their values
-CV_EXPORTS void minMax(const GpuMat& src, double* minVal, double* maxVal=0, const GpuMat& mask=GpuMat());
-CV_EXPORTS void minMax(const GpuMat& src, double* minVal, double* maxVal, const GpuMat& mask, GpuMat& buf);
-
-//! finds global minimum and maximum array elements and returns their values with locations
-CV_EXPORTS void minMaxLoc(const GpuMat& src, double* minVal, double* maxVal=0, Point* minLoc=0, Point* maxLoc=0,
- const GpuMat& mask=GpuMat());
-CV_EXPORTS void minMaxLoc(const GpuMat& src, double* minVal, double* maxVal, Point* minLoc, Point* maxLoc,
- const GpuMat& mask, GpuMat& valbuf, GpuMat& locbuf);
-
-//! counts non-zero array elements
-CV_EXPORTS int countNonZero(const GpuMat& src);
-CV_EXPORTS int countNonZero(const GpuMat& src, GpuMat& buf);
-
-//! reduces a matrix to a vector
-CV_EXPORTS void reduce(const GpuMat& mtx, GpuMat& vec, int dim, int reduceOp, int dtype = -1, Stream& stream = Stream::Null());
-
-
-///////////////////////////// Calibration 3D //////////////////////////////////
-
-CV_EXPORTS void transformPoints(const GpuMat& src, const Mat& rvec, const Mat& tvec,
- GpuMat& dst, Stream& stream = Stream::Null());
-
-CV_EXPORTS void projectPoints(const GpuMat& src, const Mat& rvec, const Mat& tvec,
- const Mat& camera_mat, const Mat& dist_coef, GpuMat& dst,
- Stream& stream = Stream::Null());
-
-CV_EXPORTS void solvePnPRansac(const Mat& object, const Mat& image, const Mat& camera_mat,
- const Mat& dist_coef, Mat& rvec, Mat& tvec, bool use_extrinsic_guess=false,
- int num_iters=100, float max_dist=8.0, int min_inlier_count=100,
- std::vector<int>* inliers=NULL);
-
-//////////////////////////////// Image Labeling ////////////////////////////////
-
-//!performs labeling via graph cuts of a 2D regular 4-connected graph.
-CV_EXPORTS void graphcut(GpuMat& terminals, GpuMat& leftTransp, GpuMat& rightTransp, GpuMat& top, GpuMat& bottom, GpuMat& labels,
- GpuMat& buf, Stream& stream = Stream::Null());
-
-//!performs labeling via graph cuts of a 2D regular 8-connected graph.
-CV_EXPORTS void graphcut(GpuMat& terminals, GpuMat& leftTransp, GpuMat& rightTransp, GpuMat& top, GpuMat& topLeft, GpuMat& topRight,
- GpuMat& bottom, GpuMat& bottomLeft, GpuMat& bottomRight,
- GpuMat& labels,
- GpuMat& buf, Stream& stream = Stream::Null());
-
-//! compute mask for Generalized Flood fill componetns labeling.
-CV_EXPORTS void connectivityMask(const GpuMat& image, GpuMat& mask, const cv::Scalar& lo, const cv::Scalar& hi, Stream& stream = Stream::Null());
-
-//! performs connected componnents labeling.
-CV_EXPORTS void labelComponents(const GpuMat& mask, GpuMat& components, int flags = 0, Stream& stream = Stream::Null());
-
-////////////////////////////////// Histograms //////////////////////////////////
-
-//! Compute levels with even distribution. levels will have 1 row and nLevels cols and CV_32SC1 type.
-CV_EXPORTS void evenLevels(GpuMat& levels, int nLevels, int lowerLevel, int upperLevel);
-//! Calculates histogram with evenly distributed bins for signle channel source.
-//! Supports CV_8UC1, CV_16UC1 and CV_16SC1 source types.
-//! Output hist will have one row and histSize cols and CV_32SC1 type.
-CV_EXPORTS void histEven(const GpuMat& src, GpuMat& hist, int histSize, int lowerLevel, int upperLevel, Stream& stream = Stream::Null());
-CV_EXPORTS void histEven(const GpuMat& src, GpuMat& hist, GpuMat& buf, int histSize, int lowerLevel, int upperLevel, Stream& stream = Stream::Null());
-//! Calculates histogram with evenly distributed bins for four-channel source.
-//! All channels of source are processed separately.
-//! Supports CV_8UC4, CV_16UC4 and CV_16SC4 source types.
-//! Output hist[i] will have one row and histSize[i] cols and CV_32SC1 type.
-CV_EXPORTS void histEven(const GpuMat& src, GpuMat hist[4], int histSize[4], int lowerLevel[4], int upperLevel[4], Stream& stream = Stream::Null());
-CV_EXPORTS void histEven(const GpuMat& src, GpuMat hist[4], GpuMat& buf, int histSize[4], int lowerLevel[4], int upperLevel[4], Stream& stream = Stream::Null());
-//! Calculates histogram with bins determined by levels array.
-//! levels must have one row and CV_32SC1 type if source has integer type or CV_32FC1 otherwise.
-//! Supports CV_8UC1, CV_16UC1, CV_16SC1 and CV_32FC1 source types.
-//! Output hist will have one row and (levels.cols-1) cols and CV_32SC1 type.
-CV_EXPORTS void histRange(const GpuMat& src, GpuMat& hist, const GpuMat& levels, Stream& stream = Stream::Null());
-CV_EXPORTS void histRange(const GpuMat& src, GpuMat& hist, const GpuMat& levels, GpuMat& buf, Stream& stream = Stream::Null());
-//! Calculates histogram with bins determined by levels array.
-//! All levels must have one row and CV_32SC1 type if source has integer type or CV_32FC1 otherwise.
-//! All channels of source are processed separately.
-//! Supports CV_8UC4, CV_16UC4, CV_16SC4 and CV_32FC4 source types.
-//! Output hist[i] will have one row and (levels[i].cols-1) cols and CV_32SC1 type.
-CV_EXPORTS void histRange(const GpuMat& src, GpuMat hist[4], const GpuMat levels[4], Stream& stream = Stream::Null());
-CV_EXPORTS void histRange(const GpuMat& src, GpuMat hist[4], const GpuMat levels[4], GpuMat& buf, Stream& stream = Stream::Null());
-
-//! Calculates histogram for 8u one channel image
-//! Output hist will have one row, 256 cols and CV32SC1 type.
-CV_EXPORTS void calcHist(const GpuMat& src, GpuMat& hist, Stream& stream = Stream::Null());
-CV_EXPORTS void calcHist(const GpuMat& src, GpuMat& hist, GpuMat& buf, Stream& stream = Stream::Null());
-
-//! normalizes the grayscale image brightness and contrast by normalizing its histogram
-CV_EXPORTS void equalizeHist(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null());
-CV_EXPORTS void equalizeHist(const GpuMat& src, GpuMat& dst, GpuMat& hist, Stream& stream = Stream::Null());
-CV_EXPORTS void equalizeHist(const GpuMat& src, GpuMat& dst, GpuMat& hist, GpuMat& buf, Stream& stream = Stream::Null());
-
-class CV_EXPORTS CLAHE : public cv::CLAHE
-{
-public:
- using cv::CLAHE::apply;
- virtual void apply(InputArray src, OutputArray dst, Stream& stream) = 0;
-};
-CV_EXPORTS Ptr<cv::gpu::CLAHE> createCLAHE(double clipLimit = 40.0, Size tileGridSize = Size(8, 8));
-
-//////////////////////////////// StereoBM_GPU ////////////////////////////////
-
-class CV_EXPORTS StereoBM_GPU
-{
-public:
- enum { BASIC_PRESET = 0, PREFILTER_XSOBEL = 1 };
-
- enum { DEFAULT_NDISP = 64, DEFAULT_WINSZ = 19 };
-
- //! the default constructor
- StereoBM_GPU();
- //! the full constructor taking the camera-specific preset, number of disparities and the SAD window size. ndisparities must be multiple of 8.
- StereoBM_GPU(int preset, int ndisparities = DEFAULT_NDISP, int winSize = DEFAULT_WINSZ);
-
- //! the stereo correspondence operator. Finds the disparity for the specified rectified stereo pair
- //! Output disparity has CV_8U type.
- void operator()(const GpuMat& left, const GpuMat& right, GpuMat& disparity, Stream& stream = Stream::Null());
-
- //! Some heuristics that tries to estmate
- // if current GPU will be faster than CPU in this algorithm.
- // It queries current active device.
- static bool checkIfGpuCallReasonable();
-
- int preset;
- int ndisp;
- int winSize;
-
- // If avergeTexThreshold == 0 => post procesing is disabled
- // If avergeTexThreshold != 0 then disparity is set 0 in each point (x,y) where for left image
- // SumOfHorizontalGradiensInWindow(x, y, winSize) < (winSize * winSize) * avergeTexThreshold
- // i.e. input left image is low textured.
- float avergeTexThreshold;
-
-private:
- GpuMat minSSD, leBuf, riBuf;
-};
-
-////////////////////////// StereoBeliefPropagation ///////////////////////////
-// "Efficient Belief Propagation for Early Vision"
-// P.Felzenszwalb
-
-class CV_EXPORTS StereoBeliefPropagation
-{
-public:
- enum { DEFAULT_NDISP = 64 };
- enum { DEFAULT_ITERS = 5 };
- enum { DEFAULT_LEVELS = 5 };
-
- static void estimateRecommendedParams(int width, int height, int& ndisp, int& iters, int& levels);
-
- //! the default constructor
- explicit StereoBeliefPropagation(int ndisp = DEFAULT_NDISP,
- int iters = DEFAULT_ITERS,
- int levels = DEFAULT_LEVELS,
- int msg_type = CV_32F);
-
- //! the full constructor taking the number of disparities, number of BP iterations on each level,
- //! number of levels, truncation of data cost, data weight,
- //! truncation of discontinuity cost and discontinuity single jump
- //! DataTerm = data_weight * min(fabs(I2-I1), max_data_term)
- //! DiscTerm = min(disc_single_jump * fabs(f1-f2), max_disc_term)
- //! please see paper for more details
- StereoBeliefPropagation(int ndisp, int iters, int levels,
- float max_data_term, float data_weight,
- float max_disc_term, float disc_single_jump,
- int msg_type = CV_32F);
-
- //! the stereo correspondence operator. Finds the disparity for the specified rectified stereo pair,
- //! if disparity is empty output type will be CV_16S else output type will be disparity.type().
- void operator()(const GpuMat& left, const GpuMat& right, GpuMat& disparity, Stream& stream = Stream::Null());
-
-
- //! version for user specified data term
- void operator()(const GpuMat& data, GpuMat& disparity, Stream& stream = Stream::Null());
-
- int ndisp;
-
- int iters;
- int levels;
-
- float max_data_term;
- float data_weight;
- float max_disc_term;
- float disc_single_jump;
-
- int msg_type;
-private:
- GpuMat u, d, l, r, u2, d2, l2, r2;
- std::vector<GpuMat> datas;
- GpuMat out;
-};
-
-/////////////////////////// StereoConstantSpaceBP ///////////////////////////
-// "A Constant-Space Belief Propagation Algorithm for Stereo Matching"
-// Qingxiong Yang, Liang Wang, Narendra Ahuja
-// http://vision.ai.uiuc.edu/~qyang6/
-
-class CV_EXPORTS StereoConstantSpaceBP
-{
-public:
- enum { DEFAULT_NDISP = 128 };
- enum { DEFAULT_ITERS = 8 };
- enum { DEFAULT_LEVELS = 4 };
- enum { DEFAULT_NR_PLANE = 4 };
-
- static void estimateRecommendedParams(int width, int height, int& ndisp, int& iters, int& levels, int& nr_plane);
-
- //! the default constructor
- explicit StereoConstantSpaceBP(int ndisp = DEFAULT_NDISP,
- int iters = DEFAULT_ITERS,
- int levels = DEFAULT_LEVELS,
- int nr_plane = DEFAULT_NR_PLANE,
- int msg_type = CV_32F);
-
- //! the full constructor taking the number of disparities, number of BP iterations on each level,
- //! number of levels, number of active disparity on the first level, truncation of data cost, data weight,
- //! truncation of discontinuity cost, discontinuity single jump and minimum disparity threshold
- StereoConstantSpaceBP(int ndisp, int iters, int levels, int nr_plane,
- float max_data_term, float data_weight, float max_disc_term, float disc_single_jump,
- int min_disp_th = 0,
- int msg_type = CV_32F);
-
- //! the stereo correspondence operator. Finds the disparity for the specified rectified stereo pair,
- //! if disparity is empty output type will be CV_16S else output type will be disparity.type().
- void operator()(const GpuMat& left, const GpuMat& right, GpuMat& disparity, Stream& stream = Stream::Null());
-
- int ndisp;
-
- int iters;
- int levels;
-
- int nr_plane;
-
- float max_data_term;
- float data_weight;
- float max_disc_term;
- float disc_single_jump;
-
- int min_disp_th;
-
- int msg_type;
-
- bool use_local_init_data_cost;
-private:
- GpuMat messages_buffers;
-
- GpuMat temp;
- GpuMat out;
-};
-
-/////////////////////////// DisparityBilateralFilter ///////////////////////////
-// Disparity map refinement using joint bilateral filtering given a single color image.
-// Qingxiong Yang, Liang Wang, Narendra Ahuja
-// http://vision.ai.uiuc.edu/~qyang6/
-
-class CV_EXPORTS DisparityBilateralFilter
-{
-public:
- enum { DEFAULT_NDISP = 64 };
- enum { DEFAULT_RADIUS = 3 };
- enum { DEFAULT_ITERS = 1 };
-
- //! the default constructor
- explicit DisparityBilateralFilter(int ndisp = DEFAULT_NDISP, int radius = DEFAULT_RADIUS, int iters = DEFAULT_ITERS);
-
- //! the full constructor taking the number of disparities, filter radius,
- //! number of iterations, truncation of data continuity, truncation of disparity continuity
- //! and filter range sigma
- DisparityBilateralFilter(int ndisp, int radius, int iters, float edge_threshold, float max_disc_threshold, float sigma_range);
-
- //! the disparity map refinement operator. Refine disparity map using joint bilateral filtering given a single color image.
- //! disparity must have CV_8U or CV_16S type, image must have CV_8UC1 or CV_8UC3 type.
- void operator()(const GpuMat& disparity, const GpuMat& image, GpuMat& dst, Stream& stream = Stream::Null());
-
-private:
- int ndisp;
- int radius;
- int iters;
-
- float edge_threshold;
- float max_disc_threshold;
- float sigma_range;
-
- GpuMat table_color;
- GpuMat table_space;
-};
-
-
-//////////////// HOG (Histogram-of-Oriented-Gradients) Descriptor and Object Detector //////////////
-struct CV_EXPORTS HOGConfidence
-{
- double scale;
- std::vector<Point> locations;
- std::vector<double> confidences;
- std::vector<double> part_scores[4];
-};
-
-struct CV_EXPORTS HOGDescriptor
-{
- enum { DEFAULT_WIN_SIGMA = -1 };
- enum { DEFAULT_NLEVELS = 64 };
- enum { DESCR_FORMAT_ROW_BY_ROW, DESCR_FORMAT_COL_BY_COL };
-
- HOGDescriptor(Size win_size=Size(64, 128), Size block_size=Size(16, 16),
- Size block_stride=Size(8, 8), Size cell_size=Size(8, 8),
- int nbins=9, double win_sigma=DEFAULT_WIN_SIGMA,
- double threshold_L2hys=0.2, bool gamma_correction=true,
- int nlevels=DEFAULT_NLEVELS);
-
- size_t getDescriptorSize() const;
- size_t getBlockHistogramSize() const;
-
- void setSVMDetector(const std::vector<float>& detector);
-
- static std::vector<float> getDefaultPeopleDetector();
- static std::vector<float> getPeopleDetector48x96();
- static std::vector<float> getPeopleDetector64x128();
-
- void detect(const GpuMat& img, std::vector<Point>& found_locations,
- double hit_threshold=0, Size win_stride=Size(),
- Size padding=Size());
-
- void detectMultiScale(const GpuMat& img, std::vector<Rect>& found_locations,
- double hit_threshold=0, Size win_stride=Size(),
- Size padding=Size(), double scale0=1.05,
- int group_threshold=2);
-
- void computeConfidence(const GpuMat& img, std::vector<Point>& hits, double hit_threshold,
- Size win_stride, Size padding, std::vector<Point>& locations, std::vector<double>& confidences);
-
- void computeConfidenceMultiScale(const GpuMat& img, std::vector<Rect>& found_locations,
- double hit_threshold, Size win_stride, Size padding,
- std::vector<HOGConfidence> &conf_out, int group_threshold);
-
- void getDescriptors(const GpuMat& img, Size win_stride,
- GpuMat& descriptors,
- int descr_format=DESCR_FORMAT_COL_BY_COL);
-
- Size win_size;
- Size block_size;
- Size block_stride;
- Size cell_size;
- int nbins;
- double win_sigma;
- double threshold_L2hys;
- bool gamma_correction;
- int nlevels;
-
-protected:
- void computeBlockHistograms(const GpuMat& img);
- void computeGradient(const GpuMat& img, GpuMat& grad, GpuMat& qangle);
-
- double getWinSigma() const;
- bool checkDetectorSize() const;
-
- static int numPartsWithin(int size, int part_size, int stride);
- static Size numPartsWithin(Size size, Size part_size, Size stride);
-
- // Coefficients of the separating plane
- float free_coef;
- GpuMat detector;
-
- // Results of the last classification step
- GpuMat labels, labels_buf;
- Mat labels_host;
-
- // Results of the last histogram evaluation step
- GpuMat block_hists, block_hists_buf;
-
- // Gradients conputation results
- GpuMat grad, qangle, grad_buf, qangle_buf;
-
- // returns subbuffer with required size, reallocates buffer if nessesary.
- static GpuMat getBuffer(const Size& sz, int type, GpuMat& buf);
- static GpuMat getBuffer(int rows, int cols, int type, GpuMat& buf);
-
- std::vector<GpuMat> image_scales;
-};
-
-
-////////////////////////////////// BruteForceMatcher //////////////////////////////////
-
-class CV_EXPORTS BFMatcher_GPU
-{
-public:
- explicit BFMatcher_GPU(int norm = cv::NORM_L2);
-
- // Add descriptors to train descriptor collection
- void add(const std::vector<GpuMat>& descCollection);
-
- // Get train descriptors collection
- const std::vector<GpuMat>& getTrainDescriptors() const;
-
- // Clear train descriptors collection
- void clear();
-
- // Return true if there are not train descriptors in collection
- bool empty() const;
-
- // Return true if the matcher supports mask in match methods
- bool isMaskSupported() const;
-
- // Find one best match for each query descriptor
- void matchSingle(const GpuMat& query, const GpuMat& train,
- GpuMat& trainIdx, GpuMat& distance,
- const GpuMat& mask = GpuMat(), Stream& stream = Stream::Null());
-
- // Download trainIdx and distance and convert it to CPU vector with DMatch
- static void matchDownload(const GpuMat& trainIdx, const GpuMat& distance, std::vector<DMatch>& matches);
- // Convert trainIdx and distance to vector with DMatch
- static void matchConvert(const Mat& trainIdx, const Mat& distance, std::vector<DMatch>& matches);
-
- // Find one best match for each query descriptor
- void match(const GpuMat& query, const GpuMat& train, std::vector<DMatch>& matches, const GpuMat& mask = GpuMat());
-
- // Make gpu collection of trains and masks in suitable format for matchCollection function
- void makeGpuCollection(GpuMat& trainCollection, GpuMat& maskCollection, const std::vector<GpuMat>& masks = std::vector<GpuMat>());
-
- // Find one best match from train collection for each query descriptor
- void matchCollection(const GpuMat& query, const GpuMat& trainCollection,
- GpuMat& trainIdx, GpuMat& imgIdx, GpuMat& distance,
- const GpuMat& masks = GpuMat(), Stream& stream = Stream::Null());
-
- // Download trainIdx, imgIdx and distance and convert it to vector with DMatch
- static void matchDownload(const GpuMat& trainIdx, const GpuMat& imgIdx, const GpuMat& distance, std::vector<DMatch>& matches);
- // Convert trainIdx, imgIdx and distance to vector with DMatch
- static void matchConvert(const Mat& trainIdx, const Mat& imgIdx, const Mat& distance, std::vector<DMatch>& matches);
-
- // Find one best match from train collection for each query descriptor.
- void match(const GpuMat& query, std::vector<DMatch>& matches, const std::vector<GpuMat>& masks = std::vector<GpuMat>());
-
- // Find k best matches for each query descriptor (in increasing order of distances)
- void knnMatchSingle(const GpuMat& query, const GpuMat& train,
- GpuMat& trainIdx, GpuMat& distance, GpuMat& allDist, int k,
- const GpuMat& mask = GpuMat(), Stream& stream = Stream::Null());
-
- // Download trainIdx and distance and convert it to vector with DMatch
- // compactResult is used when mask is not empty. If compactResult is false matches
- // vector will have the same size as queryDescriptors rows. If compactResult is true
- // matches vector will not contain matches for fully masked out query descriptors.
- static void knnMatchDownload(const GpuMat& trainIdx, const GpuMat& distance,
- std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
- // Convert trainIdx and distance to vector with DMatch
- static void knnMatchConvert(const Mat& trainIdx, const Mat& distance,
- std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
-
- // Find k best matches for each query descriptor (in increasing order of distances).
- // compactResult is used when mask is not empty. If compactResult is false matches
- // vector will have the same size as queryDescriptors rows. If compactResult is true
- // matches vector will not contain matches for fully masked out query descriptors.
- void knnMatch(const GpuMat& query, const GpuMat& train,
- std::vector< std::vector<DMatch> >& matches, int k, const GpuMat& mask = GpuMat(),
- bool compactResult = false);
-
- // Find k best matches from train collection for each query descriptor (in increasing order of distances)
- void knnMatch2Collection(const GpuMat& query, const GpuMat& trainCollection,
- GpuMat& trainIdx, GpuMat& imgIdx, GpuMat& distance,
- const GpuMat& maskCollection = GpuMat(), Stream& stream = Stream::Null());
-
- // Download trainIdx and distance and convert it to vector with DMatch
- // compactResult is used when mask is not empty. If compactResult is false matches
- // vector will have the same size as queryDescriptors rows. If compactResult is true
- // matches vector will not contain matches for fully masked out query descriptors.
- static void knnMatch2Download(const GpuMat& trainIdx, const GpuMat& imgIdx, const GpuMat& distance,
- std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
- // Convert trainIdx and distance to vector with DMatch
- static void knnMatch2Convert(const Mat& trainIdx, const Mat& imgIdx, const Mat& distance,
- std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
-
- // Find k best matches for each query descriptor (in increasing order of distances).
- // compactResult is used when mask is not empty. If compactResult is false matches
- // vector will have the same size as queryDescriptors rows. If compactResult is true
- // matches vector will not contain matches for fully masked out query descriptors.
- void knnMatch(const GpuMat& query, std::vector< std::vector<DMatch> >& matches, int k,
- const std::vector<GpuMat>& masks = std::vector<GpuMat>(), bool compactResult = false);
-
- // Find best matches for each query descriptor which have distance less than maxDistance.
- // nMatches.at<int>(0, queryIdx) will contain matches count for queryIdx.
- // carefully nMatches can be greater than trainIdx.cols - it means that matcher didn't find all matches,
- // because it didn't have enough memory.
- // If trainIdx is empty, then trainIdx and distance will be created with size nQuery x max((nTrain / 100), 10),
- // otherwize user can pass own allocated trainIdx and distance with size nQuery x nMaxMatches
- // Matches doesn't sorted.
- void radiusMatchSingle(const GpuMat& query, const GpuMat& train,
- GpuMat& trainIdx, GpuMat& distance, GpuMat& nMatches, float maxDistance,
- const GpuMat& mask = GpuMat(), Stream& stream = Stream::Null());
-
- // Download trainIdx, nMatches and distance and convert it to vector with DMatch.
- // matches will be sorted in increasing order of distances.
- // compactResult is used when mask is not empty. If compactResult is false matches
- // vector will have the same size as queryDescriptors rows. If compactResult is true
- // matches vector will not contain matches for fully masked out query descriptors.
- static void radiusMatchDownload(const GpuMat& trainIdx, const GpuMat& distance, const GpuMat& nMatches,
- std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
- // Convert trainIdx, nMatches and distance to vector with DMatch.
- static void radiusMatchConvert(const Mat& trainIdx, const Mat& distance, const Mat& nMatches,
- std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
-
- // Find best matches for each query descriptor which have distance less than maxDistance
- // in increasing order of distances).
- void radiusMatch(const GpuMat& query, const GpuMat& train,
- std::vector< std::vector<DMatch> >& matches, float maxDistance,
- const GpuMat& mask = GpuMat(), bool compactResult = false);
-
- // Find best matches for each query descriptor which have distance less than maxDistance.
- // If trainIdx is empty, then trainIdx and distance will be created with size nQuery x max((nQuery / 100), 10),
- // otherwize user can pass own allocated trainIdx and distance with size nQuery x nMaxMatches
- // Matches doesn't sorted.
- void radiusMatchCollection(const GpuMat& query, GpuMat& trainIdx, GpuMat& imgIdx, GpuMat& distance, GpuMat& nMatches, float maxDistance,
- const std::vector<GpuMat>& masks = std::vector<GpuMat>(), Stream& stream = Stream::Null());
-
- // Download trainIdx, imgIdx, nMatches and distance and convert it to vector with DMatch.
- // matches will be sorted in increasing order of distances.
- // compactResult is used when mask is not empty. If compactResult is false matches
- // vector will have the same size as queryDescriptors rows. If compactResult is true
- // matches vector will not contain matches for fully masked out query descriptors.
- static void radiusMatchDownload(const GpuMat& trainIdx, const GpuMat& imgIdx, const GpuMat& distance, const GpuMat& nMatches,
- std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
- // Convert trainIdx, nMatches and distance to vector with DMatch.
- static void radiusMatchConvert(const Mat& trainIdx, const Mat& imgIdx, const Mat& distance, const Mat& nMatches,
- std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
-
- // Find best matches from train collection for each query descriptor which have distance less than
- // maxDistance (in increasing order of distances).
- void radiusMatch(const GpuMat& query, std::vector< std::vector<DMatch> >& matches, float maxDistance,
- const std::vector<GpuMat>& masks = std::vector<GpuMat>(), bool compactResult = false);
-
- int norm;
-
-private:
- std::vector<GpuMat> trainDescCollection;
-};
-
-template <class Distance>
-class CV_EXPORTS BruteForceMatcher_GPU;
-
-template <typename T>
-class CV_EXPORTS BruteForceMatcher_GPU< L1<T> > : public BFMatcher_GPU
-{
-public:
- explicit BruteForceMatcher_GPU() : BFMatcher_GPU(NORM_L1) {}
- explicit BruteForceMatcher_GPU(L1<T> /*d*/) : BFMatcher_GPU(NORM_L1) {}
-};
-template <typename T>
-class CV_EXPORTS BruteForceMatcher_GPU< L2<T> > : public BFMatcher_GPU
-{
-public:
- explicit BruteForceMatcher_GPU() : BFMatcher_GPU(NORM_L2) {}
- explicit BruteForceMatcher_GPU(L2<T> /*d*/) : BFMatcher_GPU(NORM_L2) {}
-};
-template <> class CV_EXPORTS BruteForceMatcher_GPU< Hamming > : public BFMatcher_GPU
-{
-public:
- explicit BruteForceMatcher_GPU() : BFMatcher_GPU(NORM_HAMMING) {}
- explicit BruteForceMatcher_GPU(Hamming /*d*/) : BFMatcher_GPU(NORM_HAMMING) {}
-};
-
-////////////////////////////////// CascadeClassifier_GPU //////////////////////////////////////////
-// The cascade classifier class for object detection: supports old haar and new lbp xlm formats and nvbin for haar cascades olny.
-class CV_EXPORTS CascadeClassifier_GPU
-{
-public:
- CascadeClassifier_GPU();
- CascadeClassifier_GPU(const String& filename);
- ~CascadeClassifier_GPU();
-
- bool empty() const;
- bool load(const String& filename);
- void release();
-
- /* returns number of detected objects */
- int detectMultiScale(const GpuMat& image, GpuMat& objectsBuf, double scaleFactor = 1.2, int minNeighbors = 4, Size minSize = Size());
- int detectMultiScale(const GpuMat& image, GpuMat& objectsBuf, Size maxObjectSize, Size minSize = Size(), double scaleFactor = 1.1, int minNeighbors = 4);
-
- bool findLargestObject;
- bool visualizeInPlace;
-
- Size getClassifierSize() const;
-
-private:
- struct CascadeClassifierImpl;
- CascadeClassifierImpl* impl;
- struct HaarCascade;
- struct LbpCascade;
- friend class CascadeClassifier_GPU_LBP;
-};
-
-////////////////////////////////// FAST //////////////////////////////////////////
-
-class CV_EXPORTS FAST_GPU
-{
-public:
- enum
- {
- LOCATION_ROW = 0,
- RESPONSE_ROW,
- ROWS_COUNT
- };
-
- // all features have same size
- static const int FEATURE_SIZE = 7;
-
- explicit FAST_GPU(int threshold, bool nonmaxSupression = true, double keypointsRatio = 0.05);
-
- //! finds the keypoints using FAST detector
- //! supports only CV_8UC1 images
- void operator ()(const GpuMat& image, const GpuMat& mask, GpuMat& keypoints);
- void operator ()(const GpuMat& image, const GpuMat& mask, std::vector<KeyPoint>& keypoints);
-
- //! download keypoints from device to host memory
- static void downloadKeypoints(const GpuMat& d_keypoints, std::vector<KeyPoint>& keypoints);
-
- //! convert keypoints to KeyPoint vector
- static void convertKeypoints(const Mat& h_keypoints, std::vector<KeyPoint>& keypoints);
-
- //! release temporary buffer's memory
- void release();
-
- bool nonmaxSupression;
-
- int threshold;
-
- //! max keypoints = keypointsRatio * img.size().area()
- double keypointsRatio;
-
- //! find keypoints and compute it's response if nonmaxSupression is true
- //! return count of detected keypoints
- int calcKeyPointsLocation(const GpuMat& image, const GpuMat& mask);
-
- //! get final array of keypoints
- //! performs nonmax supression if needed
- //! return final count of keypoints
- int getKeyPoints(GpuMat& keypoints);
-
-private:
- GpuMat kpLoc_;
- int count_;
-
- GpuMat score_;
-
- GpuMat d_keypoints_;
-};
-
-////////////////////////////////// ORB //////////////////////////////////////////
-
-class CV_EXPORTS ORB_GPU
-{
-public:
- enum
- {
- X_ROW = 0,
- Y_ROW,
- RESPONSE_ROW,
- ANGLE_ROW,
- OCTAVE_ROW,
- SIZE_ROW,
- ROWS_COUNT
- };
-
- enum
- {
- DEFAULT_FAST_THRESHOLD = 20
- };
-
- //! Constructor
- explicit ORB_GPU(int nFeatures = 500, float scaleFactor = 1.2f, int nLevels = 8, int edgeThreshold = 31,
- int firstLevel = 0, int WTA_K = 2, int scoreType = 0, int patchSize = 31);
-
- //! Compute the ORB features on an image
- //! image - the image to compute the features (supports only CV_8UC1 images)
- //! mask - the mask to apply
- //! keypoints - the resulting keypoints
- void operator()(const GpuMat& image, const GpuMat& mask, std::vector<KeyPoint>& keypoints);
- void operator()(const GpuMat& image, const GpuMat& mask, GpuMat& keypoints);
-
- //! Compute the ORB features and descriptors on an image
- //! image - the image to compute the features (supports only CV_8UC1 images)
- //! mask - the mask to apply
- //! keypoints - the resulting keypoints
- //! descriptors - descriptors array
- void operator()(const GpuMat& image, const GpuMat& mask, std::vector<KeyPoint>& keypoints, GpuMat& descriptors);
- void operator()(const GpuMat& image, const GpuMat& mask, GpuMat& keypoints, GpuMat& descriptors);
-
- //! download keypoints from device to host memory
- static void downloadKeyPoints(const GpuMat& d_keypoints, std::vector<KeyPoint>& keypoints);
- //! convert keypoints to KeyPoint vector
- static void convertKeyPoints(const Mat& d_keypoints, std::vector<KeyPoint>& keypoints);
-
- //! returns the descriptor size in bytes
- inline int descriptorSize() const { return kBytes; }
-
- inline void setFastParams(int threshold, bool nonmaxSupression = true)
- {
- fastDetector_.threshold = threshold;
- fastDetector_.nonmaxSupression = nonmaxSupression;
- }
-
- //! release temporary buffer's memory
- void release();
-
- //! if true, image will be blurred before descriptors calculation
- bool blurForDescriptor;
-
-private:
- enum { kBytes = 32 };
-
- void buildScalePyramids(const GpuMat& image, const GpuMat& mask);
-
- void computeKeyPointsPyramid();
-
- void computeDescriptors(GpuMat& descriptors);
-
- void mergeKeyPoints(GpuMat& keypoints);
-
- int nFeatures_;
- float scaleFactor_;
- int nLevels_;
- int edgeThreshold_;
- int firstLevel_;
- int WTA_K_;
- int scoreType_;
- int patchSize_;
-
- // The number of desired features per scale
- std::vector<size_t> n_features_per_level_;
-
- // Points to compute BRIEF descriptors from
- GpuMat pattern_;
-
- std::vector<GpuMat> imagePyr_;
- std::vector<GpuMat> maskPyr_;
-
- GpuMat buf_;
-
- std::vector<GpuMat> keyPointsPyr_;
- std::vector<int> keyPointsCount_;
-
- FAST_GPU fastDetector_;
-
- Ptr<FilterEngine_GPU> blurFilter;
-
- GpuMat d_keypoints_;
-};
-
-////////////////////////////////// Optical Flow //////////////////////////////////////////
-
-class CV_EXPORTS BroxOpticalFlow
-{
-public:
- BroxOpticalFlow(float alpha_, float gamma_, float scale_factor_, int inner_iterations_, int outer_iterations_, int solver_iterations_) :
- alpha(alpha_), gamma(gamma_), scale_factor(scale_factor_),
- inner_iterations(inner_iterations_), outer_iterations(outer_iterations_), solver_iterations(solver_iterations_)
- {
- }
-
- //! Compute optical flow
- //! frame0 - source frame (supports only CV_32FC1 type)
- //! frame1 - frame to track (with the same size and type as frame0)
- //! u - flow horizontal component (along x axis)
- //! v - flow vertical component (along y axis)
- void operator ()(const GpuMat& frame0, const GpuMat& frame1, GpuMat& u, GpuMat& v, Stream& stream = Stream::Null());
-
- //! flow smoothness
- float alpha;
-
- //! gradient constancy importance
- float gamma;
-
- //! pyramid scale factor
- float scale_factor;
-
- //! number of lagged non-linearity iterations (inner loop)
- int inner_iterations;
-
- //! number of warping iterations (number of pyramid levels)
- int outer_iterations;
-
- //! number of linear system solver iterations
- int solver_iterations;
-
- GpuMat buf;
-};
-
-class CV_EXPORTS GoodFeaturesToTrackDetector_GPU
-{
-public:
- explicit GoodFeaturesToTrackDetector_GPU(int maxCorners = 1000, double qualityLevel = 0.01, double minDistance = 0.0,
- int blockSize = 3, bool useHarrisDetector = false, double harrisK = 0.04);
-
- //! return 1 rows matrix with CV_32FC2 type
- void operator ()(const GpuMat& image, GpuMat& corners, const GpuMat& mask = GpuMat());
-
- int maxCorners;
- double qualityLevel;
- double minDistance;
-
- int blockSize;
- bool useHarrisDetector;
- double harrisK;
-
- void releaseMemory()
- {
- Dx_.release();
- Dy_.release();
- buf_.release();
- eig_.release();
- minMaxbuf_.release();
- tmpCorners_.release();
- }
-
-private:
- GpuMat Dx_;
- GpuMat Dy_;
- GpuMat buf_;
- GpuMat eig_;
- GpuMat minMaxbuf_;
- GpuMat tmpCorners_;
-};
-
-inline GoodFeaturesToTrackDetector_GPU::GoodFeaturesToTrackDetector_GPU(int maxCorners_, double qualityLevel_, double minDistance_,
- int blockSize_, bool useHarrisDetector_, double harrisK_)
-{
- maxCorners = maxCorners_;
- qualityLevel = qualityLevel_;
- minDistance = minDistance_;
- blockSize = blockSize_;
- useHarrisDetector = useHarrisDetector_;
- harrisK = harrisK_;
-}
-
-
-class CV_EXPORTS PyrLKOpticalFlow
-{
-public:
- PyrLKOpticalFlow();
-
- void sparse(const GpuMat& prevImg, const GpuMat& nextImg, const GpuMat& prevPts, GpuMat& nextPts,
- GpuMat& status, GpuMat* err = 0);
-
- void dense(const GpuMat& prevImg, const GpuMat& nextImg, GpuMat& u, GpuMat& v, GpuMat* err = 0);
-
- void releaseMemory();
-
- Size winSize;
- int maxLevel;
- int iters;
- bool useInitialFlow;
-
-private:
- std::vector<GpuMat> prevPyr_;
- std::vector<GpuMat> nextPyr_;
-
- GpuMat buf_;
-
- GpuMat uPyr_[2];
- GpuMat vPyr_[2];
-};
-
-
-class CV_EXPORTS FarnebackOpticalFlow
-{
-public:
- FarnebackOpticalFlow()
- {
- numLevels = 5;
- pyrScale = 0.5;
- fastPyramids = false;
- winSize = 13;
- numIters = 10;
- polyN = 5;
- polySigma = 1.1;
- flags = 0;
- }
-
- int numLevels;
- double pyrScale;
- bool fastPyramids;
- int winSize;
- int numIters;
- int polyN;
- double polySigma;
- int flags;
-
- void operator ()(const GpuMat &frame0, const GpuMat &frame1, GpuMat &flowx, GpuMat &flowy, Stream &s = Stream::Null());
-
- void releaseMemory()
- {
- frames_[0].release();
- frames_[1].release();
- pyrLevel_[0].release();
- pyrLevel_[1].release();
- M_.release();
- bufM_.release();
- R_[0].release();
- R_[1].release();
- blurredFrame_[0].release();
- blurredFrame_[1].release();
- pyramid0_.clear();
- pyramid1_.clear();
- }
-
-private:
- void prepareGaussian(
- int n, double sigma, float *g, float *xg, float *xxg,
- double &ig11, double &ig03, double &ig33, double &ig55);
-
- void setPolynomialExpansionConsts(int n, double sigma);
-
- void updateFlow_boxFilter(
- const GpuMat& R0, const GpuMat& R1, GpuMat& flowx, GpuMat &flowy,
- GpuMat& M, GpuMat &bufM, int blockSize, bool updateMatrices, Stream streams[]);
-
- void updateFlow_gaussianBlur(
- const GpuMat& R0, const GpuMat& R1, GpuMat& flowx, GpuMat& flowy,
- GpuMat& M, GpuMat &bufM, int blockSize, bool updateMatrices, Stream streams[]);
-
- GpuMat frames_[2];
- GpuMat pyrLevel_[2], M_, bufM_, R_[2], blurredFrame_[2];
- std::vector<GpuMat> pyramid0_, pyramid1_;
-};
-
-
-// Implementation of the Zach, Pock and Bischof Dual TV-L1 Optical Flow method
-//
-// see reference:
-// [1] C. Zach, T. Pock and H. Bischof, "A Duality Based Approach for Realtime TV-L1 Optical Flow".
-// [2] Javier Sanchez, Enric Meinhardt-Llopis and Gabriele Facciolo. "TV-L1 Optical Flow Estimation".
-class CV_EXPORTS OpticalFlowDual_TVL1_GPU
-{
-public:
- OpticalFlowDual_TVL1_GPU();
-
- void operator ()(const GpuMat& I0, const GpuMat& I1, GpuMat& flowx, GpuMat& flowy);
-
- void collectGarbage();
-
- /**
- * Time step of the numerical scheme.
- */
- double tau;
-
- /**
- * Weight parameter for the data term, attachment parameter.
- * This is the most relevant parameter, which determines the smoothness of the output.
- * The smaller this parameter is, the smoother the solutions we obtain.
- * It depends on the range of motions of the images, so its value should be adapted to each image sequence.
- */
- double lambda;
-
- /**
- * Weight parameter for (u - v)^2, tightness parameter.
- * It serves as a link between the attachment and the regularization terms.
- * In theory, it should have a small value in order to maintain both parts in correspondence.
- * The method is stable for a large range of values of this parameter.
- */
- double theta;
+#if !defined(__OPENCV_BUILD) && !defined(OPENCV_GPU_SKIP_INCLUDE)
+ #include "opencv2/opencv_modules.hpp"
- /**
- * Number of scales used to create the pyramid of images.
- */
- int nscales;
+ #ifdef HAVE_OPENCV_GPUARITHM
+ #include "opencv2/gpuarithm.hpp"
+ #endif
- /**
- * Number of warpings per scale.
- * Represents the number of times that I1(x+u0) and grad( I1(x+u0) ) are computed per scale.
- * This is a parameter that assures the stability of the method.
- * It also affects the running time, so it is a compromise between speed and accuracy.
- */
- int warps;
+ #ifdef HAVE_OPENCV_GPUWARPING
+ #include "opencv2/gpuwarping.hpp"
+ #endif
- /**
- * Stopping criterion threshold used in the numerical scheme, which is a trade-off between precision and running time.
- * A small value will yield more accurate solutions at the expense of a slower convergence.
- */
- double epsilon;
+ #ifdef HAVE_OPENCV_GPUFILTERS
+ #include "opencv2/gpufilters.hpp"
+ #endif
- /**
- * Stopping criterion iterations number used in the numerical scheme.
- */
- int iterations;
+ #ifdef HAVE_OPENCV_GPUIMGPROC
+ #include "opencv2/gpuimgproc.hpp"
+ #endif
- double scaleStep;
+ #ifdef HAVE_OPENCV_GPUFEATURES2D
+ #include "opencv2/gpufeatures2d.hpp"
+ #endif
- bool useInitialFlow;
+ #ifdef HAVE_OPENCV_GPUOPTFLOW
+ #include "opencv2/gpuoptflow.hpp"
+ #endif
-private:
- void procOneScale(const GpuMat& I0, const GpuMat& I1, GpuMat& u1, GpuMat& u2);
-
- std::vector<GpuMat> I0s;
- std::vector<GpuMat> I1s;
- std::vector<GpuMat> u1s;
- std::vector<GpuMat> u2s;
-
- GpuMat I1x_buf;
- GpuMat I1y_buf;
-
- GpuMat I1w_buf;
- GpuMat I1wx_buf;
- GpuMat I1wy_buf;
+ #ifdef HAVE_OPENCV_GPUBGSEGM
+ #include "opencv2/gpubgsegm.hpp"
+ #endif
- GpuMat grad_buf;
- GpuMat rho_c_buf;
+ #ifdef HAVE_OPENCV_GPUSTEREO
+ #include "opencv2/gpustereo.hpp"
+ #endif
- GpuMat p11_buf;
- GpuMat p12_buf;
- GpuMat p21_buf;
- GpuMat p22_buf;
-
- GpuMat diff_buf;
- GpuMat norm_buf;
-};
+ #ifdef HAVE_OPENCV_GPUCODEC
+ #include "opencv2/gpucodec.hpp"
+ #endif
+#endif
+namespace cv { namespace gpu {
-//! Calculates optical flow for 2 images using block matching algorithm */
-CV_EXPORTS void calcOpticalFlowBM(const GpuMat& prev, const GpuMat& curr,
- Size block_size, Size shift_size, Size max_range, bool use_previous,
- GpuMat& velx, GpuMat& vely, GpuMat& buf,
- Stream& stream = Stream::Null());
+//////////////// HOG (Histogram-of-Oriented-Gradients) Descriptor and Object Detector //////////////
-class CV_EXPORTS FastOpticalFlowBM
+struct CV_EXPORTS HOGConfidence
{
-public:
- void operator ()(const GpuMat& I0, const GpuMat& I1, GpuMat& flowx, GpuMat& flowy, int search_window = 21, int block_window = 7, Stream& s = Stream::Null());
-
-private:
- GpuMat buffer;
- GpuMat extended_I0;
- GpuMat extended_I1;
+ double scale;
+ std::vector<Point> locations;
+ std::vector<double> confidences;
+ std::vector<double> part_scores[4];
};
-
-//! Interpolate frames (images) using provided optical flow (displacement field).
-//! frame0 - frame 0 (32-bit floating point images, single channel)
-//! frame1 - frame 1 (the same type and size)
-//! fu - forward horizontal displacement
-//! fv - forward vertical displacement
-//! bu - backward horizontal displacement
-//! bv - backward vertical displacement
-//! pos - new frame position
-//! newFrame - new frame
-//! buf - temporary buffer, will have width x 6*height size, CV_32FC1 type and contain 6 GpuMat;
-//! occlusion masks 0, occlusion masks 1,
-//! interpolated forward flow 0, interpolated forward flow 1,
-//! interpolated backward flow 0, interpolated backward flow 1
-//!
-CV_EXPORTS void interpolateFrames(const GpuMat& frame0, const GpuMat& frame1,
- const GpuMat& fu, const GpuMat& fv,
- const GpuMat& bu, const GpuMat& bv,
- float pos, GpuMat& newFrame, GpuMat& buf,
- Stream& stream = Stream::Null());
-
-CV_EXPORTS void createOpticalFlowNeedleMap(const GpuMat& u, const GpuMat& v, GpuMat& vertex, GpuMat& colors);
-
-
-//////////////////////// Background/foreground segmentation ////////////////////////
-
-// Foreground Object Detection from Videos Containing Complex Background.
-// Liyuan Li, Weimin Huang, Irene Y.H. Gu, and Qi Tian.
-// ACM MM2003 9p
-class CV_EXPORTS FGDStatModel
+struct CV_EXPORTS HOGDescriptor
{
-public:
- struct CV_EXPORTS Params
- {
- int Lc; // Quantized levels per 'color' component. Power of two, typically 32, 64 or 128.
- int N1c; // Number of color vectors used to model normal background color variation at a given pixel.
- int N2c; // Number of color vectors retained at given pixel. Must be > N1c, typically ~ 5/3 of N1c.
- // Used to allow the first N1c vectors to adapt over time to changing background.
-
- int Lcc; // Quantized levels per 'color co-occurrence' component. Power of two, typically 16, 32 or 64.
- int N1cc; // Number of color co-occurrence vectors used to model normal background color variation at a given pixel.
- int N2cc; // Number of color co-occurrence vectors retained at given pixel. Must be > N1cc, typically ~ 5/3 of N1cc.
- // Used to allow the first N1cc vectors to adapt over time to changing background.
-
- bool is_obj_without_holes; // If TRUE we ignore holes within foreground blobs. Defaults to TRUE.
- int perform_morphing; // Number of erode-dilate-erode foreground-blob cleanup iterations.
- // These erase one-pixel junk blobs and merge almost-touching blobs. Default value is 1.
-
- float alpha1; // How quickly we forget old background pixel values seen. Typically set to 0.1.
- float alpha2; // "Controls speed of feature learning". Depends on T. Typical value circa 0.005.
- float alpha3; // Alternate to alpha2, used (e.g.) for quicker initial convergence. Typical value 0.1.
-
- float delta; // Affects color and color co-occurrence quantization, typically set to 2.
- float T; // A percentage value which determines when new features can be recognized as new background. (Typically 0.9).
- float minArea; // Discard foreground blobs whose bounding box is smaller than this threshold.
-
- // default Params
- Params();
- };
+ enum { DEFAULT_WIN_SIGMA = -1 };
+ enum { DEFAULT_NLEVELS = 64 };
+ enum { DESCR_FORMAT_ROW_BY_ROW, DESCR_FORMAT_COL_BY_COL };
- // out_cn - channels count in output result (can be 3 or 4)
- // 4-channels require more memory, but a bit faster
- explicit FGDStatModel(int out_cn = 3);
- explicit FGDStatModel(const cv::gpu::GpuMat& firstFrame, const Params& params = Params(), int out_cn = 3);
+ HOGDescriptor(Size win_size=Size(64, 128), Size block_size=Size(16, 16),
+ Size block_stride=Size(8, 8), Size cell_size=Size(8, 8),
+ int nbins=9, double win_sigma=DEFAULT_WIN_SIGMA,
+ double threshold_L2hys=0.2, bool gamma_correction=true,
+ int nlevels=DEFAULT_NLEVELS);
- ~FGDStatModel();
+ size_t getDescriptorSize() const;
+ size_t getBlockHistogramSize() const;
- void create(const cv::gpu::GpuMat& firstFrame, const Params& params = Params());
- void release();
+ void setSVMDetector(const std::vector<float>& detector);
- int update(const cv::gpu::GpuMat& curFrame);
+ static std::vector<float> getDefaultPeopleDetector();
+ static std::vector<float> getPeopleDetector48x96();
+ static std::vector<float> getPeopleDetector64x128();
- //8UC3 or 8UC4 reference background image
- cv::gpu::GpuMat background;
+ void detect(const GpuMat& img, std::vector<Point>& found_locations,
+ double hit_threshold=0, Size win_stride=Size(),
+ Size padding=Size());
- //8UC1 foreground image
- cv::gpu::GpuMat foreground;
+ void detectMultiScale(const GpuMat& img, std::vector<Rect>& found_locations,
+ double hit_threshold=0, Size win_stride=Size(),
+ Size padding=Size(), double scale0=1.05,
+ int group_threshold=2);
- std::vector< std::vector<cv::Point> > foreground_regions;
+ void computeConfidence(const GpuMat& img, std::vector<Point>& hits, double hit_threshold,
+ Size win_stride, Size padding, std::vector<Point>& locations, std::vector<double>& confidences);
-private:
- FGDStatModel(const FGDStatModel&);
- FGDStatModel& operator=(const FGDStatModel&);
+ void computeConfidenceMultiScale(const GpuMat& img, std::vector<Rect>& found_locations,
+ double hit_threshold, Size win_stride, Size padding,
+ std::vector<HOGConfidence> &conf_out, int group_threshold);
- class Impl;
- std::auto_ptr<Impl> impl_;
-};
+ void getDescriptors(const GpuMat& img, Size win_stride,
+ GpuMat& descriptors,
+ int descr_format=DESCR_FORMAT_COL_BY_COL);
-/*!
- Gaussian Mixture-based Backbround/Foreground Segmentation Algorithm
+ Size win_size;
+ Size block_size;
+ Size block_stride;
+ Size cell_size;
+ int nbins;
+ double win_sigma;
+ double threshold_L2hys;
+ bool gamma_correction;
+ int nlevels;
- The class implements the following algorithm:
- "An improved adaptive background mixture model for real-time tracking with shadow detection"
- P. KadewTraKuPong and R. Bowden,
- Proc. 2nd European Workshp on Advanced Video-Based Surveillance Systems, 2001."
- http://personal.ee.surrey.ac.uk/Personal/R.Bowden/publications/avbs01/avbs01.pdf
-*/
-class CV_EXPORTS MOG_GPU
-{
-public:
- //! the default constructor
- MOG_GPU(int nmixtures = -1);
+protected:
+ void computeBlockHistograms(const GpuMat& img);
+ void computeGradient(const GpuMat& img, GpuMat& grad, GpuMat& qangle);
- //! re-initiaization method
- void initialize(Size frameSize, int frameType);
+ double getWinSigma() const;
+ bool checkDetectorSize() const;
- //! the update operator
- void operator()(const GpuMat& frame, GpuMat& fgmask, float learningRate = 0.0f, Stream& stream = Stream::Null());
+ static int numPartsWithin(int size, int part_size, int stride);
+ static Size numPartsWithin(Size size, Size part_size, Size stride);
- //! computes a background image which are the mean of all background gaussians
- void getBackgroundImage(GpuMat& backgroundImage, Stream& stream = Stream::Null()) const;
+ // Coefficients of the separating plane
+ float free_coef;
+ GpuMat detector;
- //! releases all inner buffers
- void release();
+ // Results of the last classification step
+ GpuMat labels, labels_buf;
+ Mat labels_host;
- int history;
- float varThreshold;
- float backgroundRatio;
- float noiseSigma;
+ // Results of the last histogram evaluation step
+ GpuMat block_hists, block_hists_buf;
-private:
- int nmixtures_;
+ // Gradients conputation results
+ GpuMat grad, qangle, grad_buf, qangle_buf;
- Size frameSize_;
- int frameType_;
- int nframes_;
+ // returns subbuffer with required size, reallocates buffer if nessesary.
+ static GpuMat getBuffer(const Size& sz, int type, GpuMat& buf);
+ static GpuMat getBuffer(int rows, int cols, int type, GpuMat& buf);
- GpuMat weight_;
- GpuMat sortKey_;
- GpuMat mean_;
- GpuMat var_;
+ std::vector<GpuMat> image_scales;
};
-/*!
- The class implements the following algorithm:
- "Improved adaptive Gausian mixture model for background subtraction"
- Z.Zivkovic
- International Conference Pattern Recognition, UK, August, 2004.
- http://www.zoranz.net/Publications/zivkovic2004ICPR.pdf
-*/
-class CV_EXPORTS MOG2_GPU
-{
-public:
- //! the default constructor
- MOG2_GPU(int nmixtures = -1);
-
- //! re-initiaization method
- void initialize(Size frameSize, int frameType);
-
- //! the update operator
- void operator()(const GpuMat& frame, GpuMat& fgmask, float learningRate = -1.0f, Stream& stream = Stream::Null());
-
- //! computes a background image which are the mean of all background gaussians
- void getBackgroundImage(GpuMat& backgroundImage, Stream& stream = Stream::Null()) const;
-
- //! releases all inner buffers
- void release();
-
- // parameters
- // you should call initialize after parameters changes
-
- int history;
-
- //! here it is the maximum allowed number of mixture components.
- //! Actual number is determined dynamically per pixel
- float varThreshold;
- // threshold on the squared Mahalanobis distance to decide if it is well described
- // by the background model or not. Related to Cthr from the paper.
- // This does not influence the update of the background. A typical value could be 4 sigma
- // and that is varThreshold=4*4=16; Corresponds to Tb in the paper.
-
- /////////////////////////
- // less important parameters - things you might change but be carefull
- ////////////////////////
-
- float backgroundRatio;
- // corresponds to fTB=1-cf from the paper
- // TB - threshold when the component becomes significant enough to be included into
- // the background model. It is the TB=1-cf from the paper. So I use cf=0.1 => TB=0.
- // For alpha=0.001 it means that the mode should exist for approximately 105 frames before
- // it is considered foreground
- // float noiseSigma;
- float varThresholdGen;
-
- //correspondts to Tg - threshold on the squared Mahalan. dist. to decide
- //when a sample is close to the existing components. If it is not close
- //to any a new component will be generated. I use 3 sigma => Tg=3*3=9.
- //Smaller Tg leads to more generated components and higher Tg might make
- //lead to small number of components but they can grow too large
- float fVarInit;
- float fVarMin;
- float fVarMax;
-
- //initial variance for the newly generated components.
- //It will will influence the speed of adaptation. A good guess should be made.
- //A simple way is to estimate the typical standard deviation from the images.
- //I used here 10 as a reasonable value
- // min and max can be used to further control the variance
- float fCT; //CT - complexity reduction prior
- //this is related to the number of samples needed to accept that a component
- //actually exists. We use CT=0.05 of all the samples. By setting CT=0 you get
- //the standard Stauffer&Grimson algorithm (maybe not exact but very similar)
-
- //shadow detection parameters
- bool bShadowDetection; //default 1 - do shadow detection
- unsigned char nShadowDetection; //do shadow detection - insert this value as the detection result - 127 default value
- float fTau;
- // Tau - shadow threshold. The shadow is detected if the pixel is darker
- //version of the background. Tau is a threshold on how much darker the shadow can be.
- //Tau= 0.5 means that if pixel is more than 2 times darker then it is not shadow
- //See: Prati,Mikic,Trivedi,Cucchiarra,"Detecting Moving Shadows...",IEEE PAMI,2003.
-
-private:
- int nmixtures_;
-
- Size frameSize_;
- int frameType_;
- int nframes_;
-
- GpuMat weight_;
- GpuMat variance_;
- GpuMat mean_;
-
- GpuMat bgmodelUsedModes_; //keep track of number of modes per pixel
-};
+//////////////////////////// CascadeClassifier ////////////////////////////
-/**
- * Background Subtractor module. Takes a series of images and returns a sequence of mask (8UC1)
- * images of the same size, where 255 indicates Foreground and 0 represents Background.
- * This class implements an algorithm described in "Visual Tracking of Human Visitors under
- * Variable-Lighting Conditions for a Responsive Audio Art Installation," A. Godbehere,
- * A. Matsukawa, K. Goldberg, American Control Conference, Montreal, June 2012.
- */
-class CV_EXPORTS GMG_GPU
+// The cascade classifier class for object detection: supports old haar and new lbp xlm formats and nvbin for haar cascades olny.
+class CV_EXPORTS CascadeClassifier_GPU
{
public:
- GMG_GPU();
-
- /**
- * Validate parameters and set up data structures for appropriate frame size.
- * @param frameSize Input frame size
- * @param min Minimum value taken on by pixels in image sequence. Usually 0
- * @param max Maximum value taken on by pixels in image sequence. e.g. 1.0 or 255
- */
- void initialize(Size frameSize, float min = 0.0f, float max = 255.0f);
-
- /**
- * Performs single-frame background subtraction and builds up a statistical background image
- * model.
- * @param frame Input frame
- * @param fgmask Output mask image representing foreground and background pixels
- * @param stream Stream for the asynchronous version
- */
- void operator ()(const GpuMat& frame, GpuMat& fgmask, float learningRate = -1.0f, Stream& stream = Stream::Null());
+ CascadeClassifier_GPU();
+ CascadeClassifier_GPU(const String& filename);
+ ~CascadeClassifier_GPU();
- //! Releases all inner buffers
+ bool empty() const;
+ bool load(const String& filename);
void release();
- //! Total number of distinct colors to maintain in histogram.
- int maxFeatures;
-
- //! Set between 0.0 and 1.0, determines how quickly features are "forgotten" from histograms.
- float learningRate;
-
- //! Number of frames of video to use to initialize histograms.
- int numInitializationFrames;
-
- //! Number of discrete levels in each channel to be used in histograms.
- int quantizationLevels;
-
- //! Prior probability that any given pixel is a background pixel. A sensitivity parameter.
- float backgroundPrior;
-
- //! Value above which pixel is determined to be FG.
- float decisionThreshold;
-
- //! Smoothing radius, in pixels, for cleaning up FG image.
- int smoothingRadius;
-
- //! Perform background model update.
- bool updateBackgroundModel;
-
-private:
- float maxVal_, minVal_;
-
- Size frameSize_;
-
- int frameNum_;
-
- GpuMat nfeatures_;
- GpuMat colors_;
- GpuMat weights_;
-
- Ptr<FilterEngine_GPU> boxFilter_;
- GpuMat buf_;
-};
-
-////////////////////////////////// Video Encoding //////////////////////////////////
-
-// Works only under Windows
-// Supports olny H264 video codec and AVI files
-class CV_EXPORTS VideoWriter_GPU
-{
-public:
- struct EncoderParams;
-
- // Callbacks for video encoder, use it if you want to work with raw video stream
- class EncoderCallBack;
-
- enum SurfaceFormat
- {
- SF_UYVY = 0,
- SF_YUY2,
- SF_YV12,
- SF_NV12,
- SF_IYUV,
- SF_BGR,
- SF_GRAY = SF_BGR
- };
-
- VideoWriter_GPU();
- VideoWriter_GPU(const String& fileName, cv::Size frameSize, double fps, SurfaceFormat format = SF_BGR);
- VideoWriter_GPU(const String& fileName, cv::Size frameSize, double fps, const EncoderParams& params, SurfaceFormat format = SF_BGR);
- VideoWriter_GPU(const cv::Ptr<EncoderCallBack>& encoderCallback, cv::Size frameSize, double fps, SurfaceFormat format = SF_BGR);
- VideoWriter_GPU(const cv::Ptr<EncoderCallBack>& encoderCallback, cv::Size frameSize, double fps, const EncoderParams& params, SurfaceFormat format = SF_BGR);
- ~VideoWriter_GPU();
-
- // all methods throws cv::Exception if error occurs
- void open(const String& fileName, cv::Size frameSize, double fps, SurfaceFormat format = SF_BGR);
- void open(const String& fileName, cv::Size frameSize, double fps, const EncoderParams& params, SurfaceFormat format = SF_BGR);
- void open(const cv::Ptr<EncoderCallBack>& encoderCallback, cv::Size frameSize, double fps, SurfaceFormat format = SF_BGR);
- void open(const cv::Ptr<EncoderCallBack>& encoderCallback, cv::Size frameSize, double fps, const EncoderParams& params, SurfaceFormat format = SF_BGR);
-
- bool isOpened() const;
- void close();
-
- void write(const cv::gpu::GpuMat& image, bool lastFrame = false);
-
- struct CV_EXPORTS EncoderParams
- {
- int P_Interval; // NVVE_P_INTERVAL,
- int IDR_Period; // NVVE_IDR_PERIOD,
- int DynamicGOP; // NVVE_DYNAMIC_GOP,
- int RCType; // NVVE_RC_TYPE,
- int AvgBitrate; // NVVE_AVG_BITRATE,
- int PeakBitrate; // NVVE_PEAK_BITRATE,
- int QP_Level_Intra; // NVVE_QP_LEVEL_INTRA,
- int QP_Level_InterP; // NVVE_QP_LEVEL_INTER_P,
- int QP_Level_InterB; // NVVE_QP_LEVEL_INTER_B,
- int DeblockMode; // NVVE_DEBLOCK_MODE,
- int ProfileLevel; // NVVE_PROFILE_LEVEL,
- int ForceIntra; // NVVE_FORCE_INTRA,
- int ForceIDR; // NVVE_FORCE_IDR,
- int ClearStat; // NVVE_CLEAR_STAT,
- int DIMode; // NVVE_SET_DEINTERLACE,
- int Presets; // NVVE_PRESETS,
- int DisableCabac; // NVVE_DISABLE_CABAC,
- int NaluFramingType; // NVVE_CONFIGURE_NALU_FRAMING_TYPE
- int DisableSPSPPS; // NVVE_DISABLE_SPS_PPS
-
- EncoderParams();
- explicit EncoderParams(const String& configFile);
-
- void load(const String& configFile);
- void save(const String& configFile) const;
- };
-
- EncoderParams getParams() const;
-
- class CV_EXPORTS EncoderCallBack
- {
- public:
- enum PicType
- {
- IFRAME = 1,
- PFRAME = 2,
- BFRAME = 3
- };
-
- virtual ~EncoderCallBack() {}
-
- // callback function to signal the start of bitstream that is to be encoded
- // must return pointer to buffer
- virtual uchar* acquireBitStream(int* bufferSize) = 0;
-
- // callback function to signal that the encoded bitstream is ready to be written to file
- virtual void releaseBitStream(unsigned char* data, int size) = 0;
+ /* returns number of detected objects */
+ int detectMultiScale(const GpuMat& image, GpuMat& objectsBuf, double scaleFactor = 1.2, int minNeighbors = 4, Size minSize = Size());
+ int detectMultiScale(const GpuMat& image, GpuMat& objectsBuf, Size maxObjectSize, Size minSize = Size(), double scaleFactor = 1.1, int minNeighbors = 4);
- // callback function to signal that the encoding operation on the frame has started
- virtual void onBeginFrame(int frameNumber, PicType picType) = 0;
+ bool findLargestObject;
+ bool visualizeInPlace;
- // callback function signals that the encoding operation on the frame has finished
- virtual void onEndFrame(int frameNumber, PicType picType) = 0;
- };
+ Size getClassifierSize() const;
private:
- VideoWriter_GPU(const VideoWriter_GPU&);
- VideoWriter_GPU& operator=(const VideoWriter_GPU&);
-
- class Impl;
- std::auto_ptr<Impl> impl_;
+ struct CascadeClassifierImpl;
+ CascadeClassifierImpl* impl;
+ struct HaarCascade;
+ struct LbpCascade;
+ friend class CascadeClassifier_GPU_LBP;
};
+//////////////////////////// Labeling ////////////////////////////
-////////////////////////////////// Video Decoding //////////////////////////////////////////
-
-namespace detail
-{
- class FrameQueue;
- class VideoParser;
-}
-
-class CV_EXPORTS VideoReader_GPU
-{
-public:
- enum Codec
- {
- MPEG1 = 0,
- MPEG2,
- MPEG4,
- VC1,
- H264,
- JPEG,
- H264_SVC,
- H264_MVC,
-
- Uncompressed_YUV420 = (('I'<<24)|('Y'<<16)|('U'<<8)|('V')), // Y,U,V (4:2:0)
- Uncompressed_YV12 = (('Y'<<24)|('V'<<16)|('1'<<8)|('2')), // Y,V,U (4:2:0)
- Uncompressed_NV12 = (('N'<<24)|('V'<<16)|('1'<<8)|('2')), // Y,UV (4:2:0)
- Uncompressed_YUYV = (('Y'<<24)|('U'<<16)|('Y'<<8)|('V')), // YUYV/YUY2 (4:2:2)
- Uncompressed_UYVY = (('U'<<24)|('Y'<<16)|('V'<<8)|('Y')), // UYVY (4:2:2)
- };
-
- enum ChromaFormat
- {
- Monochrome=0,
- YUV420,
- YUV422,
- YUV444,
- };
-
- struct FormatInfo
- {
- Codec codec;
- ChromaFormat chromaFormat;
- int width;
- int height;
- };
-
- class VideoSource;
-
- VideoReader_GPU();
- explicit VideoReader_GPU(const String& filename);
- explicit VideoReader_GPU(const cv::Ptr<VideoSource>& source);
-
- ~VideoReader_GPU();
-
- void open(const String& filename);
- void open(const cv::Ptr<VideoSource>& source);
- bool isOpened() const;
-
- void close();
-
- bool read(GpuMat& image);
-
- FormatInfo format() const;
- void dumpFormat(std::ostream& st);
+//!performs labeling via graph cuts of a 2D regular 4-connected graph.
+CV_EXPORTS void graphcut(GpuMat& terminals, GpuMat& leftTransp, GpuMat& rightTransp, GpuMat& top, GpuMat& bottom, GpuMat& labels,
+ GpuMat& buf, Stream& stream = Stream::Null());
- class CV_EXPORTS VideoSource
- {
- public:
- VideoSource() : frameQueue_(0), videoParser_(0) {}
- virtual ~VideoSource() {}
+//!performs labeling via graph cuts of a 2D regular 8-connected graph.
+CV_EXPORTS void graphcut(GpuMat& terminals, GpuMat& leftTransp, GpuMat& rightTransp, GpuMat& top, GpuMat& topLeft, GpuMat& topRight,
+ GpuMat& bottom, GpuMat& bottomLeft, GpuMat& bottomRight,
+ GpuMat& labels,
+ GpuMat& buf, Stream& stream = Stream::Null());
- virtual FormatInfo format() const = 0;
- virtual void start() = 0;
- virtual void stop() = 0;
- virtual bool isStarted() const = 0;
- virtual bool hasError() const = 0;
+//! compute mask for Generalized Flood fill componetns labeling.
+CV_EXPORTS void connectivityMask(const GpuMat& image, GpuMat& mask, const cv::Scalar& lo, const cv::Scalar& hi, Stream& stream = Stream::Null());
- void setFrameQueue(detail::FrameQueue* frameQueue) { frameQueue_ = frameQueue; }
- void setVideoParser(detail::VideoParser* videoParser) { videoParser_ = videoParser; }
+//! performs connected componnents labeling.
+CV_EXPORTS void labelComponents(const GpuMat& mask, GpuMat& components, int flags = 0, Stream& stream = Stream::Null());
- protected:
- bool parseVideoData(const uchar* data, size_t size, bool endOfStream = false);
+//////////////////////////// Calib3d ////////////////////////////
- private:
- VideoSource(const VideoSource&);
- VideoSource& operator =(const VideoSource&);
+CV_EXPORTS void transformPoints(const GpuMat& src, const Mat& rvec, const Mat& tvec,
+ GpuMat& dst, Stream& stream = Stream::Null());
- detail::FrameQueue* frameQueue_;
- detail::VideoParser* videoParser_;
- };
+CV_EXPORTS void projectPoints(const GpuMat& src, const Mat& rvec, const Mat& tvec,
+ const Mat& camera_mat, const Mat& dist_coef, GpuMat& dst,
+ Stream& stream = Stream::Null());
-private:
- VideoReader_GPU(const VideoReader_GPU&);
- VideoReader_GPU& operator =(const VideoReader_GPU&);
+CV_EXPORTS void solvePnPRansac(const Mat& object, const Mat& image, const Mat& camera_mat,
+ const Mat& dist_coef, Mat& rvec, Mat& tvec, bool use_extrinsic_guess=false,
+ int num_iters=100, float max_dist=8.0, int min_inlier_count=100,
+ std::vector<int>* inliers=NULL);
- class Impl;
- std::auto_ptr<Impl> impl_;
-};
+//////////////////////////// VStab ////////////////////////////
//! removes points (CV_32FC2, single row matrix) with zero mask value
CV_EXPORTS void compactPoints(GpuMat &points0, GpuMat &points1, const GpuMat &mask);
int left, int idx, int right, Size size, const Mat &ml, const Mat &mr,
GpuMat &mapx, GpuMat &mapy);
-} // namespace gpu
-
-} // namespace cv
+}} // namespace cv { namespace gpu {
#endif /* __OPENCV_GPU_HPP__ */
using namespace testing;
using namespace perf;
-//////////////////////////////////////////////////////////////////////
-// StereoBM
-
-typedef std::tr1::tuple<string, string> pair_string;
-DEF_PARAM_TEST_1(ImagePair, pair_string);
-
-PERF_TEST_P(ImagePair, Calib3D_StereoBM,
- Values(pair_string("gpu/perf/aloe.png", "gpu/perf/aloeR.png")))
-{
- declare.time(300.0);
-
- const cv::Mat imgLeft = readImage(GET_PARAM(0), cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(imgLeft.empty());
-
- const cv::Mat imgRight = readImage(GET_PARAM(1), cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(imgRight.empty());
-
- const int preset = 0;
- const int ndisp = 256;
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::StereoBM_GPU d_bm(preset, ndisp);
-
- const cv::gpu::GpuMat d_imgLeft(imgLeft);
- const cv::gpu::GpuMat d_imgRight(imgRight);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() d_bm(d_imgLeft, d_imgRight, dst);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- cv::Ptr<cv::StereoBM> bm = cv::createStereoBM(ndisp);
-
- cv::Mat dst;
-
- TEST_CYCLE() bm->compute(imgLeft, imgRight, dst);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// StereoBeliefPropagation
-
-PERF_TEST_P(ImagePair, Calib3D_StereoBeliefPropagation,
- Values(pair_string("gpu/stereobp/aloe-L.png", "gpu/stereobp/aloe-R.png")))
-{
- declare.time(300.0);
-
- const cv::Mat imgLeft = readImage(GET_PARAM(0));
- ASSERT_FALSE(imgLeft.empty());
-
- const cv::Mat imgRight = readImage(GET_PARAM(1));
- ASSERT_FALSE(imgRight.empty());
-
- const int ndisp = 64;
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::StereoBeliefPropagation d_bp(ndisp);
-
- const cv::gpu::GpuMat d_imgLeft(imgLeft);
- const cv::gpu::GpuMat d_imgRight(imgRight);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() d_bp(d_imgLeft, d_imgRight, dst);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// StereoConstantSpaceBP
-
-PERF_TEST_P(ImagePair, Calib3D_StereoConstantSpaceBP,
- Values(pair_string("gpu/stereobm/aloe-L.png", "gpu/stereobm/aloe-R.png")))
-{
- declare.time(300.0);
-
- const cv::Mat imgLeft = readImage(GET_PARAM(0), cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(imgLeft.empty());
-
- const cv::Mat imgRight = readImage(GET_PARAM(1), cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(imgRight.empty());
-
- const int ndisp = 128;
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::StereoConstantSpaceBP d_csbp(ndisp);
-
- const cv::gpu::GpuMat d_imgLeft(imgLeft);
- const cv::gpu::GpuMat d_imgRight(imgRight);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() d_csbp(d_imgLeft, d_imgRight, dst);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// DisparityBilateralFilter
-
-PERF_TEST_P(ImagePair, Calib3D_DisparityBilateralFilter,
- Values(pair_string("gpu/stereobm/aloe-L.png", "gpu/stereobm/aloe-disp.png")))
-{
- const cv::Mat img = readImage(GET_PARAM(0), cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(img.empty());
-
- const cv::Mat disp = readImage(GET_PARAM(1), cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(disp.empty());
-
- const int ndisp = 128;
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::DisparityBilateralFilter d_filter(ndisp);
-
- const cv::gpu::GpuMat d_img(img);
- const cv::gpu::GpuMat d_disp(disp);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() d_filter(d_disp, d_img, dst);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// TransformPoints
-
DEF_PARAM_TEST_1(Count, int);
-PERF_TEST_P(Count, Calib3D_TransformPoints,
- Values(5000, 10000, 20000))
-{
- const int count = GetParam();
-
- cv::Mat src(1, count, CV_32FC3);
- declare.in(src, WARMUP_RNG);
-
- const cv::Mat rvec = cv::Mat::ones(1, 3, CV_32FC1);
- const cv::Mat tvec = cv::Mat::ones(1, 3, CV_32FC1);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::transformPoints(d_src, rvec, tvec, dst);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
//////////////////////////////////////////////////////////////////////
// ProjectPoints
CPU_SANITY_CHECK(tvec, 1e-6);
}
}
-
-//////////////////////////////////////////////////////////////////////
-// ReprojectImageTo3D
-
-PERF_TEST_P(Sz_Depth, Calib3D_ReprojectImageTo3D,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16S)))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
-
- cv::Mat src(size, depth);
- declare.in(src, WARMUP_RNG);
-
- cv::Mat Q(4, 4, CV_32FC1);
- cv::randu(Q, 0.1, 1.0);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::reprojectImageTo3D(d_src, dst, Q);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::reprojectImageTo3D(src, dst, Q);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// DrawColorDisp
-
-PERF_TEST_P(Sz_Depth, Calib3D_DrawColorDisp,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16S)))
-{
- const cv::Size size = GET_PARAM(0);
- const int type = GET_PARAM(1);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::drawColorDisp(d_src, dst, 255);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
+++ /dev/null
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-// By downloading, copying, installing or using the software you agree to this license.
-// If you do not agree to this license, do not download, install,
-// copy or use the software.
-//
-//
-// License Agreement
-// For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-// * Redistribution's of source code must retain the above copyright notice,
-// this list of conditions and the following disclaimer.
-//
-// * Redistribution's in binary form must reproduce the above copyright notice,
-// this list of conditions and the following disclaimer in the documentation
-// and/or other materials provided with the distribution.
-//
-// * The name of the copyright holders may not be used to endorse or promote products
-// derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#include "perf_precomp.hpp"
-
-using namespace std;
-using namespace testing;
-using namespace perf;
-
-//////////////////////////////////////////////////////////////////////
-// Remap
-
-enum { HALF_SIZE=0, UPSIDE_DOWN, REFLECTION_X, REFLECTION_BOTH };
-CV_ENUM(RemapMode, HALF_SIZE, UPSIDE_DOWN, REFLECTION_X, REFLECTION_BOTH);
-
-void generateMap(cv::Mat& map_x, cv::Mat& map_y, int remapMode)
-{
- for (int j = 0; j < map_x.rows; ++j)
- {
- for (int i = 0; i < map_x.cols; ++i)
- {
- switch (remapMode)
- {
- case HALF_SIZE:
- if (i > map_x.cols*0.25 && i < map_x.cols*0.75 && j > map_x.rows*0.25 && j < map_x.rows*0.75)
- {
- map_x.at<float>(j,i) = 2.f * (i - map_x.cols * 0.25f) + 0.5f;
- map_y.at<float>(j,i) = 2.f * (j - map_x.rows * 0.25f) + 0.5f;
- }
- else
- {
- map_x.at<float>(j,i) = 0.f;
- map_y.at<float>(j,i) = 0.f;
- }
- break;
- case UPSIDE_DOWN:
- map_x.at<float>(j,i) = static_cast<float>(i);
- map_y.at<float>(j,i) = static_cast<float>(map_x.rows - j);
- break;
- case REFLECTION_X:
- map_x.at<float>(j,i) = static_cast<float>(map_x.cols - i);
- map_y.at<float>(j,i) = static_cast<float>(j);
- break;
- case REFLECTION_BOTH:
- map_x.at<float>(j,i) = static_cast<float>(map_x.cols - i);
- map_y.at<float>(j,i) = static_cast<float>(map_x.rows - j);
- break;
- } // end of switch
- }
- }
-}
-
-DEF_PARAM_TEST(Sz_Depth_Cn_Inter_Border_Mode, cv::Size, MatDepth, MatCn, Interpolation, BorderMode, RemapMode);
-
-PERF_TEST_P(Sz_Depth_Cn_Inter_Border_Mode, ImgProc_Remap,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32F),
- GPU_CHANNELS_1_3_4,
- Values(Interpolation(cv::INTER_NEAREST), Interpolation(cv::INTER_LINEAR), Interpolation(cv::INTER_CUBIC)),
- ALL_BORDER_MODES,
- RemapMode::all()))
-{
- declare.time(20.0);
-
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
- const int interpolation = GET_PARAM(3);
- const int borderMode = GET_PARAM(4);
- const int remapMode = GET_PARAM(5);
-
- const int type = CV_MAKE_TYPE(depth, channels);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- cv::Mat xmap(size, CV_32FC1);
- cv::Mat ymap(size, CV_32FC1);
- generateMap(xmap, ymap, remapMode);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- const cv::gpu::GpuMat d_xmap(xmap);
- const cv::gpu::GpuMat d_ymap(ymap);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::remap(d_src, dst, d_xmap, d_ymap, interpolation, borderMode);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::remap(src, dst, xmap, ymap, interpolation, borderMode);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// Resize
-
-DEF_PARAM_TEST(Sz_Depth_Cn_Inter_Scale, cv::Size, MatDepth, MatCn, Interpolation, double);
-
-PERF_TEST_P(Sz_Depth_Cn_Inter_Scale, ImgProc_Resize,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32F),
- GPU_CHANNELS_1_3_4,
- Values(Interpolation(cv::INTER_NEAREST), Interpolation(cv::INTER_LINEAR), Interpolation(cv::INTER_CUBIC)),
- Values(0.5, 0.3, 2.0)))
-{
- declare.time(20.0);
-
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
- const int interpolation = GET_PARAM(3);
- const double f = GET_PARAM(4);
-
- const int type = CV_MAKE_TYPE(depth, channels);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::resize(d_src, dst, cv::Size(), f, f, interpolation);
-
- GPU_SANITY_CHECK(dst, 1e-3, ERROR_RELATIVE);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::resize(src, dst, cv::Size(), f, f, interpolation);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// ResizeArea
-
-DEF_PARAM_TEST(Sz_Depth_Cn_Scale, cv::Size, MatDepth, MatCn, double);
-
-PERF_TEST_P(Sz_Depth_Cn_Scale, ImgProc_ResizeArea,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32F),
- GPU_CHANNELS_1_3_4,
- Values(0.2, 0.1, 0.05)))
-{
- declare.time(1.0);
-
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
- const int interpolation = cv::INTER_AREA;
- const double f = GET_PARAM(3);
-
- const int type = CV_MAKE_TYPE(depth, channels);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::resize(d_src, dst, cv::Size(), f, f, interpolation);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::resize(src, dst, cv::Size(), f, f, interpolation);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// WarpAffine
-
-DEF_PARAM_TEST(Sz_Depth_Cn_Inter_Border, cv::Size, MatDepth, MatCn, Interpolation, BorderMode);
-
-PERF_TEST_P(Sz_Depth_Cn_Inter_Border, ImgProc_WarpAffine,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32F),
- GPU_CHANNELS_1_3_4,
- Values(Interpolation(cv::INTER_NEAREST), Interpolation(cv::INTER_LINEAR), Interpolation(cv::INTER_CUBIC)),
- ALL_BORDER_MODES))
-{
- declare.time(20.0);
-
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
- const int interpolation = GET_PARAM(3);
- const int borderMode = GET_PARAM(4);
-
- const int type = CV_MAKE_TYPE(depth, channels);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- const double aplha = CV_PI / 4;
- const double mat[2 * 3] =
- {
- std::cos(aplha), -std::sin(aplha), src.cols / 2,
- std::sin(aplha), std::cos(aplha), 0
- };
- const cv::Mat M(2, 3, CV_64F, (void*) mat);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::warpAffine(d_src, dst, M, size, interpolation, borderMode);
-
- GPU_SANITY_CHECK(dst, 1);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::warpAffine(src, dst, M, size, interpolation, borderMode);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// WarpPerspective
-
-PERF_TEST_P(Sz_Depth_Cn_Inter_Border, ImgProc_WarpPerspective,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32F),
- GPU_CHANNELS_1_3_4,
- Values(Interpolation(cv::INTER_NEAREST), Interpolation(cv::INTER_LINEAR), Interpolation(cv::INTER_CUBIC)),
- ALL_BORDER_MODES))
-{
- declare.time(20.0);
-
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
- const int interpolation = GET_PARAM(3);
- const int borderMode = GET_PARAM(4);
-
- const int type = CV_MAKE_TYPE(depth, channels);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- const double aplha = CV_PI / 4;
- double mat[3][3] = { {std::cos(aplha), -std::sin(aplha), src.cols / 2},
- {std::sin(aplha), std::cos(aplha), 0},
- {0.0, 0.0, 1.0}};
- const cv::Mat M(3, 3, CV_64F, (void*) mat);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::warpPerspective(d_src, dst, M, size, interpolation, borderMode);
-
- GPU_SANITY_CHECK(dst, 1);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::warpPerspective(src, dst, M, size, interpolation, borderMode);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// CopyMakeBorder
-
-DEF_PARAM_TEST(Sz_Depth_Cn_Border, cv::Size, MatDepth, MatCn, BorderMode);
-
-PERF_TEST_P(Sz_Depth_Cn_Border, ImgProc_CopyMakeBorder,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32F),
- GPU_CHANNELS_1_3_4,
- ALL_BORDER_MODES))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
- const int borderMode = GET_PARAM(3);
-
- const int type = CV_MAKE_TYPE(depth, channels);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::copyMakeBorder(d_src, dst, 5, 5, 5, 5, borderMode);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::copyMakeBorder(src, dst, 5, 5, 5, 5, borderMode);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// Threshold
-
-CV_ENUM(ThreshOp, THRESH_BINARY, THRESH_BINARY_INV, THRESH_TRUNC, THRESH_TOZERO, THRESH_TOZERO_INV)
-
-DEF_PARAM_TEST(Sz_Depth_Op, cv::Size, MatDepth, ThreshOp);
-
-PERF_TEST_P(Sz_Depth_Op, ImgProc_Threshold,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32F, CV_64F),
- ThreshOp::all()))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int threshOp = GET_PARAM(2);
-
- cv::Mat src(size, depth);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::threshold(d_src, dst, 100.0, 255.0, threshOp);
-
- GPU_SANITY_CHECK(dst, 1e-10);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::threshold(src, dst, 100.0, 255.0, threshOp);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// Integral
-
-PERF_TEST_P(Sz, ImgProc_Integral,
- GPU_TYPICAL_MAT_SIZES)
-{
- const cv::Size size = GetParam();
-
- cv::Mat src(size, CV_8UC1);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
- cv::gpu::GpuMat d_buf;
-
- TEST_CYCLE() cv::gpu::integralBuffered(d_src, dst, d_buf);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::integral(src, dst);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// IntegralSqr
-
-PERF_TEST_P(Sz, ImgProc_IntegralSqr,
- GPU_TYPICAL_MAT_SIZES)
-{
- const cv::Size size = GetParam();
-
- cv::Mat src(size, CV_8UC1);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::sqrIntegral(d_src, dst);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// HistEvenC1
-
-PERF_TEST_P(Sz_Depth, ImgProc_HistEvenC1,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_16S)))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
-
- cv::Mat src(size, depth);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
- cv::gpu::GpuMat d_buf;
-
- TEST_CYCLE() cv::gpu::histEven(d_src, dst, d_buf, 30, 0, 180);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- const int hbins = 30;
- const float hranges[] = {0.0f, 180.0f};
- const int histSize[] = {hbins};
- const float* ranges[] = {hranges};
- const int channels[] = {0};
-
- cv::Mat dst;
-
- TEST_CYCLE() cv::calcHist(&src, 1, channels, cv::Mat(), dst, 1, histSize, ranges);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// HistEvenC4
-
-PERF_TEST_P(Sz_Depth, ImgProc_HistEvenC4,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_16S)))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
-
- cv::Mat src(size, CV_MAKE_TYPE(depth, 4));
- declare.in(src, WARMUP_RNG);
-
- int histSize[] = {30, 30, 30, 30};
- int lowerLevel[] = {0, 0, 0, 0};
- int upperLevel[] = {180, 180, 180, 180};
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat d_hist[4];
- cv::gpu::GpuMat d_buf;
-
- TEST_CYCLE() cv::gpu::histEven(d_src, d_hist, d_buf, histSize, lowerLevel, upperLevel);
-
- cv::Mat cpu_hist0, cpu_hist1, cpu_hist2, cpu_hist3;
- d_hist[0].download(cpu_hist0);
- d_hist[1].download(cpu_hist1);
- d_hist[2].download(cpu_hist2);
- d_hist[3].download(cpu_hist3);
- SANITY_CHECK(cpu_hist0);
- SANITY_CHECK(cpu_hist1);
- SANITY_CHECK(cpu_hist2);
- SANITY_CHECK(cpu_hist3);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// CalcHist
-
-PERF_TEST_P(Sz, ImgProc_CalcHist,
- GPU_TYPICAL_MAT_SIZES)
-{
- const cv::Size size = GetParam();
-
- cv::Mat src(size, CV_8UC1);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::calcHist(d_src, dst);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// EqualizeHist
-
-PERF_TEST_P(Sz, ImgProc_EqualizeHist,
- GPU_TYPICAL_MAT_SIZES)
-{
- const cv::Size size = GetParam();
-
- cv::Mat src(size, CV_8UC1);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
- cv::gpu::GpuMat d_hist;
- cv::gpu::GpuMat d_buf;
-
- TEST_CYCLE() cv::gpu::equalizeHist(d_src, dst, d_hist, d_buf);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::equalizeHist(src, dst);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-DEF_PARAM_TEST(Sz_ClipLimit, cv::Size, double);
-
-PERF_TEST_P(Sz_ClipLimit, ImgProc_CLAHE,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(0.0, 40.0)))
-{
- const cv::Size size = GET_PARAM(0);
- const double clipLimit = GET_PARAM(1);
-
- cv::Mat src(size, CV_8UC1);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- cv::Ptr<cv::gpu::CLAHE> clahe = cv::gpu::createCLAHE(clipLimit);
- cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() clahe->apply(d_src, dst);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- cv::Ptr<cv::CLAHE> clahe = cv::createCLAHE(clipLimit);
- cv::Mat dst;
-
- TEST_CYCLE() clahe->apply(src, dst);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// ColumnSum
-
-PERF_TEST_P(Sz, ImgProc_ColumnSum,
- GPU_TYPICAL_MAT_SIZES)
-{
- const cv::Size size = GetParam();
-
- cv::Mat src(size, CV_32FC1);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::columnSum(d_src, dst);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// Canny
-
-DEF_PARAM_TEST(Image_AppertureSz_L2gradient, string, int, bool);
-
-PERF_TEST_P(Image_AppertureSz_L2gradient, ImgProc_Canny,
- Combine(Values("perf/800x600.png", "perf/1280x1024.png", "perf/1680x1050.png"),
- Values(3, 5),
- Bool()))
-{
- const string fileName = GET_PARAM(0);
- const int apperture_size = GET_PARAM(1);
- const bool useL2gradient = GET_PARAM(2);
-
- const cv::Mat image = readImage(fileName, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(image.empty());
-
- const double low_thresh = 50.0;
- const double high_thresh = 100.0;
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_image(image);
- cv::gpu::GpuMat dst;
- cv::gpu::CannyBuf d_buf;
-
- TEST_CYCLE() cv::gpu::Canny(d_image, d_buf, dst, low_thresh, high_thresh, apperture_size, useL2gradient);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::Canny(image, dst, low_thresh, high_thresh, apperture_size, useL2gradient);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// MeanShiftFiltering
-
-DEF_PARAM_TEST_1(Image, string);
-
-PERF_TEST_P(Image, ImgProc_MeanShiftFiltering,
- Values<string>("gpu/meanshift/cones.png"))
-{
- declare.time(300.0);
-
- const cv::Mat img = readImage(GetParam());
- ASSERT_FALSE(img.empty());
-
- cv::Mat rgba;
- cv::cvtColor(img, rgba, cv::COLOR_BGR2BGRA);
-
- const int sp = 50;
- const int sr = 50;
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(rgba);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::meanShiftFiltering(d_src, dst, sp, sr);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::pyrMeanShiftFiltering(img, dst, sp, sr);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// MeanShiftProc
-
-PERF_TEST_P(Image, ImgProc_MeanShiftProc,
- Values<string>("gpu/meanshift/cones.png"))
-{
- declare.time(300.0);
-
- const cv::Mat img = readImage(GetParam());
- ASSERT_FALSE(img.empty());
-
- cv::Mat rgba;
- cv::cvtColor(img, rgba, cv::COLOR_BGR2BGRA);
-
- const int sp = 50;
- const int sr = 50;
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(rgba);
- cv::gpu::GpuMat dstr;
- cv::gpu::GpuMat dstsp;
-
- TEST_CYCLE() cv::gpu::meanShiftProc(d_src, dstr, dstsp, sp, sr);
-
- GPU_SANITY_CHECK(dstr);
- GPU_SANITY_CHECK(dstsp);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// MeanShiftSegmentation
-
-PERF_TEST_P(Image, ImgProc_MeanShiftSegmentation,
- Values<string>("gpu/meanshift/cones.png"))
-{
- declare.time(300.0);
-
- const cv::Mat img = readImage(GetParam());
- ASSERT_FALSE(img.empty());
-
- cv::Mat rgba;
- cv::cvtColor(img, rgba, cv::COLOR_BGR2BGRA);
-
- const int sp = 10;
- const int sr = 10;
- const int minsize = 20;
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(rgba);
- cv::Mat dst;
-
- TEST_CYCLE() cv::gpu::meanShiftSegmentation(d_src, dst, sp, sr, minsize);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// BlendLinear
-
-PERF_TEST_P(Sz_Depth_Cn, ImgProc_BlendLinear,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_32F),
- GPU_CHANNELS_1_3_4))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
-
- const int type = CV_MAKE_TYPE(depth, channels);
-
- cv::Mat img1(size, type);
- cv::Mat img2(size, type);
- declare.in(img1, img2, WARMUP_RNG);
-
- const cv::Mat weights1(size, CV_32FC1, cv::Scalar::all(0.5));
- const cv::Mat weights2(size, CV_32FC1, cv::Scalar::all(0.5));
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_img1(img1);
- const cv::gpu::GpuMat d_img2(img2);
- const cv::gpu::GpuMat d_weights1(weights1);
- const cv::gpu::GpuMat d_weights2(weights2);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::blendLinear(d_img1, d_img2, d_weights1, d_weights2, dst);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// Convolve
-
-DEF_PARAM_TEST(Sz_KernelSz_Ccorr, cv::Size, int, bool);
-
-PERF_TEST_P(Sz_KernelSz_Ccorr, ImgProc_Convolve,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(17, 27, 32, 64),
- Bool()))
-{
- declare.time(10.0);
-
- const cv::Size size = GET_PARAM(0);
- const int templ_size = GET_PARAM(1);
- const bool ccorr = GET_PARAM(2);
-
- const cv::Mat image(size, CV_32FC1);
- const cv::Mat templ(templ_size, templ_size, CV_32FC1);
- declare.in(image, templ, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::GpuMat d_image = cv::gpu::createContinuous(size, CV_32FC1);
- d_image.upload(image);
-
- cv::gpu::GpuMat d_templ = cv::gpu::createContinuous(templ_size, templ_size, CV_32FC1);
- d_templ.upload(templ);
-
- cv::gpu::GpuMat dst;
- cv::gpu::ConvolveBuf d_buf;
-
- TEST_CYCLE() cv::gpu::convolve(d_image, d_templ, dst, ccorr, d_buf);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- if (ccorr)
- FAIL_NO_CPU();
-
- cv::Mat dst;
-
- TEST_CYCLE() cv::filter2D(image, dst, image.depth(), templ);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-////////////////////////////////////////////////////////////////////////////////
-// MatchTemplate8U
-
-CV_ENUM(TemplateMethod, TM_SQDIFF, TM_SQDIFF_NORMED, TM_CCORR, TM_CCORR_NORMED, TM_CCOEFF, TM_CCOEFF_NORMED)
-
-DEF_PARAM_TEST(Sz_TemplateSz_Cn_Method, cv::Size, cv::Size, MatCn, TemplateMethod);
-
-PERF_TEST_P(Sz_TemplateSz_Cn_Method, ImgProc_MatchTemplate8U,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(cv::Size(5, 5), cv::Size(16, 16), cv::Size(30, 30)),
- GPU_CHANNELS_1_3_4,
- TemplateMethod::all()))
-{
- declare.time(300.0);
-
- const cv::Size size = GET_PARAM(0);
- const cv::Size templ_size = GET_PARAM(1);
- const int cn = GET_PARAM(2);
- const int method = GET_PARAM(3);
-
- cv::Mat image(size, CV_MAKE_TYPE(CV_8U, cn));
- cv::Mat templ(templ_size, CV_MAKE_TYPE(CV_8U, cn));
- declare.in(image, templ, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_image(image);
- const cv::gpu::GpuMat d_templ(templ);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::matchTemplate(d_image, d_templ, dst, method);
-
- GPU_SANITY_CHECK(dst, 1e-5, ERROR_RELATIVE);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::matchTemplate(image, templ, dst, method);
-
- CPU_SANITY_CHECK(dst);
- }
-};
-
-////////////////////////////////////////////////////////////////////////////////
-// MatchTemplate32F
-
-PERF_TEST_P(Sz_TemplateSz_Cn_Method, ImgProc_MatchTemplate32F,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(cv::Size(5, 5), cv::Size(16, 16), cv::Size(30, 30)),
- GPU_CHANNELS_1_3_4,
- Values(TemplateMethod(cv::TM_SQDIFF), TemplateMethod(cv::TM_CCORR))))
-{
- declare.time(300.0);
-
- const cv::Size size = GET_PARAM(0);
- const cv::Size templ_size = GET_PARAM(1);
- const int cn = GET_PARAM(2);
- int method = GET_PARAM(3);
-
- cv::Mat image(size, CV_MAKE_TYPE(CV_32F, cn));
- cv::Mat templ(templ_size, CV_MAKE_TYPE(CV_32F, cn));
- declare.in(image, templ, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_image(image);
- const cv::gpu::GpuMat d_templ(templ);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::matchTemplate(d_image, d_templ, dst, method);
-
- GPU_SANITY_CHECK(dst, 1e-6, ERROR_RELATIVE);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::matchTemplate(image, templ, dst, method);
-
- CPU_SANITY_CHECK(dst);
- }
-};
-
-//////////////////////////////////////////////////////////////////////
-// MulSpectrums
-
-CV_FLAGS(DftFlags, 0, DFT_INVERSE, DFT_SCALE, DFT_ROWS, DFT_COMPLEX_OUTPUT, DFT_REAL_OUTPUT)
-
-DEF_PARAM_TEST(Sz_Flags, cv::Size, DftFlags);
-
-PERF_TEST_P(Sz_Flags, ImgProc_MulSpectrums,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(0, DftFlags(cv::DFT_ROWS))))
-{
- const cv::Size size = GET_PARAM(0);
- const int flag = GET_PARAM(1);
-
- cv::Mat a(size, CV_32FC2);
- cv::Mat b(size, CV_32FC2);
- declare.in(a, b, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_a(a);
- const cv::gpu::GpuMat d_b(b);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::mulSpectrums(d_a, d_b, dst, flag);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::mulSpectrums(a, b, dst, flag);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// MulAndScaleSpectrums
-
-PERF_TEST_P(Sz, ImgProc_MulAndScaleSpectrums,
- GPU_TYPICAL_MAT_SIZES)
-{
- const cv::Size size = GetParam();
-
- const float scale = 1.f / size.area();
-
- cv::Mat src1(size, CV_32FC2);
- cv::Mat src2(size, CV_32FC2);
- declare.in(src1,src2, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src1(src1);
- const cv::gpu::GpuMat d_src2(src2);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::mulAndScaleSpectrums(d_src1, d_src2, dst, cv::DFT_ROWS, scale, false);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// Dft
-
-PERF_TEST_P(Sz_Flags, ImgProc_Dft,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(0, DftFlags(cv::DFT_ROWS), DftFlags(cv::DFT_INVERSE))))
-{
- declare.time(10.0);
-
- const cv::Size size = GET_PARAM(0);
- const int flag = GET_PARAM(1);
-
- cv::Mat src(size, CV_32FC2);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::dft(d_src, dst, size, flag);
-
- GPU_SANITY_CHECK(dst, 1e-6, ERROR_RELATIVE);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::dft(src, dst, flag);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// CornerHarris
-
-DEF_PARAM_TEST(Image_Type_Border_BlockSz_ApertureSz, string, MatType, BorderMode, int, int);
-
-PERF_TEST_P(Image_Type_Border_BlockSz_ApertureSz, ImgProc_CornerHarris,
- Combine(Values<string>("gpu/stereobm/aloe-L.png"),
- Values(CV_8UC1, CV_32FC1),
- Values(BorderMode(cv::BORDER_REFLECT101), BorderMode(cv::BORDER_REPLICATE), BorderMode(cv::BORDER_REFLECT)),
- Values(3, 5, 7),
- Values(0, 3, 5, 7)))
-{
- const string fileName = GET_PARAM(0);
- const int type = GET_PARAM(1);
- const int borderMode = GET_PARAM(2);
- const int blockSize = GET_PARAM(3);
- const int apertureSize = GET_PARAM(4);
-
- cv::Mat img = readImage(fileName, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(img.empty());
-
- img.convertTo(img, type, type == CV_32F ? 1.0 / 255.0 : 1.0);
-
- const double k = 0.5;
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_img(img);
- cv::gpu::GpuMat dst;
- cv::gpu::GpuMat d_Dx;
- cv::gpu::GpuMat d_Dy;
- cv::gpu::GpuMat d_buf;
-
- TEST_CYCLE() cv::gpu::cornerHarris(d_img, dst, d_Dx, d_Dy, d_buf, blockSize, apertureSize, k, borderMode);
-
- GPU_SANITY_CHECK(dst, 1e-4);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::cornerHarris(img, dst, blockSize, apertureSize, k, borderMode);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// CornerMinEigenVal
-
-PERF_TEST_P(Image_Type_Border_BlockSz_ApertureSz, ImgProc_CornerMinEigenVal,
- Combine(Values<string>("gpu/stereobm/aloe-L.png"),
- Values(CV_8UC1, CV_32FC1),
- Values(BorderMode(cv::BORDER_REFLECT101), BorderMode(cv::BORDER_REPLICATE), BorderMode(cv::BORDER_REFLECT)),
- Values(3, 5, 7),
- Values(0, 3, 5, 7)))
-{
- const string fileName = GET_PARAM(0);
- const int type = GET_PARAM(1);
- const int borderMode = GET_PARAM(2);
- const int blockSize = GET_PARAM(3);
- const int apertureSize = GET_PARAM(4);
-
- cv::Mat img = readImage(fileName, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(img.empty());
-
- img.convertTo(img, type, type == CV_32F ? 1.0 / 255.0 : 1.0);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_img(img);
- cv::gpu::GpuMat dst;
- cv::gpu::GpuMat d_Dx;
- cv::gpu::GpuMat d_Dy;
- cv::gpu::GpuMat d_buf;
-
- TEST_CYCLE() cv::gpu::cornerMinEigenVal(d_img, dst, d_Dx, d_Dy, d_buf, blockSize, apertureSize, borderMode);
-
- GPU_SANITY_CHECK(dst, 1e-4);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::cornerMinEigenVal(img, dst, blockSize, apertureSize, borderMode);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// BuildWarpPlaneMaps
-
-PERF_TEST_P(Sz, ImgProc_BuildWarpPlaneMaps,
- GPU_TYPICAL_MAT_SIZES)
-{
- const cv::Size size = GetParam();
-
- const cv::Mat K = cv::Mat::eye(3, 3, CV_32FC1);
- const cv::Mat R = cv::Mat::ones(3, 3, CV_32FC1);
- const cv::Mat T = cv::Mat::zeros(1, 3, CV_32F);
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::GpuMat map_x;
- cv::gpu::GpuMat map_y;
-
- TEST_CYCLE() cv::gpu::buildWarpPlaneMaps(size, cv::Rect(0, 0, size.width, size.height), K, R, T, 1.0, map_x, map_y);
-
- GPU_SANITY_CHECK(map_x);
- GPU_SANITY_CHECK(map_y);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// BuildWarpCylindricalMaps
-
-PERF_TEST_P(Sz, ImgProc_BuildWarpCylindricalMaps,
- GPU_TYPICAL_MAT_SIZES)
-{
- const cv::Size size = GetParam();
-
- const cv::Mat K = cv::Mat::eye(3, 3, CV_32FC1);
- const cv::Mat R = cv::Mat::ones(3, 3, CV_32FC1);
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::GpuMat map_x;
- cv::gpu::GpuMat map_y;
-
- TEST_CYCLE() cv::gpu::buildWarpCylindricalMaps(size, cv::Rect(0, 0, size.width, size.height), K, R, 1.0, map_x, map_y);
-
- GPU_SANITY_CHECK(map_x);
- GPU_SANITY_CHECK(map_y);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// BuildWarpSphericalMaps
-
-PERF_TEST_P(Sz, ImgProc_BuildWarpSphericalMaps,
- GPU_TYPICAL_MAT_SIZES)
-{
- const cv::Size size = GetParam();
-
- const cv::Mat K = cv::Mat::eye(3, 3, CV_32FC1);
- const cv::Mat R = cv::Mat::ones(3, 3, CV_32FC1);
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::GpuMat map_x;
- cv::gpu::GpuMat map_y;
-
- TEST_CYCLE() cv::gpu::buildWarpSphericalMaps(size, cv::Rect(0, 0, size.width, size.height), K, R, 1.0, map_x, map_y);
-
- GPU_SANITY_CHECK(map_x);
- GPU_SANITY_CHECK(map_y);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// Rotate
-
-DEF_PARAM_TEST(Sz_Depth_Cn_Inter, cv::Size, MatDepth, MatCn, Interpolation);
-
-PERF_TEST_P(Sz_Depth_Cn_Inter, ImgProc_Rotate,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32F),
- GPU_CHANNELS_1_3_4,
- Values(Interpolation(cv::INTER_NEAREST), Interpolation(cv::INTER_LINEAR), Interpolation(cv::INTER_CUBIC))))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
- const int interpolation = GET_PARAM(3);
-
- const int type = CV_MAKE_TYPE(depth, channels);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::rotate(d_src, dst, size, 30.0, 0, 0, interpolation);
-
- GPU_SANITY_CHECK(dst, 1e-3, ERROR_RELATIVE);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// PyrDown
-
-PERF_TEST_P(Sz_Depth_Cn, ImgProc_PyrDown,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32F),
- GPU_CHANNELS_1_3_4))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
-
- const int type = CV_MAKE_TYPE(depth, channels);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::pyrDown(d_src, dst);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::pyrDown(src, dst);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// PyrUp
-
-PERF_TEST_P(Sz_Depth_Cn, ImgProc_PyrUp,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32F),
- GPU_CHANNELS_1_3_4))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
-
- const int type = CV_MAKE_TYPE(depth, channels);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::pyrUp(d_src, dst);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::pyrUp(src, dst);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// CvtColor
-
-DEF_PARAM_TEST(Sz_Depth_Code, cv::Size, MatDepth, CvtColorInfo);
-
-PERF_TEST_P(Sz_Depth_Code, ImgProc_CvtColor,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_32F),
- Values(CvtColorInfo(4, 4, cv::COLOR_RGBA2BGRA),
- CvtColorInfo(4, 1, cv::COLOR_BGRA2GRAY),
- CvtColorInfo(1, 4, cv::COLOR_GRAY2BGRA),
- CvtColorInfo(3, 3, cv::COLOR_BGR2XYZ),
- CvtColorInfo(3, 3, cv::COLOR_XYZ2BGR),
- CvtColorInfo(3, 3, cv::COLOR_BGR2YCrCb),
- CvtColorInfo(3, 3, cv::COLOR_YCrCb2BGR),
- CvtColorInfo(3, 3, cv::COLOR_BGR2YUV),
- CvtColorInfo(3, 3, cv::COLOR_YUV2BGR),
- CvtColorInfo(3, 3, cv::COLOR_BGR2HSV),
- CvtColorInfo(3, 3, cv::COLOR_HSV2BGR),
- CvtColorInfo(3, 3, cv::COLOR_BGR2HLS),
- CvtColorInfo(3, 3, cv::COLOR_HLS2BGR),
- CvtColorInfo(3, 3, cv::COLOR_BGR2Lab),
- CvtColorInfo(3, 3, cv::COLOR_LBGR2Lab),
- CvtColorInfo(3, 3, cv::COLOR_BGR2Luv),
- CvtColorInfo(3, 3, cv::COLOR_LBGR2Luv),
- CvtColorInfo(3, 3, cv::COLOR_Lab2BGR),
- CvtColorInfo(3, 3, cv::COLOR_Lab2LBGR),
- CvtColorInfo(3, 3, cv::COLOR_Luv2RGB),
- CvtColorInfo(3, 3, cv::COLOR_Luv2LRGB))))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const CvtColorInfo info = GET_PARAM(2);
-
- cv::Mat src(size, CV_MAKETYPE(depth, info.scn));
- cv::randu(src, 0, depth == CV_8U ? 255.0 : 1.0);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::cvtColor(d_src, dst, info.code, info.dcn);
-
- GPU_SANITY_CHECK(dst, 1e-4);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::cvtColor(src, dst, info.code, info.dcn);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-PERF_TEST_P(Sz_Depth_Code, ImgProc_CvtColorBayer,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U),
- Values(CvtColorInfo(1, 3, cv::COLOR_BayerBG2BGR),
- CvtColorInfo(1, 3, cv::COLOR_BayerGB2BGR),
- CvtColorInfo(1, 3, cv::COLOR_BayerRG2BGR),
- CvtColorInfo(1, 3, cv::COLOR_BayerGR2BGR),
-
- CvtColorInfo(1, 1, cv::COLOR_BayerBG2GRAY),
- CvtColorInfo(1, 1, cv::COLOR_BayerGB2GRAY),
- CvtColorInfo(1, 1, cv::COLOR_BayerRG2GRAY),
- CvtColorInfo(1, 1, cv::COLOR_BayerGR2GRAY))))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const CvtColorInfo info = GET_PARAM(2);
-
- cv::Mat src(size, CV_MAKETYPE(depth, info.scn));
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::cvtColor(d_src, dst, info.code, info.dcn);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::cvtColor(src, dst, info.code, info.dcn);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-CV_ENUM(DemosaicingCode,
- COLOR_BayerBG2BGR, COLOR_BayerGB2BGR, COLOR_BayerRG2BGR, COLOR_BayerGR2BGR,
- COLOR_BayerBG2GRAY, COLOR_BayerGB2GRAY, COLOR_BayerRG2GRAY, COLOR_BayerGR2GRAY,
- COLOR_BayerBG2BGR_MHT, COLOR_BayerGB2BGR_MHT, COLOR_BayerRG2BGR_MHT, COLOR_BayerGR2BGR_MHT,
- COLOR_BayerBG2GRAY_MHT, COLOR_BayerGB2GRAY_MHT, COLOR_BayerRG2GRAY_MHT, COLOR_BayerGR2GRAY_MHT)
-
-DEF_PARAM_TEST(Sz_Code, cv::Size, DemosaicingCode);
-
-PERF_TEST_P(Sz_Code, ImgProc_Demosaicing,
- Combine(GPU_TYPICAL_MAT_SIZES,
- DemosaicingCode::all()))
-{
- const cv::Size size = GET_PARAM(0);
- const int code = GET_PARAM(1);
-
- cv::Mat src(size, CV_8UC1);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::demosaicing(d_src, dst, code);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- if (code >= cv::COLOR_COLORCVT_MAX)
- {
- FAIL_NO_CPU();
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::cvtColor(src, dst, code);
-
- CPU_SANITY_CHECK(dst);
- }
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// SwapChannels
-
-PERF_TEST_P(Sz, ImgProc_SwapChannels,
- GPU_TYPICAL_MAT_SIZES)
-{
- const cv::Size size = GetParam();
-
- cv::Mat src(size, CV_8UC4);
- declare.in(src, WARMUP_RNG);
-
- const int dstOrder[] = {2, 1, 0, 3};
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::GpuMat dst(src);
-
- TEST_CYCLE() cv::gpu::swapChannels(dst, dstOrder);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// AlphaComp
-
-CV_ENUM(AlphaOp, ALPHA_OVER, ALPHA_IN, ALPHA_OUT, ALPHA_ATOP, ALPHA_XOR, ALPHA_PLUS, ALPHA_OVER_PREMUL, ALPHA_IN_PREMUL, ALPHA_OUT_PREMUL, ALPHA_ATOP_PREMUL, ALPHA_XOR_PREMUL, ALPHA_PLUS_PREMUL, ALPHA_PREMUL)
-
-DEF_PARAM_TEST(Sz_Type_Op, cv::Size, MatType, AlphaOp);
-
-PERF_TEST_P(Sz_Type_Op, ImgProc_AlphaComp,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8UC4, CV_16UC4, CV_32SC4, CV_32FC4),
- AlphaOp::all()))
-{
- const cv::Size size = GET_PARAM(0);
- const int type = GET_PARAM(1);
- const int alpha_op = GET_PARAM(2);
-
- cv::Mat img1(size, type);
- cv::Mat img2(size, type);
- declare.in(img1, img2, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_img1(img1);
- const cv::gpu::GpuMat d_img2(img2);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::alphaComp(d_img1, d_img2, dst, alpha_op);
-
- GPU_SANITY_CHECK(dst, 1e-3, ERROR_RELATIVE);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// ImagePyramidBuild
-
-PERF_TEST_P(Sz_Depth_Cn, ImgProc_ImagePyramidBuild,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32F),
- GPU_CHANNELS_1_3_4))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
-
- const int type = CV_MAKE_TYPE(depth, channels);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- const int nLayers = 5;
- const cv::Size dstSize(size.width / 2 + 10, size.height / 2 + 10);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
-
- cv::gpu::ImagePyramid d_pyr;
-
- TEST_CYCLE() d_pyr.build(d_src, nLayers);
-
- cv::gpu::GpuMat dst;
- d_pyr.getLayer(dst, dstSize);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// ImagePyramidGetLayer
-
-PERF_TEST_P(Sz_Depth_Cn, ImgProc_ImagePyramidGetLayer,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32F),
- GPU_CHANNELS_1_3_4))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
-
- const int type = CV_MAKE_TYPE(depth, channels);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- const int nLayers = 3;
- const cv::Size dstSize(size.width / 2 + 10, size.height / 2 + 10);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- cv::gpu::ImagePyramid d_pyr(d_src, nLayers);
-
- TEST_CYCLE() d_pyr.getLayer(dst, dstSize);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// HoughLines
-
-namespace
-{
- struct Vec4iComparator
- {
- bool operator()(const cv::Vec4i& a, const cv::Vec4i b) const
- {
- if (a[0] != b[0]) return a[0] < b[0];
- else if(a[1] != b[1]) return a[1] < b[1];
- else if(a[2] != b[2]) return a[2] < b[2];
- else return a[3] < b[3];
- }
- };
- struct Vec3fComparator
- {
- bool operator()(const cv::Vec3f& a, const cv::Vec3f b) const
- {
- if(a[0] != b[0]) return a[0] < b[0];
- else if(a[1] != b[1]) return a[1] < b[1];
- else return a[2] < b[2];
- }
- };
- struct Vec2fComparator
- {
- bool operator()(const cv::Vec2f& a, const cv::Vec2f b) const
- {
- if(a[0] != b[0]) return a[0] < b[0];
- else return a[1] < b[1];
- }
- };
-}
-
-PERF_TEST_P(Sz, ImgProc_HoughLines,
- GPU_TYPICAL_MAT_SIZES)
-{
- declare.time(30.0);
-
- const cv::Size size = GetParam();
-
- const float rho = 1.0f;
- const float theta = static_cast<float>(CV_PI / 180.0);
- const int threshold = 300;
-
- cv::Mat src(size, CV_8UC1, cv::Scalar::all(0));
- cv::line(src, cv::Point(0, 100), cv::Point(src.cols, 100), cv::Scalar::all(255), 1);
- cv::line(src, cv::Point(0, 200), cv::Point(src.cols, 200), cv::Scalar::all(255), 1);
- cv::line(src, cv::Point(0, 400), cv::Point(src.cols, 400), cv::Scalar::all(255), 1);
- cv::line(src, cv::Point(100, 0), cv::Point(100, src.rows), cv::Scalar::all(255), 1);
- cv::line(src, cv::Point(200, 0), cv::Point(200, src.rows), cv::Scalar::all(255), 1);
- cv::line(src, cv::Point(400, 0), cv::Point(400, src.rows), cv::Scalar::all(255), 1);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat d_lines;
- cv::gpu::HoughLinesBuf d_buf;
-
- TEST_CYCLE() cv::gpu::HoughLines(d_src, d_lines, d_buf, rho, theta, threshold);
-
- cv::Mat gpu_lines(d_lines.row(0));
- cv::Vec2f* begin = gpu_lines.ptr<cv::Vec2f>(0);
- cv::Vec2f* end = begin + gpu_lines.cols;
- std::sort(begin, end, Vec2fComparator());
- SANITY_CHECK(gpu_lines);
- }
- else
- {
- std::vector<cv::Vec2f> cpu_lines;
-
- TEST_CYCLE() cv::HoughLines(src, cpu_lines, rho, theta, threshold);
-
- SANITY_CHECK(cpu_lines);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// HoughLinesP
-
-DEF_PARAM_TEST_1(Image, std::string);
-
-PERF_TEST_P(Image, ImgProc_HoughLinesP,
- testing::Values("cv/shared/pic5.png", "stitching/a1.png"))
-{
- declare.time(30.0);
-
- const std::string fileName = getDataPath(GetParam());
-
- const float rho = 1.0f;
- const float theta = static_cast<float>(CV_PI / 180.0);
- const int threshold = 100;
- const int minLineLenght = 50;
- const int maxLineGap = 5;
-
- const cv::Mat image = cv::imread(fileName, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(image.empty());
-
- cv::Mat mask;
- cv::Canny(image, mask, 50, 100);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_mask(mask);
- cv::gpu::GpuMat d_lines;
- cv::gpu::HoughLinesBuf d_buf;
-
- TEST_CYCLE() cv::gpu::HoughLinesP(d_mask, d_lines, d_buf, rho, theta, minLineLenght, maxLineGap);
-
- cv::Mat gpu_lines(d_lines);
- cv::Vec4i* begin = gpu_lines.ptr<cv::Vec4i>();
- cv::Vec4i* end = begin + gpu_lines.cols;
- std::sort(begin, end, Vec4iComparator());
- SANITY_CHECK(gpu_lines);
- }
- else
- {
- std::vector<cv::Vec4i> cpu_lines;
-
- TEST_CYCLE() cv::HoughLinesP(mask, cpu_lines, rho, theta, threshold, minLineLenght, maxLineGap);
-
- SANITY_CHECK(cpu_lines);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// HoughCircles
-
-DEF_PARAM_TEST(Sz_Dp_MinDist, cv::Size, float, float);
-
-PERF_TEST_P(Sz_Dp_MinDist, ImgProc_HoughCircles,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(1.0f, 2.0f, 4.0f),
- Values(1.0f)))
-{
- declare.time(30.0);
-
- const cv::Size size = GET_PARAM(0);
- const float dp = GET_PARAM(1);
- const float minDist = GET_PARAM(2);
-
- const int minRadius = 10;
- const int maxRadius = 30;
- const int cannyThreshold = 100;
- const int votesThreshold = 15;
-
- cv::Mat src(size, CV_8UC1, cv::Scalar::all(0));
- cv::circle(src, cv::Point(100, 100), 20, cv::Scalar::all(255), -1);
- cv::circle(src, cv::Point(200, 200), 25, cv::Scalar::all(255), -1);
- cv::circle(src, cv::Point(200, 100), 25, cv::Scalar::all(255), -1);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat d_circles;
- cv::gpu::HoughCirclesBuf d_buf;
-
- TEST_CYCLE() cv::gpu::HoughCircles(d_src, d_circles, d_buf, cv::HOUGH_GRADIENT, dp, minDist, cannyThreshold, votesThreshold, minRadius, maxRadius);
-
- cv::Mat gpu_circles(d_circles);
- cv::Vec3f* begin = gpu_circles.ptr<cv::Vec3f>(0);
- cv::Vec3f* end = begin + gpu_circles.cols;
- std::sort(begin, end, Vec3fComparator());
- SANITY_CHECK(gpu_circles);
- }
- else
- {
- std::vector<cv::Vec3f> cpu_circles;
-
- TEST_CYCLE() cv::HoughCircles(src, cpu_circles, cv::HOUGH_GRADIENT, dp, minDist, cannyThreshold, votesThreshold, minRadius, maxRadius);
-
- SANITY_CHECK(cpu_circles);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// GeneralizedHough
-
-enum { GHT_POSITION = cv::GeneralizedHough::GHT_POSITION,
- GHT_SCALE = cv::GeneralizedHough::GHT_SCALE,
- GHT_ROTATION = cv::GeneralizedHough::GHT_ROTATION
- };
-
-CV_FLAGS(GHMethod, GHT_POSITION, GHT_SCALE, GHT_ROTATION);
-
-DEF_PARAM_TEST(Method_Sz, GHMethod, cv::Size);
-
-PERF_TEST_P(Method_Sz, ImgProc_GeneralizedHough,
- Combine(Values(GHMethod(GHT_POSITION), GHMethod(GHT_POSITION | GHT_SCALE), GHMethod(GHT_POSITION | GHT_ROTATION), GHMethod(GHT_POSITION | GHT_SCALE | GHT_ROTATION)),
- GPU_TYPICAL_MAT_SIZES))
-{
- declare.time(10);
-
- const int method = GET_PARAM(0);
- const cv::Size imageSize = GET_PARAM(1);
-
- const cv::Mat templ = readImage("cv/shared/templ.png", cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(templ.empty());
-
- cv::Mat image(imageSize, CV_8UC1, cv::Scalar::all(0));
- templ.copyTo(image(cv::Rect(50, 50, templ.cols, templ.rows)));
-
- cv::RNG rng(123456789);
- const int objCount = rng.uniform(5, 15);
- for (int i = 0; i < objCount; ++i)
- {
- double scale = rng.uniform(0.7, 1.3);
- bool rotate = 1 == rng.uniform(0, 2);
-
- cv::Mat obj;
- cv::resize(templ, obj, cv::Size(), scale, scale);
- if (rotate)
- obj = obj.t();
-
- cv::Point pos;
-
- pos.x = rng.uniform(0, image.cols - obj.cols);
- pos.y = rng.uniform(0, image.rows - obj.rows);
-
- cv::Mat roi = image(cv::Rect(pos, obj.size()));
- cv::add(roi, obj, roi);
- }
-
- cv::Mat edges;
- cv::Canny(image, edges, 50, 100);
-
- cv::Mat dx, dy;
- cv::Sobel(image, dx, CV_32F, 1, 0);
- cv::Sobel(image, dy, CV_32F, 0, 1);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_edges(edges);
- const cv::gpu::GpuMat d_dx(dx);
- const cv::gpu::GpuMat d_dy(dy);
- cv::gpu::GpuMat posAndVotes;
-
- cv::Ptr<cv::gpu::GeneralizedHough_GPU> d_hough = cv::gpu::GeneralizedHough_GPU::create(method);
- if (method & GHT_ROTATION)
- {
- d_hough->set("maxAngle", 90.0);
- d_hough->set("angleStep", 2.0);
- }
-
- d_hough->setTemplate(cv::gpu::GpuMat(templ));
-
- TEST_CYCLE() d_hough->detect(d_edges, d_dx, d_dy, posAndVotes);
-
- const cv::gpu::GpuMat positions(1, posAndVotes.cols, CV_32FC4, posAndVotes.data);
- GPU_SANITY_CHECK(positions);
- }
- else
- {
- cv::Mat positions;
-
- cv::Ptr<cv::GeneralizedHough> hough = cv::GeneralizedHough::create(method);
- if (method & GHT_ROTATION)
- {
- hough->set("maxAngle", 90.0);
- hough->set("angleStep", 2.0);
- }
-
- hough->setTemplate(templ);
-
- TEST_CYCLE() hough->detect(edges, dx, dy, positions);
-
- CPU_SANITY_CHECK(positions);
- }
-}
#ifndef __OPENCV_PERF_PRECOMP_HPP__
#define __OPENCV_PERF_PRECOMP_HPP__
-#include <cstdio>
-#include <iostream>
-
#include "opencv2/ts.hpp"
#include "opencv2/ts/gpu_perf.hpp"
-#include "opencv2/core.hpp"
-#include "opencv2/highgui.hpp"
#include "opencv2/gpu.hpp"
#include "opencv2/calib3d.hpp"
-#include "opencv2/imgproc.hpp"
-#include "opencv2/video.hpp"
-#include "opencv2/photo.hpp"
-
-#include "opencv2/core/gpu_private.hpp"
+#include "opencv2/objdetect.hpp"
#ifdef GTEST_CREATE_SHARED_LIBRARY
#error no modules except ts should have GTEST_CREATE_SHARED_LIBRARY defined
+++ /dev/null
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-// By downloading, copying, installing or using the software you agree to this license.
-// If you do not agree to this license, do not download, install,
-// copy or use the software.
-//
-//
-// License Agreement
-// For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-// * Redistribution's of source code must retain the above copyright notice,
-// this list of conditions and the following disclaimer.
-//
-// * Redistribution's in binary form must reproduce the above copyright notice,
-// this list of conditions and the following disclaimer in the documentation
-// and/or other materials provided with the distribution.
-//
-// * The name of the copyright holders may not be used to endorse or promote products
-// derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#include "perf_precomp.hpp"
-#include "opencv2/legacy.hpp"
-
-using namespace std;
-using namespace testing;
-using namespace perf;
-
-#if defined(HAVE_XINE) || \
- defined(HAVE_GSTREAMER) || \
- defined(HAVE_QUICKTIME) || \
- defined(HAVE_AVFOUNDATION) || \
- defined(HAVE_FFMPEG) || \
- defined(WIN32) /* assume that we have ffmpeg */
-
-# define BUILD_WITH_VIDEO_INPUT_SUPPORT 1
-#else
-# define BUILD_WITH_VIDEO_INPUT_SUPPORT 0
-#endif
-
-namespace cv
-{
- template<> void Ptr<CvBGStatModel>::delete_obj()
- {
- cvReleaseBGStatModel(&obj);
- }
-}
-
-//////////////////////////////////////////////////////
-// InterpolateFrames
-
-typedef pair<string, string> pair_string;
-
-DEF_PARAM_TEST_1(ImagePair, pair_string);
-
-PERF_TEST_P(ImagePair, Video_InterpolateFrames,
- Values<pair_string>(make_pair("gpu/opticalflow/frame0.png", "gpu/opticalflow/frame1.png")))
-{
- cv::Mat frame0 = readImage(GetParam().first, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(frame0.empty());
-
- cv::Mat frame1 = readImage(GetParam().second, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(frame1.empty());
-
- frame0.convertTo(frame0, CV_32FC1, 1.0 / 255.0);
- frame1.convertTo(frame1, CV_32FC1, 1.0 / 255.0);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_frame0(frame0);
- const cv::gpu::GpuMat d_frame1(frame1);
- cv::gpu::GpuMat d_fu, d_fv;
- cv::gpu::GpuMat d_bu, d_bv;
-
- cv::gpu::BroxOpticalFlow d_flow(0.197f /*alpha*/, 50.0f /*gamma*/, 0.8f /*scale_factor*/,
- 10 /*inner_iterations*/, 77 /*outer_iterations*/, 10 /*solver_iterations*/);
-
- d_flow(d_frame0, d_frame1, d_fu, d_fv);
- d_flow(d_frame1, d_frame0, d_bu, d_bv);
-
- cv::gpu::GpuMat newFrame;
- cv::gpu::GpuMat d_buf;
-
- TEST_CYCLE() cv::gpu::interpolateFrames(d_frame0, d_frame1, d_fu, d_fv, d_bu, d_bv, 0.5f, newFrame, d_buf);
-
- GPU_SANITY_CHECK(newFrame);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////
-// CreateOpticalFlowNeedleMap
-
-PERF_TEST_P(ImagePair, Video_CreateOpticalFlowNeedleMap,
- Values<pair_string>(make_pair("gpu/opticalflow/frame0.png", "gpu/opticalflow/frame1.png")))
-{
- cv::Mat frame0 = readImage(GetParam().first, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(frame0.empty());
-
- cv::Mat frame1 = readImage(GetParam().second, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(frame1.empty());
-
- frame0.convertTo(frame0, CV_32FC1, 1.0 / 255.0);
- frame1.convertTo(frame1, CV_32FC1, 1.0 / 255.0);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_frame0(frame0);
- const cv::gpu::GpuMat d_frame1(frame1);
- cv::gpu::GpuMat u;
- cv::gpu::GpuMat v;
-
- cv::gpu::BroxOpticalFlow d_flow(0.197f /*alpha*/, 50.0f /*gamma*/, 0.8f /*scale_factor*/,
- 10 /*inner_iterations*/, 77 /*outer_iterations*/, 10 /*solver_iterations*/);
-
- d_flow(d_frame0, d_frame1, u, v);
-
- cv::gpu::GpuMat vertex, colors;
-
- TEST_CYCLE() cv::gpu::createOpticalFlowNeedleMap(u, v, vertex, colors);
-
- GPU_SANITY_CHECK(vertex);
- GPU_SANITY_CHECK(colors);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////
-// GoodFeaturesToTrack
-
-DEF_PARAM_TEST(Image_MinDistance, string, double);
-
-PERF_TEST_P(Image_MinDistance, Video_GoodFeaturesToTrack,
- Combine(Values<string>("gpu/perf/aloe.png"),
- Values(0.0, 3.0)))
-{
- const string fileName = GET_PARAM(0);
- const double minDistance = GET_PARAM(1);
-
- const cv::Mat image = readImage(fileName, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(image.empty());
-
- const int maxCorners = 8000;
- const double qualityLevel = 0.01;
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::GoodFeaturesToTrackDetector_GPU d_detector(maxCorners, qualityLevel, minDistance);
-
- const cv::gpu::GpuMat d_image(image);
- cv::gpu::GpuMat pts;
-
- TEST_CYCLE() d_detector(d_image, pts);
-
- GPU_SANITY_CHECK(pts);
- }
- else
- {
- cv::Mat pts;
-
- TEST_CYCLE() cv::goodFeaturesToTrack(image, pts, maxCorners, qualityLevel, minDistance);
-
- CPU_SANITY_CHECK(pts);
- }
-}
-
-//////////////////////////////////////////////////////
-// BroxOpticalFlow
-
-PERF_TEST_P(ImagePair, Video_BroxOpticalFlow,
- Values<pair_string>(make_pair("gpu/opticalflow/frame0.png", "gpu/opticalflow/frame1.png")))
-{
- declare.time(300);
-
- cv::Mat frame0 = readImage(GetParam().first, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(frame0.empty());
-
- cv::Mat frame1 = readImage(GetParam().second, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(frame1.empty());
-
- frame0.convertTo(frame0, CV_32FC1, 1.0 / 255.0);
- frame1.convertTo(frame1, CV_32FC1, 1.0 / 255.0);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_frame0(frame0);
- const cv::gpu::GpuMat d_frame1(frame1);
- cv::gpu::GpuMat u;
- cv::gpu::GpuMat v;
-
- cv::gpu::BroxOpticalFlow d_flow(0.197f /*alpha*/, 50.0f /*gamma*/, 0.8f /*scale_factor*/,
- 10 /*inner_iterations*/, 77 /*outer_iterations*/, 10 /*solver_iterations*/);
-
- TEST_CYCLE() d_flow(d_frame0, d_frame1, u, v);
-
- GPU_SANITY_CHECK(u);
- GPU_SANITY_CHECK(v);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////
-// PyrLKOpticalFlowSparse
-
-DEF_PARAM_TEST(ImagePair_Gray_NPts_WinSz_Levels_Iters, pair_string, bool, int, int, int, int);
-
-PERF_TEST_P(ImagePair_Gray_NPts_WinSz_Levels_Iters, Video_PyrLKOpticalFlowSparse,
- Combine(Values<pair_string>(make_pair("gpu/opticalflow/frame0.png", "gpu/opticalflow/frame1.png")),
- Bool(),
- Values(8000),
- Values(21),
- Values(1, 3),
- Values(1, 30)))
-{
- declare.time(20.0);
-
- const pair_string imagePair = GET_PARAM(0);
- const bool useGray = GET_PARAM(1);
- const int points = GET_PARAM(2);
- const int winSize = GET_PARAM(3);
- const int levels = GET_PARAM(4);
- const int iters = GET_PARAM(5);
-
- const cv::Mat frame0 = readImage(imagePair.first, useGray ? cv::IMREAD_GRAYSCALE : cv::IMREAD_COLOR);
- ASSERT_FALSE(frame0.empty());
-
- const cv::Mat frame1 = readImage(imagePair.second, useGray ? cv::IMREAD_GRAYSCALE : cv::IMREAD_COLOR);
- ASSERT_FALSE(frame1.empty());
-
- cv::Mat gray_frame;
- if (useGray)
- gray_frame = frame0;
- else
- cv::cvtColor(frame0, gray_frame, cv::COLOR_BGR2GRAY);
-
- cv::Mat pts;
- cv::goodFeaturesToTrack(gray_frame, pts, points, 0.01, 0.0);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_pts(pts.reshape(2, 1));
-
- cv::gpu::PyrLKOpticalFlow d_pyrLK;
- d_pyrLK.winSize = cv::Size(winSize, winSize);
- d_pyrLK.maxLevel = levels - 1;
- d_pyrLK.iters = iters;
-
- const cv::gpu::GpuMat d_frame0(frame0);
- const cv::gpu::GpuMat d_frame1(frame1);
- cv::gpu::GpuMat nextPts;
- cv::gpu::GpuMat status;
-
- TEST_CYCLE() d_pyrLK.sparse(d_frame0, d_frame1, d_pts, nextPts, status);
-
- GPU_SANITY_CHECK(nextPts);
- GPU_SANITY_CHECK(status);
- }
- else
- {
- cv::Mat nextPts;
- cv::Mat status;
-
- TEST_CYCLE()
- {
- cv::calcOpticalFlowPyrLK(frame0, frame1, pts, nextPts, status, cv::noArray(),
- cv::Size(winSize, winSize), levels - 1,
- cv::TermCriteria(cv::TermCriteria::COUNT + cv::TermCriteria::EPS, iters, 0.01));
- }
-
- CPU_SANITY_CHECK(nextPts);
- CPU_SANITY_CHECK(status);
- }
-}
-
-//////////////////////////////////////////////////////
-// PyrLKOpticalFlowDense
-
-DEF_PARAM_TEST(ImagePair_WinSz_Levels_Iters, pair_string, int, int, int);
-
-PERF_TEST_P(ImagePair_WinSz_Levels_Iters, Video_PyrLKOpticalFlowDense,
- Combine(Values<pair_string>(make_pair("gpu/opticalflow/frame0.png", "gpu/opticalflow/frame1.png")),
- Values(3, 5, 7, 9, 13, 17, 21),
- Values(1, 3),
- Values(1, 10)))
-{
- declare.time(30);
-
- const pair_string imagePair = GET_PARAM(0);
- const int winSize = GET_PARAM(1);
- const int levels = GET_PARAM(2);
- const int iters = GET_PARAM(3);
-
- const cv::Mat frame0 = readImage(imagePair.first, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(frame0.empty());
-
- const cv::Mat frame1 = readImage(imagePair.second, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(frame1.empty());
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_frame0(frame0);
- const cv::gpu::GpuMat d_frame1(frame1);
- cv::gpu::GpuMat u;
- cv::gpu::GpuMat v;
-
- cv::gpu::PyrLKOpticalFlow d_pyrLK;
- d_pyrLK.winSize = cv::Size(winSize, winSize);
- d_pyrLK.maxLevel = levels - 1;
- d_pyrLK.iters = iters;
-
- TEST_CYCLE() d_pyrLK.dense(d_frame0, d_frame1, u, v);
-
- GPU_SANITY_CHECK(u);
- GPU_SANITY_CHECK(v);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////
-// FarnebackOpticalFlow
-
-PERF_TEST_P(ImagePair, Video_FarnebackOpticalFlow,
- Values<pair_string>(make_pair("gpu/opticalflow/frame0.png", "gpu/opticalflow/frame1.png")))
-{
- declare.time(10);
-
- const cv::Mat frame0 = readImage(GetParam().first, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(frame0.empty());
-
- const cv::Mat frame1 = readImage(GetParam().second, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(frame1.empty());
-
- const int numLevels = 5;
- const double pyrScale = 0.5;
- const int winSize = 13;
- const int numIters = 10;
- const int polyN = 5;
- const double polySigma = 1.1;
- const int flags = 0;
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_frame0(frame0);
- const cv::gpu::GpuMat d_frame1(frame1);
- cv::gpu::GpuMat u;
- cv::gpu::GpuMat v;
-
- cv::gpu::FarnebackOpticalFlow d_farneback;
- d_farneback.numLevels = numLevels;
- d_farneback.pyrScale = pyrScale;
- d_farneback.winSize = winSize;
- d_farneback.numIters = numIters;
- d_farneback.polyN = polyN;
- d_farneback.polySigma = polySigma;
- d_farneback.flags = flags;
-
- TEST_CYCLE() d_farneback(d_frame0, d_frame1, u, v);
-
- GPU_SANITY_CHECK(u, 1e-4);
- GPU_SANITY_CHECK(v, 1e-4);
- }
- else
- {
- cv::Mat flow;
-
- TEST_CYCLE() cv::calcOpticalFlowFarneback(frame0, frame1, flow, pyrScale, numLevels, winSize, numIters, polyN, polySigma, flags);
-
- CPU_SANITY_CHECK(flow);
- }
-}
-
-//////////////////////////////////////////////////////
-// OpticalFlowDual_TVL1
-
-PERF_TEST_P(ImagePair, Video_OpticalFlowDual_TVL1,
- Values<pair_string>(make_pair("gpu/opticalflow/frame0.png", "gpu/opticalflow/frame1.png")))
-{
- declare.time(20);
-
- const cv::Mat frame0 = readImage(GetParam().first, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(frame0.empty());
-
- const cv::Mat frame1 = readImage(GetParam().second, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(frame1.empty());
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_frame0(frame0);
- const cv::gpu::GpuMat d_frame1(frame1);
- cv::gpu::GpuMat u;
- cv::gpu::GpuMat v;
-
- cv::gpu::OpticalFlowDual_TVL1_GPU d_alg;
-
- TEST_CYCLE() d_alg(d_frame0, d_frame1, u, v);
-
- GPU_SANITY_CHECK(u, 1e-2);
- GPU_SANITY_CHECK(v, 1e-2);
- }
- else
- {
- cv::Mat flow;
-
- cv::Ptr<cv::DenseOpticalFlow> alg = cv::createOptFlow_DualTVL1();
- alg->set("medianFiltering", 1);
- alg->set("innerIterations", 1);
- alg->set("outerIterations", 300);
-
- TEST_CYCLE() alg->calc(frame0, frame1, flow);
-
- CPU_SANITY_CHECK(flow);
- }
-}
-
-//////////////////////////////////////////////////////
-// OpticalFlowBM
-
-void calcOpticalFlowBM(const cv::Mat& prev, const cv::Mat& curr,
- cv::Size bSize, cv::Size shiftSize, cv::Size maxRange, int usePrevious,
- cv::Mat& velx, cv::Mat& vely)
-{
- cv::Size sz((curr.cols - bSize.width + shiftSize.width)/shiftSize.width, (curr.rows - bSize.height + shiftSize.height)/shiftSize.height);
-
- velx.create(sz, CV_32FC1);
- vely.create(sz, CV_32FC1);
-
- CvMat cvprev = prev;
- CvMat cvcurr = curr;
-
- CvMat cvvelx = velx;
- CvMat cvvely = vely;
-
- cvCalcOpticalFlowBM(&cvprev, &cvcurr, bSize, shiftSize, maxRange, usePrevious, &cvvelx, &cvvely);
-}
-
-PERF_TEST_P(ImagePair, Video_OpticalFlowBM,
- Values<pair_string>(make_pair("gpu/opticalflow/frame0.png", "gpu/opticalflow/frame1.png")))
-{
- declare.time(400);
-
- const cv::Mat frame0 = readImage(GetParam().first, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(frame0.empty());
-
- const cv::Mat frame1 = readImage(GetParam().second, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(frame1.empty());
-
- const cv::Size block_size(16, 16);
- const cv::Size shift_size(1, 1);
- const cv::Size max_range(16, 16);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_frame0(frame0);
- const cv::gpu::GpuMat d_frame1(frame1);
- cv::gpu::GpuMat u, v, buf;
-
- TEST_CYCLE() cv::gpu::calcOpticalFlowBM(d_frame0, d_frame1, block_size, shift_size, max_range, false, u, v, buf);
-
- GPU_SANITY_CHECK(u);
- GPU_SANITY_CHECK(v);
- }
- else
- {
- cv::Mat u, v;
-
- TEST_CYCLE() calcOpticalFlowBM(frame0, frame1, block_size, shift_size, max_range, false, u, v);
-
- CPU_SANITY_CHECK(u);
- CPU_SANITY_CHECK(v);
- }
-}
-
-PERF_TEST_P(ImagePair, Video_FastOpticalFlowBM,
- Values<pair_string>(make_pair("gpu/opticalflow/frame0.png", "gpu/opticalflow/frame1.png")))
-{
- declare.time(400);
-
- const cv::Mat frame0 = readImage(GetParam().first, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(frame0.empty());
-
- const cv::Mat frame1 = readImage(GetParam().second, cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(frame1.empty());
-
- const cv::Size block_size(16, 16);
- const cv::Size shift_size(1, 1);
- const cv::Size max_range(16, 16);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_frame0(frame0);
- const cv::gpu::GpuMat d_frame1(frame1);
- cv::gpu::GpuMat u, v;
-
- cv::gpu::FastOpticalFlowBM fastBM;
-
- TEST_CYCLE() fastBM(d_frame0, d_frame1, u, v, max_range.width, block_size.width);
-
- GPU_SANITY_CHECK(u, 2);
- GPU_SANITY_CHECK(v, 2);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////
-// FGDStatModel
-
-#if BUILD_WITH_VIDEO_INPUT_SUPPORT
-
-DEF_PARAM_TEST_1(Video, string);
-
-PERF_TEST_P(Video, Video_FGDStatModel,
- Values(string("gpu/video/768x576.avi")))
-{
- declare.time(60);
-
- const string inputFile = perf::TestBase::getDataPath(GetParam());
-
- cv::VideoCapture cap(inputFile);
- ASSERT_TRUE(cap.isOpened());
-
- cv::Mat frame;
- cap >> frame;
- ASSERT_FALSE(frame.empty());
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::GpuMat d_frame(frame);
-
- cv::gpu::FGDStatModel d_model(4);
- d_model.create(d_frame);
-
- for (int i = 0; i < 10; ++i)
- {
- cap >> frame;
- ASSERT_FALSE(frame.empty());
-
- d_frame.upload(frame);
-
- startTimer(); next();
- d_model.update(d_frame);
- stopTimer();
- }
-
- const cv::gpu::GpuMat background = d_model.background;
- const cv::gpu::GpuMat foreground = d_model.foreground;
-
- GPU_SANITY_CHECK(background, 1e-2, ERROR_RELATIVE);
- GPU_SANITY_CHECK(foreground, 1e-2, ERROR_RELATIVE);
- }
- else
- {
- IplImage ipl_frame = frame;
- cv::Ptr<CvBGStatModel> model(cvCreateFGDStatModel(&ipl_frame));
-
- for (int i = 0; i < 10; ++i)
- {
- cap >> frame;
- ASSERT_FALSE(frame.empty());
-
- ipl_frame = frame;
-
- startTimer(); next();
- cvUpdateBGStatModel(&ipl_frame, model);
- stopTimer();
- }
-
- const cv::Mat background = cv::cvarrToMat(model->background);
- const cv::Mat foreground = cv::cvarrToMat(model->foreground);
-
- CPU_SANITY_CHECK(background);
- CPU_SANITY_CHECK(foreground);
- }
-}
-
-#endif
-
-//////////////////////////////////////////////////////
-// MOG
-
-#if BUILD_WITH_VIDEO_INPUT_SUPPORT
-
-DEF_PARAM_TEST(Video_Cn_LearningRate, string, MatCn, double);
-
-PERF_TEST_P(Video_Cn_LearningRate, Video_MOG,
- Combine(Values("gpu/video/768x576.avi", "gpu/video/1920x1080.avi"),
- GPU_CHANNELS_1_3_4,
- Values(0.0, 0.01)))
-{
- const string inputFile = perf::TestBase::getDataPath(GET_PARAM(0));
- const int cn = GET_PARAM(1);
- const float learningRate = static_cast<float>(GET_PARAM(2));
-
- cv::VideoCapture cap(inputFile);
- ASSERT_TRUE(cap.isOpened());
-
- cv::Mat frame;
-
- cap >> frame;
- ASSERT_FALSE(frame.empty());
-
- if (cn != 3)
- {
- cv::Mat temp;
- if (cn == 1)
- cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
- else
- cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
- cv::swap(temp, frame);
- }
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::GpuMat d_frame(frame);
- cv::gpu::MOG_GPU d_mog;
- cv::gpu::GpuMat foreground;
-
- d_mog(d_frame, foreground, learningRate);
-
- for (int i = 0; i < 10; ++i)
- {
- cap >> frame;
- ASSERT_FALSE(frame.empty());
-
- if (cn != 3)
- {
- cv::Mat temp;
- if (cn == 1)
- cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
- else
- cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
- cv::swap(temp, frame);
- }
-
- d_frame.upload(frame);
-
- startTimer(); next();
- d_mog(d_frame, foreground, learningRate);
- stopTimer();
- }
-
- GPU_SANITY_CHECK(foreground);
- }
- else
- {
- cv::Ptr<cv::BackgroundSubtractor> mog = cv::createBackgroundSubtractorMOG();
- cv::Mat foreground;
-
- mog->apply(frame, foreground, learningRate);
-
- for (int i = 0; i < 10; ++i)
- {
- cap >> frame;
- ASSERT_FALSE(frame.empty());
-
- if (cn != 3)
- {
- cv::Mat temp;
- if (cn == 1)
- cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
- else
- cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
- cv::swap(temp, frame);
- }
-
- startTimer(); next();
- mog->apply(frame, foreground, learningRate);
- stopTimer();
- }
-
- CPU_SANITY_CHECK(foreground);
- }
-}
-
-#endif
-
-//////////////////////////////////////////////////////
-// MOG2
-
-#if BUILD_WITH_VIDEO_INPUT_SUPPORT
-
-DEF_PARAM_TEST(Video_Cn, string, int);
-
-PERF_TEST_P(Video_Cn, Video_MOG2,
- Combine(Values("gpu/video/768x576.avi", "gpu/video/1920x1080.avi"),
- GPU_CHANNELS_1_3_4))
-{
- const string inputFile = perf::TestBase::getDataPath(GET_PARAM(0));
- const int cn = GET_PARAM(1);
-
- cv::VideoCapture cap(inputFile);
- ASSERT_TRUE(cap.isOpened());
-
- cv::Mat frame;
-
- cap >> frame;
- ASSERT_FALSE(frame.empty());
-
- if (cn != 3)
- {
- cv::Mat temp;
- if (cn == 1)
- cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
- else
- cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
- cv::swap(temp, frame);
- }
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::MOG2_GPU d_mog2;
- d_mog2.bShadowDetection = false;
-
- cv::gpu::GpuMat d_frame(frame);
- cv::gpu::GpuMat foreground;
-
- d_mog2(d_frame, foreground);
-
- for (int i = 0; i < 10; ++i)
- {
- cap >> frame;
- ASSERT_FALSE(frame.empty());
-
- if (cn != 3)
- {
- cv::Mat temp;
- if (cn == 1)
- cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
- else
- cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
- cv::swap(temp, frame);
- }
-
- d_frame.upload(frame);
-
- startTimer(); next();
- d_mog2(d_frame, foreground);
- stopTimer();
- }
-
- GPU_SANITY_CHECK(foreground);
- }
- else
- {
- cv::Ptr<cv::BackgroundSubtractor> mog2 = cv::createBackgroundSubtractorMOG2();
- mog2->set("detectShadows", false);
-
- cv::Mat foreground;
-
- mog2->apply(frame, foreground);
-
- for (int i = 0; i < 10; ++i)
- {
- cap >> frame;
- ASSERT_FALSE(frame.empty());
-
- if (cn != 3)
- {
- cv::Mat temp;
- if (cn == 1)
- cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
- else
- cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
- cv::swap(temp, frame);
- }
-
- startTimer(); next();
- mog2->apply(frame, foreground);
- stopTimer();
- }
-
- CPU_SANITY_CHECK(foreground);
- }
-}
-
-#endif
-
-//////////////////////////////////////////////////////
-// MOG2GetBackgroundImage
-
-#if BUILD_WITH_VIDEO_INPUT_SUPPORT
-
-PERF_TEST_P(Video_Cn, Video_MOG2GetBackgroundImage,
- Combine(Values("gpu/video/768x576.avi", "gpu/video/1920x1080.avi"),
- GPU_CHANNELS_1_3_4))
-{
- const string inputFile = perf::TestBase::getDataPath(GET_PARAM(0));
- const int cn = GET_PARAM(1);
-
- cv::VideoCapture cap(inputFile);
- ASSERT_TRUE(cap.isOpened());
-
- cv::Mat frame;
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::GpuMat d_frame;
- cv::gpu::MOG2_GPU d_mog2;
- cv::gpu::GpuMat d_foreground;
-
- for (int i = 0; i < 10; ++i)
- {
- cap >> frame;
- ASSERT_FALSE(frame.empty());
-
- if (cn != 3)
- {
- cv::Mat temp;
- if (cn == 1)
- cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
- else
- cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
- cv::swap(temp, frame);
- }
-
- d_frame.upload(frame);
-
- d_mog2(d_frame, d_foreground);
- }
-
- cv::gpu::GpuMat background;
-
- TEST_CYCLE() d_mog2.getBackgroundImage(background);
-
- GPU_SANITY_CHECK(background, 1);
- }
- else
- {
- cv::Ptr<cv::BackgroundSubtractor> mog2 = cv::createBackgroundSubtractorMOG2();
- cv::Mat foreground;
-
- for (int i = 0; i < 10; ++i)
- {
- cap >> frame;
- ASSERT_FALSE(frame.empty());
-
- if (cn != 3)
- {
- cv::Mat temp;
- if (cn == 1)
- cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
- else
- cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
- cv::swap(temp, frame);
- }
-
- mog2->apply(frame, foreground);
- }
-
- cv::Mat background;
-
- TEST_CYCLE() mog2->getBackgroundImage(background);
-
- CPU_SANITY_CHECK(background);
- }
-}
-
-#endif
-
-//////////////////////////////////////////////////////
-// GMG
-
-#if BUILD_WITH_VIDEO_INPUT_SUPPORT
-
-DEF_PARAM_TEST(Video_Cn_MaxFeatures, string, MatCn, int);
-
-PERF_TEST_P(Video_Cn_MaxFeatures, Video_GMG,
- Combine(Values(string("gpu/video/768x576.avi")),
- GPU_CHANNELS_1_3_4,
- Values(20, 40, 60)))
-{
- const std::string inputFile = perf::TestBase::getDataPath(GET_PARAM(0));
- const int cn = GET_PARAM(1);
- const int maxFeatures = GET_PARAM(2);
-
- cv::VideoCapture cap(inputFile);
- ASSERT_TRUE(cap.isOpened());
-
- cv::Mat frame;
- cap >> frame;
- ASSERT_FALSE(frame.empty());
-
- if (cn != 3)
- {
- cv::Mat temp;
- if (cn == 1)
- cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
- else
- cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
- cv::swap(temp, frame);
- }
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::GpuMat d_frame(frame);
- cv::gpu::GpuMat foreground;
-
- cv::gpu::GMG_GPU d_gmg;
- d_gmg.maxFeatures = maxFeatures;
-
- d_gmg(d_frame, foreground);
-
- for (int i = 0; i < 150; ++i)
- {
- cap >> frame;
- if (frame.empty())
- {
- cap.release();
- cap.open(inputFile);
- cap >> frame;
- }
-
- if (cn != 3)
- {
- cv::Mat temp;
- if (cn == 1)
- cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
- else
- cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
- cv::swap(temp, frame);
- }
-
- d_frame.upload(frame);
-
- startTimer(); next();
- d_gmg(d_frame, foreground);
- stopTimer();
- }
-
- GPU_SANITY_CHECK(foreground);
- }
- else
- {
- cv::Mat foreground;
- cv::Mat zeros(frame.size(), CV_8UC1, cv::Scalar::all(0));
-
- cv::Ptr<cv::BackgroundSubtractor> gmg = cv::createBackgroundSubtractorGMG();
- gmg->set("maxFeatures", maxFeatures);
- //gmg.initialize(frame.size(), 0.0, 255.0);
-
- gmg->apply(frame, foreground);
-
- for (int i = 0; i < 150; ++i)
- {
- cap >> frame;
- if (frame.empty())
- {
- cap.release();
- cap.open(inputFile);
- cap >> frame;
- }
-
- if (cn != 3)
- {
- cv::Mat temp;
- if (cn == 1)
- cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
- else
- cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
- cv::swap(temp, frame);
- }
-
- startTimer(); next();
- gmg->apply(frame, foreground);
- stopTimer();
- }
-
- CPU_SANITY_CHECK(foreground);
- }
-}
-
-#endif
-
-//////////////////////////////////////////////////////
-// VideoReader
-
-#if defined(HAVE_NVCUVID) && BUILD_WITH_VIDEO_INPUT_SUPPORT
-
-PERF_TEST_P(Video, DISABLED_Video_VideoReader, Values("gpu/video/768x576.avi", "gpu/video/1920x1080.avi"))
-{
- declare.time(20);
-
- const string inputFile = perf::TestBase::getDataPath(GetParam());
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::VideoReader_GPU d_reader(inputFile);
- ASSERT_TRUE( d_reader.isOpened() );
-
- cv::gpu::GpuMat frame;
-
- TEST_CYCLE_N(10) d_reader.read(frame);
-
- GPU_SANITY_CHECK(frame);
- }
- else
- {
- cv::VideoCapture reader(inputFile);
- ASSERT_TRUE( reader.isOpened() );
-
- cv::Mat frame;
-
- TEST_CYCLE_N(10) reader >> frame;
-
- CPU_SANITY_CHECK(frame);
- }
-}
-
-#endif
-
-//////////////////////////////////////////////////////
-// VideoWriter
-
-#if defined(HAVE_NVCUVID) && defined(WIN32)
-
-PERF_TEST_P(Video, DISABLED_Video_VideoWriter, Values("gpu/video/768x576.avi", "gpu/video/1920x1080.avi"))
-{
- declare.time(30);
-
- const string inputFile = perf::TestBase::getDataPath(GetParam());
- const string outputFile = cv::tempfile(".avi");
-
- const double FPS = 25.0;
-
- cv::VideoCapture reader(inputFile);
- ASSERT_TRUE( reader.isOpened() );
-
- cv::Mat frame;
-
- if (PERF_RUN_GPU())
- {
- cv::gpu::VideoWriter_GPU d_writer;
-
- cv::gpu::GpuMat d_frame;
-
- for (int i = 0; i < 10; ++i)
- {
- reader >> frame;
- ASSERT_FALSE(frame.empty());
-
- d_frame.upload(frame);
-
- if (!d_writer.isOpened())
- d_writer.open(outputFile, frame.size(), FPS);
-
- startTimer(); next();
- d_writer.write(d_frame);
- stopTimer();
- }
- }
- else
- {
- cv::VideoWriter writer;
-
- for (int i = 0; i < 10; ++i)
- {
- reader >> frame;
- ASSERT_FALSE(frame.empty());
-
- if (!writer.isOpened())
- writer.open(outputFile, CV_FOURCC('X', 'V', 'I', 'D'), FPS, frame.size());
-
- startTimer(); next();
- writer.write(frame);
- stopTimer();
- }
- }
-
- SANITY_CHECK(frame);
-}
-
-#endif
-set(PERF4AU_REQUIRED_DEPS opencv_core opencv_imgproc opencv_highgui opencv_video opencv_legacy opencv_gpu opencv_ts)
+set(PERF4AU_REQUIRED_DEPS opencv_core opencv_imgproc opencv_highgui opencv_video opencv_legacy opencv_ml opencv_ts opencv_gpufilters opencv_gpuimgproc opencv_gpuoptflow)
ocv_check_dependencies(${PERF4AU_REQUIRED_DEPS})
set_target_properties(${the_target} PROPERTIES LINK_FLAGS "/NODEFAULTLIB:atlthunk.lib /NODEFAULTLIB:atlsd.lib /DEBUG")
endif()
endif()
-
#include "cvconfig.h"
#include "opencv2/core.hpp"
-#include "opencv2/gpu.hpp"
+#include "opencv2/gpuimgproc.hpp"
+#include "opencv2/gpuoptflow.hpp"
#include "opencv2/highgui.hpp"
#include "opencv2/video.hpp"
#include "opencv2/legacy.hpp"
+++ /dev/null
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-// By downloading, copying, installing or using the software you agree to this license.
-// If you do not agree to this license, do not download, install,
-// copy or use the software.
-//
-//
-// License Agreement
-// For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-// * Redistribution's of source code must retain the above copyright notice,
-// this list of conditions and the following disclaimer.
-//
-// * Redistribution's in binary form must reproduce the above copyright notice,
-// this list of conditions and the following disclaimer in the documentation
-// and/or other materials provided with the distribution.
-//
-// * The name of the copyright holders may not be used to endorse or promote products
-// derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#include "precomp.hpp"
-
-using namespace cv;
-using namespace cv::gpu;
-
-#if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
-
-void cv::gpu::gemm(const GpuMat&, const GpuMat&, double, const GpuMat&, double, GpuMat&, int, Stream&) { throw_no_cuda(); }
-void cv::gpu::transpose(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::flip(const GpuMat&, GpuMat&, int, Stream&) { throw_no_cuda(); }
-void cv::gpu::LUT(const GpuMat&, const Mat&, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::magnitude(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::magnitudeSqr(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::magnitude(const GpuMat&, const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::magnitudeSqr(const GpuMat&, const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::phase(const GpuMat&, const GpuMat&, GpuMat&, bool, Stream&) { throw_no_cuda(); }
-void cv::gpu::cartToPolar(const GpuMat&, const GpuMat&, GpuMat&, GpuMat&, bool, Stream&) { throw_no_cuda(); }
-void cv::gpu::polarToCart(const GpuMat&, const GpuMat&, GpuMat&, GpuMat&, bool, Stream&) { throw_no_cuda(); }
-void cv::gpu::normalize(const GpuMat&, GpuMat&, double, double, int, int, const GpuMat&) { throw_no_cuda(); }
-void cv::gpu::normalize(const GpuMat&, GpuMat&, double, double, int, int, const GpuMat&, GpuMat&, GpuMat&) { throw_no_cuda(); }
-
-#else /* !defined (HAVE_CUDA) */
-
-////////////////////////////////////////////////////////////////////////
-// gemm
-
-void cv::gpu::gemm(const GpuMat& src1, const GpuMat& src2, double alpha, const GpuMat& src3, double beta, GpuMat& dst, int flags, Stream& stream)
-{
-#ifndef HAVE_CUBLAS
- (void)src1;
- (void)src2;
- (void)alpha;
- (void)src3;
- (void)beta;
- (void)dst;
- (void)flags;
- (void)stream;
- CV_Error(cv::Error::StsNotImplemented, "The library was build without CUBLAS");
-#else
- // CUBLAS works with column-major matrices
-
- CV_Assert(src1.type() == CV_32FC1 || src1.type() == CV_32FC2 || src1.type() == CV_64FC1 || src1.type() == CV_64FC2);
- CV_Assert(src2.type() == src1.type() && (src3.empty() || src3.type() == src1.type()));
-
- if (src1.depth() == CV_64F)
- {
- if (!deviceSupports(NATIVE_DOUBLE))
- CV_Error(cv::Error::StsUnsupportedFormat, "The device doesn't support double");
- }
-
- bool tr1 = (flags & GEMM_1_T) != 0;
- bool tr2 = (flags & GEMM_2_T) != 0;
- bool tr3 = (flags & GEMM_3_T) != 0;
-
- if (src1.type() == CV_64FC2)
- {
- if (tr1 || tr2 || tr3)
- CV_Error(cv::Error::StsNotImplemented, "transpose operation doesn't implemented for CV_64FC2 type");
- }
-
- Size src1Size = tr1 ? Size(src1.rows, src1.cols) : src1.size();
- Size src2Size = tr2 ? Size(src2.rows, src2.cols) : src2.size();
- Size src3Size = tr3 ? Size(src3.rows, src3.cols) : src3.size();
- Size dstSize(src2Size.width, src1Size.height);
-
- CV_Assert(src1Size.width == src2Size.height);
- CV_Assert(src3.empty() || src3Size == dstSize);
-
- dst.create(dstSize, src1.type());
-
- if (beta != 0)
- {
- if (src3.empty())
- {
- if (stream)
- stream.enqueueMemSet(dst, Scalar::all(0));
- else
- dst.setTo(Scalar::all(0));
- }
- else
- {
- if (tr3)
- {
- transpose(src3, dst, stream);
- }
- else
- {
- if (stream)
- stream.enqueueCopy(src3, dst);
- else
- src3.copyTo(dst);
- }
- }
- }
-
- cublasHandle_t handle;
- cublasSafeCall( cublasCreate_v2(&handle) );
-
- cublasSafeCall( cublasSetStream_v2(handle, StreamAccessor::getStream(stream)) );
-
- cublasSafeCall( cublasSetPointerMode_v2(handle, CUBLAS_POINTER_MODE_HOST) );
-
- const float alphaf = static_cast<float>(alpha);
- const float betaf = static_cast<float>(beta);
-
- const cuComplex alphacf = make_cuComplex(alphaf, 0);
- const cuComplex betacf = make_cuComplex(betaf, 0);
-
- const cuDoubleComplex alphac = make_cuDoubleComplex(alpha, 0);
- const cuDoubleComplex betac = make_cuDoubleComplex(beta, 0);
-
- cublasOperation_t transa = tr2 ? CUBLAS_OP_T : CUBLAS_OP_N;
- cublasOperation_t transb = tr1 ? CUBLAS_OP_T : CUBLAS_OP_N;
-
- switch (src1.type())
- {
- case CV_32FC1:
- cublasSafeCall( cublasSgemm_v2(handle, transa, transb, tr2 ? src2.rows : src2.cols, tr1 ? src1.cols : src1.rows, tr2 ? src2.cols : src2.rows,
- &alphaf,
- src2.ptr<float>(), static_cast<int>(src2.step / sizeof(float)),
- src1.ptr<float>(), static_cast<int>(src1.step / sizeof(float)),
- &betaf,
- dst.ptr<float>(), static_cast<int>(dst.step / sizeof(float))) );
- break;
-
- case CV_64FC1:
- cublasSafeCall( cublasDgemm_v2(handle, transa, transb, tr2 ? src2.rows : src2.cols, tr1 ? src1.cols : src1.rows, tr2 ? src2.cols : src2.rows,
- &alpha,
- src2.ptr<double>(), static_cast<int>(src2.step / sizeof(double)),
- src1.ptr<double>(), static_cast<int>(src1.step / sizeof(double)),
- &beta,
- dst.ptr<double>(), static_cast<int>(dst.step / sizeof(double))) );
- break;
-
- case CV_32FC2:
- cublasSafeCall( cublasCgemm_v2(handle, transa, transb, tr2 ? src2.rows : src2.cols, tr1 ? src1.cols : src1.rows, tr2 ? src2.cols : src2.rows,
- &alphacf,
- src2.ptr<cuComplex>(), static_cast<int>(src2.step / sizeof(cuComplex)),
- src1.ptr<cuComplex>(), static_cast<int>(src1.step / sizeof(cuComplex)),
- &betacf,
- dst.ptr<cuComplex>(), static_cast<int>(dst.step / sizeof(cuComplex))) );
- break;
-
- case CV_64FC2:
- cublasSafeCall( cublasZgemm_v2(handle, transa, transb, tr2 ? src2.rows : src2.cols, tr1 ? src1.cols : src1.rows, tr2 ? src2.cols : src2.rows,
- &alphac,
- src2.ptr<cuDoubleComplex>(), static_cast<int>(src2.step / sizeof(cuDoubleComplex)),
- src1.ptr<cuDoubleComplex>(), static_cast<int>(src1.step / sizeof(cuDoubleComplex)),
- &betac,
- dst.ptr<cuDoubleComplex>(), static_cast<int>(dst.step / sizeof(cuDoubleComplex))) );
- break;
- }
-
- cublasSafeCall( cublasDestroy_v2(handle) );
-#endif
-}
-
-////////////////////////////////////////////////////////////////////////
-// transpose
-
-void cv::gpu::transpose(const GpuMat& src, GpuMat& dst, Stream& s)
-{
- CV_Assert(src.elemSize() == 1 || src.elemSize() == 4 || src.elemSize() == 8);
-
- dst.create( src.cols, src.rows, src.type() );
-
- cudaStream_t stream = StreamAccessor::getStream(s);
-
- if (src.elemSize() == 1)
- {
- NppStreamHandler h(stream);
-
- NppiSize sz;
- sz.width = src.cols;
- sz.height = src.rows;
-
- nppSafeCall( nppiTranspose_8u_C1R(src.ptr<Npp8u>(), static_cast<int>(src.step),
- dst.ptr<Npp8u>(), static_cast<int>(dst.step), sz) );
- }
- else if (src.elemSize() == 4)
- {
- NppStStreamHandler h(stream);
-
- NcvSize32u sz;
- sz.width = src.cols;
- sz.height = src.rows;
-
- ncvSafeCall( nppiStTranspose_32u_C1R(const_cast<Ncv32u*>(src.ptr<Ncv32u>()), static_cast<int>(src.step),
- dst.ptr<Ncv32u>(), static_cast<int>(dst.step), sz) );
- }
- else // if (src.elemSize() == 8)
- {
- if (!deviceSupports(NATIVE_DOUBLE))
- CV_Error(cv::Error::StsUnsupportedFormat, "The device doesn't support double");
-
- NppStStreamHandler h(stream);
-
- NcvSize32u sz;
- sz.width = src.cols;
- sz.height = src.rows;
-
- ncvSafeCall( nppiStTranspose_64u_C1R(const_cast<Ncv64u*>(src.ptr<Ncv64u>()), static_cast<int>(src.step),
- dst.ptr<Ncv64u>(), static_cast<int>(dst.step), sz) );
- }
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
-}
-
-////////////////////////////////////////////////////////////////////////
-// flip
-
-namespace
-{
- template<int DEPTH> struct NppTypeTraits;
- template<> struct NppTypeTraits<CV_8U> { typedef Npp8u npp_t; };
- template<> struct NppTypeTraits<CV_8S> { typedef Npp8s npp_t; };
- template<> struct NppTypeTraits<CV_16U> { typedef Npp16u npp_t; };
- template<> struct NppTypeTraits<CV_16S> { typedef Npp16s npp_t; };
- template<> struct NppTypeTraits<CV_32S> { typedef Npp32s npp_t; };
- template<> struct NppTypeTraits<CV_32F> { typedef Npp32f npp_t; };
- template<> struct NppTypeTraits<CV_64F> { typedef Npp64f npp_t; };
-
- template <int DEPTH> struct NppMirrorFunc
- {
- typedef typename NppTypeTraits<DEPTH>::npp_t npp_t;
-
- typedef NppStatus (*func_t)(const npp_t* pSrc, int nSrcStep, npp_t* pDst, int nDstStep, NppiSize oROI, NppiAxis flip);
- };
-
- template <int DEPTH, typename NppMirrorFunc<DEPTH>::func_t func> struct NppMirror
- {
- typedef typename NppMirrorFunc<DEPTH>::npp_t npp_t;
-
- static void call(const GpuMat& src, GpuMat& dst, int flipCode, cudaStream_t stream)
- {
- NppStreamHandler h(stream);
-
- NppiSize sz;
- sz.width = src.cols;
- sz.height = src.rows;
-
- nppSafeCall( func(src.ptr<npp_t>(), static_cast<int>(src.step),
- dst.ptr<npp_t>(), static_cast<int>(dst.step), sz,
- (flipCode == 0 ? NPP_HORIZONTAL_AXIS : (flipCode > 0 ? NPP_VERTICAL_AXIS : NPP_BOTH_AXIS))) );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
- };
-}
-
-void cv::gpu::flip(const GpuMat& src, GpuMat& dst, int flipCode, Stream& stream)
-{
- typedef void (*func_t)(const GpuMat& src, GpuMat& dst, int flipCode, cudaStream_t stream);
- static const func_t funcs[6][4] =
- {
- {NppMirror<CV_8U, nppiMirror_8u_C1R>::call, 0, NppMirror<CV_8U, nppiMirror_8u_C3R>::call, NppMirror<CV_8U, nppiMirror_8u_C4R>::call},
- {0,0,0,0},
- {NppMirror<CV_16U, nppiMirror_16u_C1R>::call, 0, NppMirror<CV_16U, nppiMirror_16u_C3R>::call, NppMirror<CV_16U, nppiMirror_16u_C4R>::call},
- {0,0,0,0},
- {NppMirror<CV_32S, nppiMirror_32s_C1R>::call, 0, NppMirror<CV_32S, nppiMirror_32s_C3R>::call, NppMirror<CV_32S, nppiMirror_32s_C4R>::call},
- {NppMirror<CV_32F, nppiMirror_32f_C1R>::call, 0, NppMirror<CV_32F, nppiMirror_32f_C3R>::call, NppMirror<CV_32F, nppiMirror_32f_C4R>::call}
- };
-
- CV_Assert(src.depth() == CV_8U || src.depth() == CV_16U || src.depth() == CV_32S || src.depth() == CV_32F);
- CV_Assert(src.channels() == 1 || src.channels() == 3 || src.channels() == 4);
-
- dst.create(src.size(), src.type());
-
- funcs[src.depth()][src.channels() - 1](src, dst, flipCode, StreamAccessor::getStream(stream));
-}
-
-////////////////////////////////////////////////////////////////////////
-// LUT
-
-void cv::gpu::LUT(const GpuMat& src, const Mat& lut, GpuMat& dst, Stream& s)
-{
- const int cn = src.channels();
-
- CV_Assert( src.type() == CV_8UC1 || src.type() == CV_8UC3 );
- CV_Assert( lut.depth() == CV_8U );
- CV_Assert( lut.channels() == 1 || lut.channels() == cn );
- CV_Assert( lut.rows * lut.cols == 256 && lut.isContinuous() );
-
- dst.create(src.size(), CV_MAKE_TYPE(lut.depth(), cn));
-
- NppiSize sz;
- sz.height = src.rows;
- sz.width = src.cols;
-
- Mat nppLut;
- lut.convertTo(nppLut, CV_32S);
-
- int nValues3[] = {256, 256, 256};
-
- Npp32s pLevels[256];
- for (int i = 0; i < 256; ++i)
- pLevels[i] = i;
-
- const Npp32s* pLevels3[3];
-
-#if (CUDA_VERSION <= 4020)
- pLevels3[0] = pLevels3[1] = pLevels3[2] = pLevels;
-#else
- GpuMat d_pLevels;
- d_pLevels.upload(Mat(1, 256, CV_32S, pLevels));
- pLevels3[0] = pLevels3[1] = pLevels3[2] = d_pLevels.ptr<Npp32s>();
-#endif
-
- cudaStream_t stream = StreamAccessor::getStream(s);
- NppStreamHandler h(stream);
-
- if (src.type() == CV_8UC1)
- {
-#if (CUDA_VERSION <= 4020)
- nppSafeCall( nppiLUT_Linear_8u_C1R(src.ptr<Npp8u>(), static_cast<int>(src.step),
- dst.ptr<Npp8u>(), static_cast<int>(dst.step), sz, nppLut.ptr<Npp32s>(), pLevels, 256) );
-#else
- GpuMat d_nppLut(Mat(1, 256, CV_32S, nppLut.data));
- nppSafeCall( nppiLUT_Linear_8u_C1R(src.ptr<Npp8u>(), static_cast<int>(src.step),
- dst.ptr<Npp8u>(), static_cast<int>(dst.step), sz, d_nppLut.ptr<Npp32s>(), d_pLevels.ptr<Npp32s>(), 256) );
-#endif
- }
- else
- {
- const Npp32s* pValues3[3];
-
- Mat nppLut3[3];
- if (nppLut.channels() == 1)
- {
-#if (CUDA_VERSION <= 4020)
- pValues3[0] = pValues3[1] = pValues3[2] = nppLut.ptr<Npp32s>();
-#else
- GpuMat d_nppLut(Mat(1, 256, CV_32S, nppLut.data));
- pValues3[0] = pValues3[1] = pValues3[2] = d_nppLut.ptr<Npp32s>();
-#endif
- }
- else
- {
- cv::split(nppLut, nppLut3);
-
-#if (CUDA_VERSION <= 4020)
- pValues3[0] = nppLut3[0].ptr<Npp32s>();
- pValues3[1] = nppLut3[1].ptr<Npp32s>();
- pValues3[2] = nppLut3[2].ptr<Npp32s>();
-#else
- GpuMat d_nppLut0(Mat(1, 256, CV_32S, nppLut3[0].data));
- GpuMat d_nppLut1(Mat(1, 256, CV_32S, nppLut3[1].data));
- GpuMat d_nppLut2(Mat(1, 256, CV_32S, nppLut3[2].data));
-
- pValues3[0] = d_nppLut0.ptr<Npp32s>();
- pValues3[1] = d_nppLut1.ptr<Npp32s>();
- pValues3[2] = d_nppLut2.ptr<Npp32s>();
-#endif
- }
-
- nppSafeCall( nppiLUT_Linear_8u_C3R(src.ptr<Npp8u>(), static_cast<int>(src.step),
- dst.ptr<Npp8u>(), static_cast<int>(dst.step), sz, pValues3, pLevels3, nValues3) );
- }
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
-}
-
-////////////////////////////////////////////////////////////////////////
-// NPP magnitide
-
-namespace
-{
- typedef NppStatus (*nppMagnitude_t)(const Npp32fc* pSrc, int nSrcStep, Npp32f* pDst, int nDstStep, NppiSize oSizeROI);
-
- inline void npp_magnitude(const GpuMat& src, GpuMat& dst, nppMagnitude_t func, cudaStream_t stream)
- {
- CV_Assert(src.type() == CV_32FC2);
-
- dst.create(src.size(), CV_32FC1);
-
- NppiSize sz;
- sz.width = src.cols;
- sz.height = src.rows;
-
- NppStreamHandler h(stream);
-
- nppSafeCall( func(src.ptr<Npp32fc>(), static_cast<int>(src.step), dst.ptr<Npp32f>(), static_cast<int>(dst.step), sz) );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
-}
-
-void cv::gpu::magnitude(const GpuMat& src, GpuMat& dst, Stream& stream)
-{
- npp_magnitude(src, dst, nppiMagnitude_32fc32f_C1R, StreamAccessor::getStream(stream));
-}
-
-void cv::gpu::magnitudeSqr(const GpuMat& src, GpuMat& dst, Stream& stream)
-{
- npp_magnitude(src, dst, nppiMagnitudeSqr_32fc32f_C1R, StreamAccessor::getStream(stream));
-}
-
-////////////////////////////////////////////////////////////////////////
-// Polar <-> Cart
-
-namespace cv { namespace gpu { namespace cudev
-{
- namespace mathfunc
- {
- void cartToPolar_gpu(PtrStepSzf x, PtrStepSzf y, PtrStepSzf mag, bool magSqr, PtrStepSzf angle, bool angleInDegrees, cudaStream_t stream);
- void polarToCart_gpu(PtrStepSzf mag, PtrStepSzf angle, PtrStepSzf x, PtrStepSzf y, bool angleInDegrees, cudaStream_t stream);
- }
-}}}
-
-namespace
-{
- inline void cartToPolar_caller(const GpuMat& x, const GpuMat& y, GpuMat* mag, bool magSqr, GpuMat* angle, bool angleInDegrees, cudaStream_t stream)
- {
- using namespace ::cv::gpu::cudev::mathfunc;
-
- CV_Assert(x.size() == y.size() && x.type() == y.type());
- CV_Assert(x.depth() == CV_32F);
-
- if (mag)
- mag->create(x.size(), x.type());
- if (angle)
- angle->create(x.size(), x.type());
-
- GpuMat x1cn = x.reshape(1);
- GpuMat y1cn = y.reshape(1);
- GpuMat mag1cn = mag ? mag->reshape(1) : GpuMat();
- GpuMat angle1cn = angle ? angle->reshape(1) : GpuMat();
-
- cartToPolar_gpu(x1cn, y1cn, mag1cn, magSqr, angle1cn, angleInDegrees, stream);
- }
-
- inline void polarToCart_caller(const GpuMat& mag, const GpuMat& angle, GpuMat& x, GpuMat& y, bool angleInDegrees, cudaStream_t stream)
- {
- using namespace ::cv::gpu::cudev::mathfunc;
-
- CV_Assert((mag.empty() || mag.size() == angle.size()) && mag.type() == angle.type());
- CV_Assert(mag.depth() == CV_32F);
-
- x.create(mag.size(), mag.type());
- y.create(mag.size(), mag.type());
-
- GpuMat mag1cn = mag.reshape(1);
- GpuMat angle1cn = angle.reshape(1);
- GpuMat x1cn = x.reshape(1);
- GpuMat y1cn = y.reshape(1);
-
- polarToCart_gpu(mag1cn, angle1cn, x1cn, y1cn, angleInDegrees, stream);
- }
-}
-
-void cv::gpu::magnitude(const GpuMat& x, const GpuMat& y, GpuMat& dst, Stream& stream)
-{
- cartToPolar_caller(x, y, &dst, false, 0, false, StreamAccessor::getStream(stream));
-}
-
-void cv::gpu::magnitudeSqr(const GpuMat& x, const GpuMat& y, GpuMat& dst, Stream& stream)
-{
- cartToPolar_caller(x, y, &dst, true, 0, false, StreamAccessor::getStream(stream));
-}
-
-void cv::gpu::phase(const GpuMat& x, const GpuMat& y, GpuMat& angle, bool angleInDegrees, Stream& stream)
-{
- cartToPolar_caller(x, y, 0, false, &angle, angleInDegrees, StreamAccessor::getStream(stream));
-}
-
-void cv::gpu::cartToPolar(const GpuMat& x, const GpuMat& y, GpuMat& mag, GpuMat& angle, bool angleInDegrees, Stream& stream)
-{
- cartToPolar_caller(x, y, &mag, false, &angle, angleInDegrees, StreamAccessor::getStream(stream));
-}
-
-void cv::gpu::polarToCart(const GpuMat& magnitude, const GpuMat& angle, GpuMat& x, GpuMat& y, bool angleInDegrees, Stream& stream)
-{
- polarToCart_caller(magnitude, angle, x, y, angleInDegrees, StreamAccessor::getStream(stream));
-}
-
-////////////////////////////////////////////////////////////////////////
-// normalize
-
-void cv::gpu::normalize(const GpuMat& src, GpuMat& dst, double a, double b, int norm_type, int dtype, const GpuMat& mask)
-{
- GpuMat norm_buf;
- GpuMat cvt_buf;
- normalize(src, dst, a, b, norm_type, dtype, mask, norm_buf, cvt_buf);
-}
-
-void cv::gpu::normalize(const GpuMat& src, GpuMat& dst, double a, double b, int norm_type, int dtype, const GpuMat& mask, GpuMat& norm_buf, GpuMat& cvt_buf)
-{
- double scale = 1, shift = 0;
- if (norm_type == NORM_MINMAX)
- {
- double smin = 0, smax = 0;
- double dmin = std::min(a, b), dmax = std::max(a, b);
- minMax(src, &smin, &smax, mask, norm_buf);
- scale = (dmax - dmin) * (smax - smin > std::numeric_limits<double>::epsilon() ? 1.0 / (smax - smin) : 0.0);
- shift = dmin - smin * scale;
- }
- else if (norm_type == NORM_L2 || norm_type == NORM_L1 || norm_type == NORM_INF)
- {
- scale = norm(src, norm_type, mask, norm_buf);
- scale = scale > std::numeric_limits<double>::epsilon() ? a / scale : 0.0;
- shift = 0;
- }
- else
- {
- CV_Error(cv::Error::StsBadArg, "Unknown/unsupported norm type");
- }
-
- if (mask.empty())
- {
- src.convertTo(dst, dtype, scale, shift);
- }
- else
- {
- src.convertTo(cvt_buf, dtype, scale, shift);
- cvt_buf.copyTo(dst, mask);
- }
-}
-
-#endif /* !defined (HAVE_CUDA) */
// Find the best hypothesis index
Point best_idx;
double best_score;
- minMaxLoc(d_hypothesis_scores, NULL, &best_score, NULL, &best_idx);
+ gpu::minMaxLoc(d_hypothesis_scores, NULL, &best_score, NULL, &best_idx);
int num_inliers = static_cast<int>(best_score);
// Extract the best hypothesis data
//M*/
#include "precomp.hpp"
-#include <vector>
-#include <iostream>
#include "opencv2/objdetect/objdetect_c.h"
using namespace cv;
virtual bool read(const String& classifierAsXml) = 0;
};
+#ifndef HAVE_OPENCV_GPULEGACY
+
+struct cv::gpu::CascadeClassifier_GPU::HaarCascade : cv::gpu::CascadeClassifier_GPU::CascadeClassifierImpl
+{
+public:
+ HaarCascade()
+ {
+ throw_no_cuda();
+ }
+
+ unsigned int process(const GpuMat&, GpuMat&, float, int, bool, bool, cv::Size, cv::Size)
+ {
+ throw_no_cuda();
+ return 0;
+ }
+
+ cv::Size getClassifierCvSize() const
+ {
+ throw_no_cuda();
+ return cv::Size();
+ }
+
+ bool read(const String&)
+ {
+ throw_no_cuda();
+ return false;
+ }
+};
+
+#else
+
struct cv::gpu::CascadeClassifier_GPU::HaarCascade : cv::gpu::CascadeClassifier_GPU::CascadeClassifierImpl
{
public:
virtual ~HaarCascade(){}
};
+#endif
+
cv::Size operator -(const cv::Size& a, const cv::Size& b)
{
return cv::Size(a.width - b.width, a.height - b.height);
resuzeBuffer.create(frame, CV_8UC1);
integral.create(frame.height + 1, integralFactor * (frame.width + 1), CV_32SC1);
+
+#ifdef HAVE_OPENCV_GPULEGACY
NcvSize32u roiSize;
roiSize.width = frame.width;
roiSize.height = frame.height;
Ncv32u bufSize;
ncvSafeCall( nppiStIntegralGetSize_8u32u(roiSize, &bufSize, prop) );
integralBuffer.create(1, bufSize, CV_8UC1);
+#endif
candidates.create(1 , frame.width >> 1, CV_32SC4);
}
}
#endif
-
-//////////////////////////////////////////////////////////////////////////////////////////////////////
-
-#if defined (HAVE_CUDA)
-
-struct RectConvert
-{
- Rect operator()(const NcvRect32u& nr) const { return Rect(nr.x, nr.y, nr.width, nr.height); }
- NcvRect32u operator()(const Rect& nr) const
- {
- NcvRect32u rect;
- rect.x = nr.x;
- rect.y = nr.y;
- rect.width = nr.width;
- rect.height = nr.height;
- return rect;
- }
-};
-
-void groupRectangles(std::vector<NcvRect32u> &hypotheses, int groupThreshold, double eps, std::vector<Ncv32u> *weights)
-{
- std::vector<Rect> rects(hypotheses.size());
- std::transform(hypotheses.begin(), hypotheses.end(), rects.begin(), RectConvert());
-
- if (weights)
- {
- std::vector<int> weights_int;
- weights_int.assign(weights->begin(), weights->end());
- cv::groupRectangles(rects, weights_int, groupThreshold, eps);
- }
- else
- {
- cv::groupRectangles(rects, groupThreshold, eps);
- }
- std::transform(rects.begin(), rects.end(), hypotheses.begin(), RectConvert());
- hypotheses.resize(rects.size());
-}
-
-NCVStatus loadFromXML(const String &filename,
- HaarClassifierCascadeDescriptor &haar,
- std::vector<HaarStage64> &haarStages,
- std::vector<HaarClassifierNode128> &haarClassifierNodes,
- std::vector<HaarFeature64> &haarFeatures)
-{
- NCVStatus ncvStat;
-
- haar.NumStages = 0;
- haar.NumClassifierRootNodes = 0;
- haar.NumClassifierTotalNodes = 0;
- haar.NumFeatures = 0;
- haar.ClassifierSize.width = 0;
- haar.ClassifierSize.height = 0;
- haar.bHasStumpsOnly = true;
- haar.bNeedsTiltedII = false;
- Ncv32u curMaxTreeDepth;
-
- std::vector<char> xmlFileCont;
-
- std::vector<HaarClassifierNode128> h_TmpClassifierNotRootNodes;
- haarStages.resize(0);
- haarClassifierNodes.resize(0);
- haarFeatures.resize(0);
-
- Ptr<CvHaarClassifierCascade> oldCascade = (CvHaarClassifierCascade*)cvLoad(filename.c_str(), 0, 0, 0);
- if (oldCascade.empty())
- {
- return NCV_HAAR_XML_LOADING_EXCEPTION;
- }
-
- haar.ClassifierSize.width = oldCascade->orig_window_size.width;
- haar.ClassifierSize.height = oldCascade->orig_window_size.height;
-
- int stagesCound = oldCascade->count;
- for(int s = 0; s < stagesCound; ++s) // by stages
- {
- HaarStage64 curStage;
- curStage.setStartClassifierRootNodeOffset(static_cast<Ncv32u>(haarClassifierNodes.size()));
-
- curStage.setStageThreshold(oldCascade->stage_classifier[s].threshold);
-
- int treesCount = oldCascade->stage_classifier[s].count;
- for(int t = 0; t < treesCount; ++t) // by trees
- {
- Ncv32u nodeId = 0;
- CvHaarClassifier* tree = &oldCascade->stage_classifier[s].classifier[t];
-
- int nodesCount = tree->count;
- for(int n = 0; n < nodesCount; ++n) //by features
- {
- CvHaarFeature* feature = &tree->haar_feature[n];
-
- HaarClassifierNode128 curNode;
- curNode.setThreshold(tree->threshold[n]);
-
- NcvBool bIsLeftNodeLeaf = false;
- NcvBool bIsRightNodeLeaf = false;
-
- HaarClassifierNodeDescriptor32 nodeLeft;
- if ( tree->left[n] <= 0 )
- {
- Ncv32f leftVal = tree->alpha[-tree->left[n]];
- ncvStat = nodeLeft.create(leftVal);
- ncvAssertReturn(ncvStat == NCV_SUCCESS, ncvStat);
- bIsLeftNodeLeaf = true;
- }
- else
- {
- Ncv32u leftNodeOffset = tree->left[n];
- nodeLeft.create((Ncv32u)(h_TmpClassifierNotRootNodes.size() + leftNodeOffset - 1));
- haar.bHasStumpsOnly = false;
- }
- curNode.setLeftNodeDesc(nodeLeft);
-
- HaarClassifierNodeDescriptor32 nodeRight;
- if ( tree->right[n] <= 0 )
- {
- Ncv32f rightVal = tree->alpha[-tree->right[n]];
- ncvStat = nodeRight.create(rightVal);
- ncvAssertReturn(ncvStat == NCV_SUCCESS, ncvStat);
- bIsRightNodeLeaf = true;
- }
- else
- {
- Ncv32u rightNodeOffset = tree->right[n];
- nodeRight.create((Ncv32u)(h_TmpClassifierNotRootNodes.size() + rightNodeOffset - 1));
- haar.bHasStumpsOnly = false;
- }
- curNode.setRightNodeDesc(nodeRight);
-
- Ncv32u tiltedVal = feature->tilted;
- haar.bNeedsTiltedII = (tiltedVal != 0);
-
- Ncv32u featureId = 0;
- for(int l = 0; l < CV_HAAR_FEATURE_MAX; ++l) //by rects
- {
- Ncv32u rectX = feature->rect[l].r.x;
- Ncv32u rectY = feature->rect[l].r.y;
- Ncv32u rectWidth = feature->rect[l].r.width;
- Ncv32u rectHeight = feature->rect[l].r.height;
-
- Ncv32f rectWeight = feature->rect[l].weight;
-
- if (rectWeight == 0/* && rectX == 0 &&rectY == 0 && rectWidth == 0 && rectHeight == 0*/)
- break;
-
- HaarFeature64 curFeature;
- ncvStat = curFeature.setRect(rectX, rectY, rectWidth, rectHeight, haar.ClassifierSize.width, haar.ClassifierSize.height);
- curFeature.setWeight(rectWeight);
- ncvAssertReturn(NCV_SUCCESS == ncvStat, ncvStat);
- haarFeatures.push_back(curFeature);
-
- featureId++;
- }
-
- HaarFeatureDescriptor32 tmpFeatureDesc;
- ncvStat = tmpFeatureDesc.create(haar.bNeedsTiltedII, bIsLeftNodeLeaf, bIsRightNodeLeaf,
- featureId, static_cast<Ncv32u>(haarFeatures.size()) - featureId);
- ncvAssertReturn(NCV_SUCCESS == ncvStat, ncvStat);
- curNode.setFeatureDesc(tmpFeatureDesc);
-
- if (!nodeId)
- {
- //root node
- haarClassifierNodes.push_back(curNode);
- curMaxTreeDepth = 1;
- }
- else
- {
- //other node
- h_TmpClassifierNotRootNodes.push_back(curNode);
- curMaxTreeDepth++;
- }
-
- nodeId++;
- }
- }
-
- curStage.setNumClassifierRootNodes(treesCount);
- haarStages.push_back(curStage);
- }
-
- //fill in cascade stats
- haar.NumStages = static_cast<Ncv32u>(haarStages.size());
- haar.NumClassifierRootNodes = static_cast<Ncv32u>(haarClassifierNodes.size());
- haar.NumClassifierTotalNodes = static_cast<Ncv32u>(haar.NumClassifierRootNodes + h_TmpClassifierNotRootNodes.size());
- haar.NumFeatures = static_cast<Ncv32u>(haarFeatures.size());
-
- //merge root and leaf nodes in one classifiers array
- Ncv32u offsetRoot = static_cast<Ncv32u>(haarClassifierNodes.size());
- for (Ncv32u i=0; i<haarClassifierNodes.size(); i++)
- {
- HaarFeatureDescriptor32 featureDesc = haarClassifierNodes[i].getFeatureDesc();
-
- HaarClassifierNodeDescriptor32 nodeLeft = haarClassifierNodes[i].getLeftNodeDesc();
- if (!featureDesc.isLeftNodeLeaf())
- {
- Ncv32u newOffset = nodeLeft.getNextNodeOffset() + offsetRoot;
- nodeLeft.create(newOffset);
- }
- haarClassifierNodes[i].setLeftNodeDesc(nodeLeft);
-
- HaarClassifierNodeDescriptor32 nodeRight = haarClassifierNodes[i].getRightNodeDesc();
- if (!featureDesc.isRightNodeLeaf())
- {
- Ncv32u newOffset = nodeRight.getNextNodeOffset() + offsetRoot;
- nodeRight.create(newOffset);
- }
- haarClassifierNodes[i].setRightNodeDesc(nodeRight);
- }
-
- for (Ncv32u i=0; i<h_TmpClassifierNotRootNodes.size(); i++)
- {
- HaarFeatureDescriptor32 featureDesc = h_TmpClassifierNotRootNodes[i].getFeatureDesc();
-
- HaarClassifierNodeDescriptor32 nodeLeft = h_TmpClassifierNotRootNodes[i].getLeftNodeDesc();
- if (!featureDesc.isLeftNodeLeaf())
- {
- Ncv32u newOffset = nodeLeft.getNextNodeOffset() + offsetRoot;
- nodeLeft.create(newOffset);
- }
- h_TmpClassifierNotRootNodes[i].setLeftNodeDesc(nodeLeft);
-
- HaarClassifierNodeDescriptor32 nodeRight = h_TmpClassifierNotRootNodes[i].getRightNodeDesc();
- if (!featureDesc.isRightNodeLeaf())
- {
- Ncv32u newOffset = nodeRight.getNextNodeOffset() + offsetRoot;
- nodeRight.create(newOffset);
- }
- h_TmpClassifierNotRootNodes[i].setRightNodeDesc(nodeRight);
-
- haarClassifierNodes.push_back(h_TmpClassifierNotRootNodes[i]);
- }
-
- return NCV_SUCCESS;
-}
-
-#endif /* HAVE_CUDA */
+++ /dev/null
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-// By downloading, copying, installing or using the software you agree to this license.
-// If you do not agree to this license, do not download, install,
-// copy or use the software.
-//
-//
-// License Agreement
-// For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-// * Redistribution's of source code must retain the above copyright notice,
-// this list of conditions and the following disclaimer.
-//
-// * Redistribution's in binary form must reproduce the above copyright notice,
-// this list of conditions and the following disclaimer in the documentation
-// and/or other materials provided with the distribution.
-//
-// * The name of the copyright holders may not be used to endorse or promote products
-// derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-/*
- * NV12ToARGB color space conversion CUDA kernel
- *
- * This sample uses CUDA to perform a simple NV12 (YUV 4:2:0 planar)
- * source and converts to output in ARGB format
- */
-
-#if !defined CUDA_DISABLER
-
-#include "opencv2/core/cuda/common.hpp"
-
-namespace cv { namespace gpu { namespace cudev {
- namespace video_decoding
- {
- __constant__ uint constAlpha = ((uint)0xff << 24);
-
- __constant__ float constHueColorSpaceMat[9];
-
- void loadHueCSC(float hueCSC[9])
- {
- cudaSafeCall( cudaMemcpyToSymbol(constHueColorSpaceMat, hueCSC, 9 * sizeof(float)) );
- }
-
- __device__ void YUV2RGB(const uint* yuvi, float* red, float* green, float* blue)
- {
- float luma, chromaCb, chromaCr;
-
- // Prepare for hue adjustment
- luma = (float)yuvi[0];
- chromaCb = (float)((int)yuvi[1] - 512.0f);
- chromaCr = (float)((int)yuvi[2] - 512.0f);
-
- // Convert YUV To RGB with hue adjustment
- *red = (luma * constHueColorSpaceMat[0]) +
- (chromaCb * constHueColorSpaceMat[1]) +
- (chromaCr * constHueColorSpaceMat[2]);
-
- *green = (luma * constHueColorSpaceMat[3]) +
- (chromaCb * constHueColorSpaceMat[4]) +
- (chromaCr * constHueColorSpaceMat[5]);
-
- *blue = (luma * constHueColorSpaceMat[6]) +
- (chromaCb * constHueColorSpaceMat[7]) +
- (chromaCr * constHueColorSpaceMat[8]);
- }
-
- __device__ uint RGBAPACK_10bit(float red, float green, float blue, uint alpha)
- {
- uint ARGBpixel = 0;
-
- // Clamp final 10 bit results
- red = ::fmin(::fmax(red, 0.0f), 1023.f);
- green = ::fmin(::fmax(green, 0.0f), 1023.f);
- blue = ::fmin(::fmax(blue, 0.0f), 1023.f);
-
- // Convert to 8 bit unsigned integers per color component
- ARGBpixel = (((uint)blue >> 2) |
- (((uint)green >> 2) << 8) |
- (((uint)red >> 2) << 16) |
- (uint)alpha);
-
- return ARGBpixel;
- }
-
- // CUDA kernel for outputing the final ARGB output from NV12
-
- #define COLOR_COMPONENT_BIT_SIZE 10
- #define COLOR_COMPONENT_MASK 0x3FF
-
- __global__ void NV12ToARGB(uchar* srcImage, size_t nSourcePitch,
- uint* dstImage, size_t nDestPitch,
- uint width, uint height)
- {
- // Pad borders with duplicate pixels, and we multiply by 2 because we process 2 pixels per thread
- const int x = blockIdx.x * (blockDim.x << 1) + (threadIdx.x << 1);
- const int y = blockIdx.y * blockDim.y + threadIdx.y;
-
- if (x >= width || y >= height)
- return;
-
- // Read 2 Luma components at a time, so we don't waste processing since CbCr are decimated this way.
- // if we move to texture we could read 4 luminance values
-
- uint yuv101010Pel[2];
-
- yuv101010Pel[0] = (srcImage[y * nSourcePitch + x ]) << 2;
- yuv101010Pel[1] = (srcImage[y * nSourcePitch + x + 1]) << 2;
-
- const size_t chromaOffset = nSourcePitch * height;
-
- const int y_chroma = y >> 1;
-
- if (y & 1) // odd scanline ?
- {
- uint chromaCb = srcImage[chromaOffset + y_chroma * nSourcePitch + x ];
- uint chromaCr = srcImage[chromaOffset + y_chroma * nSourcePitch + x + 1];
-
- if (y_chroma < ((height >> 1) - 1)) // interpolate chroma vertically
- {
- chromaCb = (chromaCb + srcImage[chromaOffset + (y_chroma + 1) * nSourcePitch + x ] + 1) >> 1;
- chromaCr = (chromaCr + srcImage[chromaOffset + (y_chroma + 1) * nSourcePitch + x + 1] + 1) >> 1;
- }
-
- yuv101010Pel[0] |= (chromaCb << ( COLOR_COMPONENT_BIT_SIZE + 2));
- yuv101010Pel[0] |= (chromaCr << ((COLOR_COMPONENT_BIT_SIZE << 1) + 2));
-
- yuv101010Pel[1] |= (chromaCb << ( COLOR_COMPONENT_BIT_SIZE + 2));
- yuv101010Pel[1] |= (chromaCr << ((COLOR_COMPONENT_BIT_SIZE << 1) + 2));
- }
- else
- {
- yuv101010Pel[0] |= ((uint)srcImage[chromaOffset + y_chroma * nSourcePitch + x ] << ( COLOR_COMPONENT_BIT_SIZE + 2));
- yuv101010Pel[0] |= ((uint)srcImage[chromaOffset + y_chroma * nSourcePitch + x + 1] << ((COLOR_COMPONENT_BIT_SIZE << 1) + 2));
-
- yuv101010Pel[1] |= ((uint)srcImage[chromaOffset + y_chroma * nSourcePitch + x ] << ( COLOR_COMPONENT_BIT_SIZE + 2));
- yuv101010Pel[1] |= ((uint)srcImage[chromaOffset + y_chroma * nSourcePitch + x + 1] << ((COLOR_COMPONENT_BIT_SIZE << 1) + 2));
- }
-
- // this steps performs the color conversion
- uint yuvi[6];
- float red[2], green[2], blue[2];
-
- yuvi[0] = (yuv101010Pel[0] & COLOR_COMPONENT_MASK );
- yuvi[1] = ((yuv101010Pel[0] >> COLOR_COMPONENT_BIT_SIZE) & COLOR_COMPONENT_MASK);
- yuvi[2] = ((yuv101010Pel[0] >> (COLOR_COMPONENT_BIT_SIZE << 1)) & COLOR_COMPONENT_MASK);
-
- yuvi[3] = (yuv101010Pel[1] & COLOR_COMPONENT_MASK );
- yuvi[4] = ((yuv101010Pel[1] >> COLOR_COMPONENT_BIT_SIZE) & COLOR_COMPONENT_MASK);
- yuvi[5] = ((yuv101010Pel[1] >> (COLOR_COMPONENT_BIT_SIZE << 1)) & COLOR_COMPONENT_MASK);
-
- // YUV to RGB Transformation conversion
- YUV2RGB(&yuvi[0], &red[0], &green[0], &blue[0]);
- YUV2RGB(&yuvi[3], &red[1], &green[1], &blue[1]);
-
- // Clamp the results to RGBA
-
- const size_t dstImagePitch = nDestPitch >> 2;
-
- dstImage[y * dstImagePitch + x ] = RGBAPACK_10bit(red[0], green[0], blue[0], constAlpha);
- dstImage[y * dstImagePitch + x + 1 ] = RGBAPACK_10bit(red[1], green[1], blue[1], constAlpha);
- }
-
- void NV12ToARGB_gpu(const PtrStepb decodedFrame, PtrStepSz<uint> interopFrame, cudaStream_t stream)
- {
- dim3 block(32, 8);
- dim3 grid(divUp(interopFrame.cols, 2 * block.x), divUp(interopFrame.rows, block.y));
-
- NV12ToARGB<<<grid, block, 0, stream>>>(decodedFrame.data, decodedFrame.step, interopFrame.data, interopFrame.step,
- interopFrame.cols, interopFrame.rows);
-
- cudaSafeCall( cudaGetLastError() );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
- }
-}}}
-
-#endif /* CUDA_DISABLER */
+++ /dev/null
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-// By downloading, copying, installing or using the software you agree to this license.
-// If you do not agree to this license, do not download, install,
-// copy or use the software.
-//
-//
-// License Agreement
-// For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-// * Redistribution's of source code must retain the above copyright notice,
-// this list of conditions and the following disclaimer.
-//
-// * Redistribution's in binary form must reproduce the above copyright notice,
-// this list of conditions and the following disclaimer in the documentation
-// and/or other materials provided with the distribution.
-//
-// * The name of the copyright holders may not be used to endorse or promote products
-// derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#if !defined CUDA_DISABLER
-
-#include "opencv2/core/cuda/common.hpp"
-#include "opencv2/core/cuda/functional.hpp"
-#include "opencv2/core/cuda/vec_math.hpp"
-#include "opencv2/core/cuda/transform.hpp"
-#include "opencv2/core/cuda/limits.hpp"
-#include "opencv2/core/cuda/saturate_cast.hpp"
-#include "opencv2/core/cuda/simd_functions.hpp"
-
-using namespace cv::gpu;
-using namespace cv::gpu::cudev;
-
-namespace arithm
-{
- template <size_t src_size, size_t dst_size> struct ArithmFuncTraits
- {
- enum { simple_block_dim_x = 32 };
- enum { simple_block_dim_y = 8 };
-
- enum { smart_block_dim_x = 32 };
- enum { smart_block_dim_y = 8 };
- enum { smart_shift = 1 };
- };
-
- template <> struct ArithmFuncTraits<1, 1>
- {
- enum { simple_block_dim_x = 32 };
- enum { simple_block_dim_y = 8 };
-
- enum { smart_block_dim_x = 32 };
- enum { smart_block_dim_y = 8 };
- enum { smart_shift = 4 };
- };
- template <> struct ArithmFuncTraits<1, 2>
- {
- enum { simple_block_dim_x = 32 };
- enum { simple_block_dim_y = 8 };
-
- enum { smart_block_dim_x = 32 };
- enum { smart_block_dim_y = 8 };
- enum { smart_shift = 4 };
- };
- template <> struct ArithmFuncTraits<1, 4>
- {
- enum { simple_block_dim_x = 32 };
- enum { simple_block_dim_y = 8 };
-
- enum { smart_block_dim_x = 32 };
- enum { smart_block_dim_y = 8 };
- enum { smart_shift = 4 };
- };
-
- template <> struct ArithmFuncTraits<2, 1>
- {
- enum { simple_block_dim_x = 32 };
- enum { simple_block_dim_y = 8 };
-
- enum { smart_block_dim_x = 32 };
- enum { smart_block_dim_y = 8 };
- enum { smart_shift = 4 };
- };
- template <> struct ArithmFuncTraits<2, 2>
- {
- enum { simple_block_dim_x = 32 };
- enum { simple_block_dim_y = 8 };
-
- enum { smart_block_dim_x = 32 };
- enum { smart_block_dim_y = 8 };
- enum { smart_shift = 4 };
- };
- template <> struct ArithmFuncTraits<2, 4>
- {
- enum { simple_block_dim_x = 32 };
- enum { simple_block_dim_y = 8 };
-
- enum { smart_block_dim_x = 32 };
- enum { smart_block_dim_y = 8 };
- enum { smart_shift = 4 };
- };
-
- template <> struct ArithmFuncTraits<4, 1>
- {
- enum { simple_block_dim_x = 32 };
- enum { simple_block_dim_y = 8 };
-
- enum { smart_block_dim_x = 32 };
- enum { smart_block_dim_y = 8 };
- enum { smart_shift = 4 };
- };
- template <> struct ArithmFuncTraits<4, 2>
- {
- enum { simple_block_dim_x = 32 };
- enum { simple_block_dim_y = 8 };
-
- enum { smart_block_dim_x = 32 };
- enum { smart_block_dim_y = 8 };
- enum { smart_shift = 4 };
- };
- template <> struct ArithmFuncTraits<4, 4>
- {
- enum { simple_block_dim_x = 32 };
- enum { simple_block_dim_y = 8 };
-
- enum { smart_block_dim_x = 32 };
- enum { smart_block_dim_y = 8 };
- enum { smart_shift = 4 };
- };
-}
-
-//////////////////////////////////////////////////////////////////////////
-// addMat
-
-namespace arithm
-{
- struct VAdd4 : binary_function<uint, uint, uint>
- {
- __device__ __forceinline__ uint operator ()(uint a, uint b) const
- {
- return vadd4(a, b);
- }
-
- __host__ __device__ __forceinline__ VAdd4() {}
- __host__ __device__ __forceinline__ VAdd4(const VAdd4&) {}
- };
-
- ////////////////////////////////////
-
- struct VAdd2 : binary_function<uint, uint, uint>
- {
- __device__ __forceinline__ uint operator ()(uint a, uint b) const
- {
- return vadd2(a, b);
- }
-
- __host__ __device__ __forceinline__ VAdd2() {}
- __host__ __device__ __forceinline__ VAdd2(const VAdd2&) {}
- };
-
- ////////////////////////////////////
-
- template <typename T, typename D> struct AddMat : binary_function<T, T, D>
- {
- __device__ __forceinline__ D operator ()(T a, T b) const
- {
- return saturate_cast<D>(a + b);
- }
-
- __host__ __device__ __forceinline__ AddMat() {}
- __host__ __device__ __forceinline__ AddMat(const AddMat&) {}
- };
-}
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <> struct TransformFunctorTraits< arithm::VAdd4 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
- {
- };
-
- ////////////////////////////////////
-
- template <> struct TransformFunctorTraits< arithm::VAdd2 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
- {
- };
-
- ////////////////////////////////////
-
- template <typename T, typename D> struct TransformFunctorTraits< arithm::AddMat<T, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
- {
- };
-}}}
-
-namespace arithm
-{
- void addMat_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
- {
- cudev::transform(src1, src2, dst, VAdd4(), WithOutMask(), stream);
- }
-
- void addMat_v2(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
- {
- cudev::transform(src1, src2, dst, VAdd2(), WithOutMask(), stream);
- }
-
- template <typename T, typename D>
- void addMat(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream)
- {
- if (mask.data)
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<D>) dst, AddMat<T, D>(), mask, stream);
- else
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<D>) dst, AddMat<T, D>(), WithOutMask(), stream);
- }
-
- template void addMat<uchar, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<uchar, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<uchar, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<uchar, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<uchar, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<uchar, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<uchar, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- template void addMat<schar, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<schar, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<schar, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<schar, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<schar, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<schar, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<schar, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void addMat<ushort, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addMat<ushort, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<ushort, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<ushort, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<ushort, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<ushort, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<ushort, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void addMat<short, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addMat<short, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<short, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<short, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<short, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<short, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<short, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void addMat<int, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addMat<int, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addMat<int, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addMat<int, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<int, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<int, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<int, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void addMat<float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addMat<float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addMat<float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addMat<float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addMat<float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<float, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void addMat<double, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addMat<double, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addMat<double, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addMat<double, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addMat<double, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addMat<double, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addMat<double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// addScalar
-
-namespace arithm
-{
- template <typename T, typename S, typename D> struct AddScalar : unary_function<T, D>
- {
- S val;
-
- explicit AddScalar(S val_) : val(val_) {}
-
- __device__ __forceinline__ D operator ()(T a) const
- {
- return saturate_cast<D>(a + val);
- }
- };
-}
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <typename T, typename S, typename D> struct TransformFunctorTraits< arithm::AddScalar<T, S, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
- {
- };
-}}}
-
-namespace arithm
-{
- template <typename T, typename S, typename D>
- void addScalar(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream)
- {
- AddScalar<T, S, D> op(static_cast<S>(val));
-
- if (mask.data)
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<D>) dst, op, mask, stream);
- else
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
- }
-
- template void addScalar<uchar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<uchar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<uchar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<uchar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<uchar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<uchar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<uchar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- template void addScalar<schar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<schar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<schar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<schar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<schar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<schar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<schar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void addScalar<ushort, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addScalar<ushort, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<ushort, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<ushort, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<ushort, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<ushort, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<ushort, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void addScalar<short, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addScalar<short, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<short, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<short, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<short, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<short, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<short, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void addScalar<int, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addScalar<int, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addScalar<int, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addScalar<int, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<int, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<int, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<int, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void addScalar<float, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addScalar<float, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addScalar<float, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addScalar<float, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addScalar<float, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<float, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<float, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void addScalar<double, double, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addScalar<double, double, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addScalar<double, double, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addScalar<double, double, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addScalar<double, double, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void addScalar<double, double, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void addScalar<double, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// subMat
-
-namespace arithm
-{
- struct VSub4 : binary_function<uint, uint, uint>
- {
- __device__ __forceinline__ uint operator ()(uint a, uint b) const
- {
- return vsub4(a, b);
- }
-
- __host__ __device__ __forceinline__ VSub4() {}
- __host__ __device__ __forceinline__ VSub4(const VSub4&) {}
- };
-
- ////////////////////////////////////
-
- struct VSub2 : binary_function<uint, uint, uint>
- {
- __device__ __forceinline__ uint operator ()(uint a, uint b) const
- {
- return vsub2(a, b);
- }
-
- __host__ __device__ __forceinline__ VSub2() {}
- __host__ __device__ __forceinline__ VSub2(const VSub2&) {}
- };
-
- ////////////////////////////////////
-
- template <typename T, typename D> struct SubMat : binary_function<T, T, D>
- {
- __device__ __forceinline__ D operator ()(T a, T b) const
- {
- return saturate_cast<D>(a - b);
- }
-
- __host__ __device__ __forceinline__ SubMat() {}
- __host__ __device__ __forceinline__ SubMat(const SubMat&) {}
- };
-}
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <> struct TransformFunctorTraits< arithm::VSub4 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
- {
- };
-
- ////////////////////////////////////
-
- template <> struct TransformFunctorTraits< arithm::VSub2 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
- {
- };
-
- ////////////////////////////////////
-
- template <typename T, typename D> struct TransformFunctorTraits< arithm::SubMat<T, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
- {
- };
-}}}
-
-namespace arithm
-{
- void subMat_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
- {
- cudev::transform(src1, src2, dst, VSub4(), WithOutMask(), stream);
- }
-
- void subMat_v2(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
- {
- cudev::transform(src1, src2, dst, VSub2(), WithOutMask(), stream);
- }
-
- template <typename T, typename D>
- void subMat(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream)
- {
- if (mask.data)
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<D>) dst, SubMat<T, D>(), mask, stream);
- else
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<D>) dst, SubMat<T, D>(), WithOutMask(), stream);
- }
-
- template void subMat<uchar, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<uchar, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<uchar, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<uchar, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<uchar, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<uchar, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<uchar, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- template void subMat<schar, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<schar, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<schar, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<schar, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<schar, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<schar, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<schar, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void subMat<ushort, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subMat<ushort, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<ushort, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<ushort, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<ushort, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<ushort, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<ushort, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void subMat<short, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subMat<short, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<short, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<short, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<short, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<short, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<short, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void subMat<int, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subMat<int, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subMat<int, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subMat<int, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<int, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<int, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<int, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void subMat<float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subMat<float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subMat<float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subMat<float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subMat<float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<float, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void subMat<double, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subMat<double, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subMat<double, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subMat<double, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subMat<double, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subMat<double, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subMat<double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// subScalar
-
-namespace arithm
-{
- template <typename T, typename S, typename D>
- void subScalar(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream)
- {
- AddScalar<T, S, D> op(-static_cast<S>(val));
-
- if (mask.data)
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<D>) dst, op, mask, stream);
- else
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
- }
-
- template void subScalar<uchar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<uchar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<uchar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<uchar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<uchar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<uchar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<uchar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- template void subScalar<schar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<schar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<schar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<schar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<schar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<schar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<schar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void subScalar<ushort, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subScalar<ushort, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<ushort, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<ushort, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<ushort, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<ushort, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<ushort, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void subScalar<short, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subScalar<short, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<short, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<short, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<short, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<short, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<short, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void subScalar<int, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subScalar<int, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subScalar<int, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subScalar<int, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<int, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<int, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<int, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void subScalar<float, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subScalar<float, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subScalar<float, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subScalar<float, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subScalar<float, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<float, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<float, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- //template void subScalar<double, double, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subScalar<double, double, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subScalar<double, double, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subScalar<double, double, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subScalar<double, double, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- //template void subScalar<double, double, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void subScalar<double, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// mulMat
-
-namespace arithm
-{
- struct Mul_8uc4_32f : binary_function<uint, float, uint>
- {
- __device__ __forceinline__ uint operator ()(uint a, float b) const
- {
- uint res = 0;
-
- res |= (saturate_cast<uchar>((0xffu & (a )) * b) );
- res |= (saturate_cast<uchar>((0xffu & (a >> 8)) * b) << 8);
- res |= (saturate_cast<uchar>((0xffu & (a >> 16)) * b) << 16);
- res |= (saturate_cast<uchar>((0xffu & (a >> 24)) * b) << 24);
-
- return res;
- }
-
- __host__ __device__ __forceinline__ Mul_8uc4_32f() {}
- __host__ __device__ __forceinline__ Mul_8uc4_32f(const Mul_8uc4_32f&) {}
- };
-
- struct Mul_16sc4_32f : binary_function<short4, float, short4>
- {
- __device__ __forceinline__ short4 operator ()(short4 a, float b) const
- {
- return make_short4(saturate_cast<short>(a.x * b), saturate_cast<short>(a.y * b),
- saturate_cast<short>(a.z * b), saturate_cast<short>(a.w * b));
- }
-
- __host__ __device__ __forceinline__ Mul_16sc4_32f() {}
- __host__ __device__ __forceinline__ Mul_16sc4_32f(const Mul_16sc4_32f&) {}
- };
-
- template <typename T, typename D> struct Mul : binary_function<T, T, D>
- {
- __device__ __forceinline__ D operator ()(T a, T b) const
- {
- return saturate_cast<D>(a * b);
- }
-
- __host__ __device__ __forceinline__ Mul() {}
- __host__ __device__ __forceinline__ Mul(const Mul&) {}
- };
-
- template <typename T, typename S, typename D> struct MulScale : binary_function<T, T, D>
- {
- S scale;
-
- explicit MulScale(S scale_) : scale(scale_) {}
-
- __device__ __forceinline__ D operator ()(T a, T b) const
- {
- return saturate_cast<D>(scale * a * b);
- }
- };
-}
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <> struct TransformFunctorTraits<arithm::Mul_8uc4_32f> : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
- {
- };
-
- template <typename T, typename D> struct TransformFunctorTraits< arithm::Mul<T, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
- {
- };
-
- template <typename T, typename S, typename D> struct TransformFunctorTraits< arithm::MulScale<T, S, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
- {
- };
-}}}
-
-namespace arithm
-{
- void mulMat_8uc4_32f(PtrStepSz<uint> src1, PtrStepSzf src2, PtrStepSz<uint> dst, cudaStream_t stream)
- {
- cudev::transform(src1, src2, dst, Mul_8uc4_32f(), WithOutMask(), stream);
- }
-
- void mulMat_16sc4_32f(PtrStepSz<short4> src1, PtrStepSzf src2, PtrStepSz<short4> dst, cudaStream_t stream)
- {
- cudev::transform(src1, src2, dst, Mul_16sc4_32f(), WithOutMask(), stream);
- }
-
- template <typename T, typename S, typename D>
- void mulMat(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream)
- {
- if (scale == 1)
- {
- Mul<T, D> op;
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
- }
- else
- {
- MulScale<T, S, D> op(static_cast<S>(scale));
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
- }
- }
-
- template void mulMat<uchar, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<uchar, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<uchar, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<uchar, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<uchar, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<uchar, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<uchar, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
-
- template void mulMat<schar, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<schar, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<schar, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<schar, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<schar, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<schar, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<schar, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
-
- //template void mulMat<ushort, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void mulMat<ushort, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<ushort, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<ushort, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<ushort, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<ushort, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<ushort, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
-
- //template void mulMat<short, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void mulMat<short, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<short, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<short, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<short, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<short, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<short, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
-
- //template void mulMat<int, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void mulMat<int, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void mulMat<int, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void mulMat<int, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<int, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<int, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<int, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
-
- //template void mulMat<float, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void mulMat<float, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void mulMat<float, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void mulMat<float, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void mulMat<float, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<float, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<float, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
-
- //template void mulMat<double, double, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void mulMat<double, double, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void mulMat<double, double, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void mulMat<double, double, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void mulMat<double, double, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void mulMat<double, double, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void mulMat<double, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// mulScalar
-
-namespace arithm
-{
- template <typename T, typename S, typename D> struct MulScalar : unary_function<T, D>
- {
- S val;
-
- explicit MulScalar(S val_) : val(val_) {}
-
- __device__ __forceinline__ D operator ()(T a) const
- {
- return saturate_cast<D>(a * val);
- }
- };
-}
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <typename T, typename S, typename D> struct TransformFunctorTraits< arithm::MulScalar<T, S, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
- {
- };
-}}}
-
-namespace arithm
-{
- template <typename T, typename S, typename D>
- void mulScalar(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream)
- {
- MulScalar<T, S, D> op(static_cast<S>(val));
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
- }
-
- template void mulScalar<uchar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<uchar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<uchar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<uchar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<uchar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<uchar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<uchar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-
- template void mulScalar<schar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<schar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<schar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<schar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<schar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<schar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<schar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-
- //template void mulScalar<ushort, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void mulScalar<ushort, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<ushort, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<ushort, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<ushort, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<ushort, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<ushort, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-
- //template void mulScalar<short, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void mulScalar<short, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<short, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<short, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<short, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<short, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<short, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-
- //template void mulScalar<int, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void mulScalar<int, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void mulScalar<int, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void mulScalar<int, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<int, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<int, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<int, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-
- //template void mulScalar<float, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void mulScalar<float, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void mulScalar<float, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void mulScalar<float, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void mulScalar<float, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<float, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<float, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-
- //template void mulScalar<double, double, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void mulScalar<double, double, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void mulScalar<double, double, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void mulScalar<double, double, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void mulScalar<double, double, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void mulScalar<double, double, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void mulScalar<double, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// divMat
-
-namespace arithm
-{
- struct Div_8uc4_32f : binary_function<uint, float, uint>
- {
- __device__ __forceinline__ uint operator ()(uint a, float b) const
- {
- uint res = 0;
-
- if (b != 0)
- {
- b = 1.0f / b;
- res |= (saturate_cast<uchar>((0xffu & (a )) * b) );
- res |= (saturate_cast<uchar>((0xffu & (a >> 8)) * b) << 8);
- res |= (saturate_cast<uchar>((0xffu & (a >> 16)) * b) << 16);
- res |= (saturate_cast<uchar>((0xffu & (a >> 24)) * b) << 24);
- }
-
- return res;
- }
- };
-
- struct Div_16sc4_32f : binary_function<short4, float, short4>
- {
- __device__ __forceinline__ short4 operator ()(short4 a, float b) const
- {
- return b != 0 ? make_short4(saturate_cast<short>(a.x / b), saturate_cast<short>(a.y / b),
- saturate_cast<short>(a.z / b), saturate_cast<short>(a.w / b))
- : make_short4(0,0,0,0);
- }
- };
-
- template <typename T, typename D> struct Div : binary_function<T, T, D>
- {
- __device__ __forceinline__ D operator ()(T a, T b) const
- {
- return b != 0 ? saturate_cast<D>(a / b) : 0;
- }
-
- __host__ __device__ __forceinline__ Div() {}
- __host__ __device__ __forceinline__ Div(const Div&) {}
- };
- template <typename T> struct Div<T, float> : binary_function<T, T, float>
- {
- __device__ __forceinline__ float operator ()(T a, T b) const
- {
- return b != 0 ? static_cast<float>(a) / b : 0;
- }
-
- __host__ __device__ __forceinline__ Div() {}
- __host__ __device__ __forceinline__ Div(const Div&) {}
- };
- template <typename T> struct Div<T, double> : binary_function<T, T, double>
- {
- __device__ __forceinline__ double operator ()(T a, T b) const
- {
- return b != 0 ? static_cast<double>(a) / b : 0;
- }
-
- __host__ __device__ __forceinline__ Div() {}
- __host__ __device__ __forceinline__ Div(const Div&) {}
- };
-
- template <typename T, typename S, typename D> struct DivScale : binary_function<T, T, D>
- {
- S scale;
-
- explicit DivScale(S scale_) : scale(scale_) {}
-
- __device__ __forceinline__ D operator ()(T a, T b) const
- {
- return b != 0 ? saturate_cast<D>(scale * a / b) : 0;
- }
- };
-}
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <> struct TransformFunctorTraits<arithm::Div_8uc4_32f> : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
- {
- };
-
- template <typename T, typename D> struct TransformFunctorTraits< arithm::Div<T, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
- {
- };
-
- template <typename T, typename S, typename D> struct TransformFunctorTraits< arithm::DivScale<T, S, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
- {
- };
-}}}
-
-namespace arithm
-{
- void divMat_8uc4_32f(PtrStepSz<uint> src1, PtrStepSzf src2, PtrStepSz<uint> dst, cudaStream_t stream)
- {
- cudev::transform(src1, src2, dst, Div_8uc4_32f(), WithOutMask(), stream);
- }
-
- void divMat_16sc4_32f(PtrStepSz<short4> src1, PtrStepSzf src2, PtrStepSz<short4> dst, cudaStream_t stream)
- {
- cudev::transform(src1, src2, dst, Div_16sc4_32f(), WithOutMask(), stream);
- }
-
- template <typename T, typename S, typename D>
- void divMat(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream)
- {
- if (scale == 1)
- {
- Div<T, D> op;
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
- }
- else
- {
- DivScale<T, S, D> op(static_cast<S>(scale));
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
- }
- }
-
- template void divMat<uchar, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<uchar, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<uchar, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<uchar, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<uchar, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<uchar, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<uchar, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
-
- template void divMat<schar, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<schar, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<schar, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<schar, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<schar, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<schar, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<schar, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
-
- //template void divMat<ushort, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void divMat<ushort, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<ushort, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<ushort, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<ushort, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<ushort, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<ushort, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
-
- //template void divMat<short, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void divMat<short, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<short, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<short, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<short, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<short, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<short, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
-
- //template void divMat<int, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void divMat<int, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void divMat<int, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void divMat<int, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<int, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<int, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<int, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
-
- //template void divMat<float, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void divMat<float, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void divMat<float, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void divMat<float, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void divMat<float, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<float, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<float, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
-
- //template void divMat<double, double, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void divMat<double, double, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void divMat<double, double, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void divMat<double, double, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void divMat<double, double, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- //template void divMat<double, double, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
- template void divMat<double, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// divScalar
-
-namespace arithm
-{
- template <typename T, typename S, typename D>
- void divScalar(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream)
- {
- MulScalar<T, S, D> op(static_cast<S>(1.0 / val));
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
- }
-
- template void divScalar<uchar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<uchar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<uchar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<uchar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<uchar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<uchar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<uchar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-
- template void divScalar<schar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<schar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<schar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<schar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<schar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<schar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<schar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-
- //template void divScalar<ushort, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divScalar<ushort, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<ushort, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<ushort, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<ushort, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<ushort, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<ushort, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-
- //template void divScalar<short, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divScalar<short, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<short, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<short, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<short, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<short, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<short, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-
- //template void divScalar<int, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divScalar<int, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divScalar<int, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divScalar<int, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<int, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<int, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<int, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-
- //template void divScalar<float, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divScalar<float, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divScalar<float, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divScalar<float, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divScalar<float, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<float, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<float, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-
- //template void divScalar<double, double, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divScalar<double, double, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divScalar<double, double, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divScalar<double, double, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divScalar<double, double, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divScalar<double, double, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divScalar<double, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// divInv
-
-namespace arithm
-{
- template <typename T, typename S, typename D> struct DivInv : unary_function<T, D>
- {
- S val;
-
- explicit DivInv(S val_) : val(val_) {}
-
- __device__ __forceinline__ D operator ()(T a) const
- {
- return a != 0 ? saturate_cast<D>(val / a) : 0;
- }
- };
-}
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <typename T, typename S, typename D> struct TransformFunctorTraits< arithm::DivInv<T, S, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
- {
- };
-}}}
-
-namespace arithm
-{
- template <typename T, typename S, typename D>
- void divInv(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream)
- {
- DivInv<T, S, D> op(static_cast<S>(val));
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
- }
-
- template void divInv<uchar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<uchar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<uchar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<uchar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<uchar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<uchar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<uchar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-
- template void divInv<schar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<schar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<schar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<schar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<schar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<schar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<schar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-
- //template void divInv<ushort, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divInv<ushort, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<ushort, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<ushort, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<ushort, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<ushort, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<ushort, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-
- //template void divInv<short, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divInv<short, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<short, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<short, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<short, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<short, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<short, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-
- //template void divInv<int, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divInv<int, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divInv<int, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divInv<int, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<int, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<int, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<int, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-
- //template void divInv<float, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divInv<float, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divInv<float, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divInv<float, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divInv<float, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<float, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<float, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-
- //template void divInv<double, double, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divInv<double, double, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divInv<double, double, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divInv<double, double, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divInv<double, double, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- //template void divInv<double, double, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
- template void divInv<double, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// absDiffMat
-
-namespace arithm
-{
- struct VAbsDiff4 : binary_function<uint, uint, uint>
- {
- __device__ __forceinline__ uint operator ()(uint a, uint b) const
- {
- return vabsdiff4(a, b);
- }
-
- __host__ __device__ __forceinline__ VAbsDiff4() {}
- __host__ __device__ __forceinline__ VAbsDiff4(const VAbsDiff4&) {}
- };
-
- ////////////////////////////////////
-
- struct VAbsDiff2 : binary_function<uint, uint, uint>
- {
- __device__ __forceinline__ uint operator ()(uint a, uint b) const
- {
- return vabsdiff2(a, b);
- }
-
- __host__ __device__ __forceinline__ VAbsDiff2() {}
- __host__ __device__ __forceinline__ VAbsDiff2(const VAbsDiff2&) {}
- };
-
- ////////////////////////////////////
-
- __device__ __forceinline__ int _abs(int a)
- {
- return ::abs(a);
- }
- __device__ __forceinline__ float _abs(float a)
- {
- return ::fabsf(a);
- }
- __device__ __forceinline__ double _abs(double a)
- {
- return ::fabs(a);
- }
-
- template <typename T> struct AbsDiffMat : binary_function<T, T, T>
- {
- __device__ __forceinline__ T operator ()(T a, T b) const
- {
- return saturate_cast<T>(_abs(a - b));
- }
-
- __host__ __device__ __forceinline__ AbsDiffMat() {}
- __host__ __device__ __forceinline__ AbsDiffMat(const AbsDiffMat&) {}
- };
-}
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <> struct TransformFunctorTraits< arithm::VAbsDiff4 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
- {
- };
-
- ////////////////////////////////////
-
- template <> struct TransformFunctorTraits< arithm::VAbsDiff2 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
- {
- };
-
- ////////////////////////////////////
-
- template <typename T> struct TransformFunctorTraits< arithm::AbsDiffMat<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-}}}
-
-namespace arithm
-{
- void absDiffMat_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
- {
- cudev::transform(src1, src2, dst, VAbsDiff4(), WithOutMask(), stream);
- }
-
- void absDiffMat_v2(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
- {
- cudev::transform(src1, src2, dst, VAbsDiff2(), WithOutMask(), stream);
- }
-
- template <typename T>
- void absDiffMat(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream)
- {
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<T>) dst, AbsDiffMat<T>(), WithOutMask(), stream);
- }
-
- template void absDiffMat<uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void absDiffMat<schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void absDiffMat<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void absDiffMat<short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void absDiffMat<int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void absDiffMat<float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void absDiffMat<double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// absDiffScalar
-
-namespace arithm
-{
- template <typename T, typename S> struct AbsDiffScalar : unary_function<T, T>
- {
- S val;
-
- explicit AbsDiffScalar(S val_) : val(val_) {}
-
- __device__ __forceinline__ T operator ()(T a) const
- {
- abs_func<S> f;
- return saturate_cast<T>(f(a - val));
- }
- };
-}
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <typename T, typename S> struct TransformFunctorTraits< arithm::AbsDiffScalar<T, S> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-}}}
-
-namespace arithm
-{
- template <typename T, typename S>
- void absDiffScalar(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream)
- {
- AbsDiffScalar<T, S> op(static_cast<S>(val));
-
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) dst, op, WithOutMask(), stream);
- }
-
- template void absDiffScalar<uchar, float>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
- template void absDiffScalar<schar, float>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
- template void absDiffScalar<ushort, float>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
- template void absDiffScalar<short, float>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
- template void absDiffScalar<int, float>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
- template void absDiffScalar<float, float>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
- template void absDiffScalar<double, double>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// absMat
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <typename T> struct TransformFunctorTraits< abs_func<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-}}}
-
-namespace arithm
-{
- template <typename T>
- void absMat(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream)
- {
- cudev::transform((PtrStepSz<T>) src, (PtrStepSz<T>) dst, abs_func<T>(), WithOutMask(), stream);
- }
-
- template void absMat<uchar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void absMat<schar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void absMat<ushort>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void absMat<short>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void absMat<int>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void absMat<float>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void absMat<double>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// sqrMat
-
-namespace arithm
-{
- template <typename T> struct Sqr : unary_function<T, T>
- {
- __device__ __forceinline__ T operator ()(T x) const
- {
- return saturate_cast<T>(x * x);
- }
-
- __host__ __device__ __forceinline__ Sqr() {}
- __host__ __device__ __forceinline__ Sqr(const Sqr&) {}
- };
-}
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <typename T> struct TransformFunctorTraits< arithm::Sqr<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-}}}
-
-namespace arithm
-{
- template <typename T>
- void sqrMat(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream)
- {
- cudev::transform((PtrStepSz<T>) src, (PtrStepSz<T>) dst, Sqr<T>(), WithOutMask(), stream);
- }
-
- template void sqrMat<uchar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void sqrMat<schar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void sqrMat<ushort>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void sqrMat<short>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void sqrMat<int>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void sqrMat<float>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void sqrMat<double>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// sqrtMat
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <typename T> struct TransformFunctorTraits< sqrt_func<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-}}}
-
-namespace arithm
-{
- template <typename T>
- void sqrtMat(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream)
- {
- cudev::transform((PtrStepSz<T>) src, (PtrStepSz<T>) dst, sqrt_func<T>(), WithOutMask(), stream);
- }
-
- template void sqrtMat<uchar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void sqrtMat<schar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void sqrtMat<ushort>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void sqrtMat<short>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void sqrtMat<int>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void sqrtMat<float>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void sqrtMat<double>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// logMat
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <typename T> struct TransformFunctorTraits< log_func<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-}}}
-
-namespace arithm
-{
- template <typename T>
- void logMat(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream)
- {
- cudev::transform((PtrStepSz<T>) src, (PtrStepSz<T>) dst, log_func<T>(), WithOutMask(), stream);
- }
-
- template void logMat<uchar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void logMat<schar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void logMat<ushort>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void logMat<short>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void logMat<int>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void logMat<float>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void logMat<double>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// expMat
-
-namespace arithm
-{
- template <typename T> struct Exp : unary_function<T, T>
- {
- __device__ __forceinline__ T operator ()(T x) const
- {
- exp_func<T> f;
- return saturate_cast<T>(f(x));
- }
-
- __host__ __device__ __forceinline__ Exp() {}
- __host__ __device__ __forceinline__ Exp(const Exp&) {}
- };
-}
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <typename T> struct TransformFunctorTraits< arithm::Exp<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-}}}
-
-namespace arithm
-{
- template <typename T>
- void expMat(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream)
- {
- cudev::transform((PtrStepSz<T>) src, (PtrStepSz<T>) dst, Exp<T>(), WithOutMask(), stream);
- }
-
- template void expMat<uchar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void expMat<schar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void expMat<ushort>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void expMat<short>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void expMat<int>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void expMat<float>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void expMat<double>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////////////////
-// cmpMat
-
-namespace arithm
-{
- struct VCmpEq4 : binary_function<uint, uint, uint>
- {
- __device__ __forceinline__ uint operator ()(uint a, uint b) const
- {
- return vcmpeq4(a, b);
- }
-
- __host__ __device__ __forceinline__ VCmpEq4() {}
- __host__ __device__ __forceinline__ VCmpEq4(const VCmpEq4&) {}
- };
- struct VCmpNe4 : binary_function<uint, uint, uint>
- {
- __device__ __forceinline__ uint operator ()(uint a, uint b) const
- {
- return vcmpne4(a, b);
- }
-
- __host__ __device__ __forceinline__ VCmpNe4() {}
- __host__ __device__ __forceinline__ VCmpNe4(const VCmpNe4&) {}
- };
- struct VCmpLt4 : binary_function<uint, uint, uint>
- {
- __device__ __forceinline__ uint operator ()(uint a, uint b) const
- {
- return vcmplt4(a, b);
- }
-
- __host__ __device__ __forceinline__ VCmpLt4() {}
- __host__ __device__ __forceinline__ VCmpLt4(const VCmpLt4&) {}
- };
- struct VCmpLe4 : binary_function<uint, uint, uint>
- {
- __device__ __forceinline__ uint operator ()(uint a, uint b) const
- {
- return vcmple4(a, b);
- }
-
- __host__ __device__ __forceinline__ VCmpLe4() {}
- __host__ __device__ __forceinline__ VCmpLe4(const VCmpLe4&) {}
- };
-
- ////////////////////////////////////
-
- template <class Op, typename T>
- struct Cmp : binary_function<T, T, uchar>
- {
- __device__ __forceinline__ uchar operator()(T a, T b) const
- {
- Op op;
- return -op(a, b);
- }
- };
-}
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <> struct TransformFunctorTraits< arithm::VCmpEq4 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
- {
- };
- template <> struct TransformFunctorTraits< arithm::VCmpNe4 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
- {
- };
- template <> struct TransformFunctorTraits< arithm::VCmpLt4 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
- {
- };
- template <> struct TransformFunctorTraits< arithm::VCmpLe4 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
- {
- };
-
- ////////////////////////////////////
-
- template <class Op, typename T> struct TransformFunctorTraits< arithm::Cmp<Op, T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(uchar)>
- {
- };
-}}}
-
-namespace arithm
-{
- void cmpMatEq_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
- {
- cudev::transform(src1, src2, dst, VCmpEq4(), WithOutMask(), stream);
- }
- void cmpMatNe_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
- {
- cudev::transform(src1, src2, dst, VCmpNe4(), WithOutMask(), stream);
- }
- void cmpMatLt_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
- {
- cudev::transform(src1, src2, dst, VCmpLt4(), WithOutMask(), stream);
- }
- void cmpMatLe_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
- {
- cudev::transform(src1, src2, dst, VCmpLe4(), WithOutMask(), stream);
- }
-
- template <template <typename> class Op, typename T>
- void cmpMat(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream)
- {
- Cmp<Op<T>, T> op;
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, dst, op, WithOutMask(), stream);
- }
-
- template <typename T> void cmpMatEq(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream)
- {
- cmpMat<equal_to, T>(src1, src2, dst, stream);
- }
- template <typename T> void cmpMatNe(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream)
- {
- cmpMat<not_equal_to, T>(src1, src2, dst, stream);
- }
- template <typename T> void cmpMatLt(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream)
- {
- cmpMat<less, T>(src1, src2, dst, stream);
- }
- template <typename T> void cmpMatLe(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream)
- {
- cmpMat<less_equal, T>(src1, src2, dst, stream);
- }
-
- template void cmpMatEq<uchar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatEq<schar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatEq<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatEq<short >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatEq<int >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatEq<float >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatEq<double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
-
- template void cmpMatNe<uchar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatNe<schar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatNe<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatNe<short >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatNe<int >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatNe<float >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatNe<double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
-
- template void cmpMatLt<uchar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatLt<schar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatLt<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatLt<short >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatLt<int >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatLt<float >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatLt<double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
-
- template void cmpMatLe<uchar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatLe<schar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatLe<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatLe<short >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatLe<int >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatLe<float >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void cmpMatLe<double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////////////////
-// cmpScalar
-
-namespace arithm
-{
-#define TYPE_VEC(type, cn) typename TypeVec<type, cn>::vec_type
-
- template <class Op, typename T, int cn> struct CmpScalar;
- template <class Op, typename T>
- struct CmpScalar<Op, T, 1> : unary_function<T, uchar>
- {
- const T val;
-
- __host__ explicit CmpScalar(T val_) : val(val_) {}
-
- __device__ __forceinline__ uchar operator()(T src) const
- {
- Cmp<Op, T> op;
- return op(src, val);
- }
- };
- template <class Op, typename T>
- struct CmpScalar<Op, T, 2> : unary_function<TYPE_VEC(T, 2), TYPE_VEC(uchar, 2)>
- {
- const TYPE_VEC(T, 2) val;
-
- __host__ explicit CmpScalar(TYPE_VEC(T, 2) val_) : val(val_) {}
-
- __device__ __forceinline__ TYPE_VEC(uchar, 2) operator()(const TYPE_VEC(T, 2) & src) const
- {
- Cmp<Op, T> op;
- return VecTraits<TYPE_VEC(uchar, 2)>::make(op(src.x, val.x), op(src.y, val.y));
- }
- };
- template <class Op, typename T>
- struct CmpScalar<Op, T, 3> : unary_function<TYPE_VEC(T, 3), TYPE_VEC(uchar, 3)>
- {
- const TYPE_VEC(T, 3) val;
-
- __host__ explicit CmpScalar(TYPE_VEC(T, 3) val_) : val(val_) {}
-
- __device__ __forceinline__ TYPE_VEC(uchar, 3) operator()(const TYPE_VEC(T, 3) & src) const
- {
- Cmp<Op, T> op;
- return VecTraits<TYPE_VEC(uchar, 3)>::make(op(src.x, val.x), op(src.y, val.y), op(src.z, val.z));
- }
- };
- template <class Op, typename T>
- struct CmpScalar<Op, T, 4> : unary_function<TYPE_VEC(T, 4), TYPE_VEC(uchar, 4)>
- {
- const TYPE_VEC(T, 4) val;
-
- __host__ explicit CmpScalar(TYPE_VEC(T, 4) val_) : val(val_) {}
-
- __device__ __forceinline__ TYPE_VEC(uchar, 4) operator()(const TYPE_VEC(T, 4) & src) const
- {
- Cmp<Op, T> op;
- return VecTraits<TYPE_VEC(uchar, 4)>::make(op(src.x, val.x), op(src.y, val.y), op(src.z, val.z), op(src.w, val.w));
- }
- };
-
-#undef TYPE_VEC
-}
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <class Op, typename T> struct TransformFunctorTraits< arithm::CmpScalar<Op, T, 1> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(uchar)>
- {
- };
-}}}
-
-namespace arithm
-{
- template <template <typename> class Op, typename T, int cn>
- void cmpScalar(PtrStepSzb src, double val[4], PtrStepSzb dst, cudaStream_t stream)
- {
- typedef typename TypeVec<T, cn>::vec_type src_t;
- typedef typename TypeVec<uchar, cn>::vec_type dst_t;
-
- T sval[] = {static_cast<T>(val[0]), static_cast<T>(val[1]), static_cast<T>(val[2]), static_cast<T>(val[3])};
- src_t val1 = VecTraits<src_t>::make(sval);
-
- CmpScalar<Op<T>, T, cn> op(val1);
- cudev::transform((PtrStepSz<src_t>) src, (PtrStepSz<dst_t>) dst, op, WithOutMask(), stream);
- }
-
- template <typename T> void cmpScalarEq(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream)
- {
- typedef void (*func_t)(PtrStepSzb src, double val[4], PtrStepSzb dst, cudaStream_t stream);
- static const func_t funcs[] =
- {
- 0,
- cmpScalar<equal_to, T, 1>,
- cmpScalar<equal_to, T, 2>,
- cmpScalar<equal_to, T, 3>,
- cmpScalar<equal_to, T, 4>
- };
-
- funcs[cn](src, val, dst, stream);
- }
- template <typename T> void cmpScalarNe(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream)
- {
- typedef void (*func_t)(PtrStepSzb src, double val[4], PtrStepSzb dst, cudaStream_t stream);
- static const func_t funcs[] =
- {
- 0,
- cmpScalar<not_equal_to, T, 1>,
- cmpScalar<not_equal_to, T, 2>,
- cmpScalar<not_equal_to, T, 3>,
- cmpScalar<not_equal_to, T, 4>
- };
-
- funcs[cn](src, val, dst, stream);
- }
- template <typename T> void cmpScalarLt(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream)
- {
- typedef void (*func_t)(PtrStepSzb src, double val[4], PtrStepSzb dst, cudaStream_t stream);
- static const func_t funcs[] =
- {
- 0,
- cmpScalar<less, T, 1>,
- cmpScalar<less, T, 2>,
- cmpScalar<less, T, 3>,
- cmpScalar<less, T, 4>
- };
-
- funcs[cn](src, val, dst, stream);
- }
- template <typename T> void cmpScalarLe(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream)
- {
- typedef void (*func_t)(PtrStepSzb src, double val[4], PtrStepSzb dst, cudaStream_t stream);
- static const func_t funcs[] =
- {
- 0,
- cmpScalar<less_equal, T, 1>,
- cmpScalar<less_equal, T, 2>,
- cmpScalar<less_equal, T, 3>,
- cmpScalar<less_equal, T, 4>
- };
-
- funcs[cn](src, val, dst, stream);
- }
- template <typename T> void cmpScalarGt(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream)
- {
- typedef void (*func_t)(PtrStepSzb src, double val[4], PtrStepSzb dst, cudaStream_t stream);
- static const func_t funcs[] =
- {
- 0,
- cmpScalar<greater, T, 1>,
- cmpScalar<greater, T, 2>,
- cmpScalar<greater, T, 3>,
- cmpScalar<greater, T, 4>
- };
-
- funcs[cn](src, val, dst, stream);
- }
- template <typename T> void cmpScalarGe(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream)
- {
- typedef void (*func_t)(PtrStepSzb src, double val[4], PtrStepSzb dst, cudaStream_t stream);
- static const func_t funcs[] =
- {
- 0,
- cmpScalar<greater_equal, T, 1>,
- cmpScalar<greater_equal, T, 2>,
- cmpScalar<greater_equal, T, 3>,
- cmpScalar<greater_equal, T, 4>
- };
-
- funcs[cn](src, val, dst, stream);
- }
-
- template void cmpScalarEq<uchar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarEq<schar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarEq<ushort>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarEq<short >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarEq<int >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarEq<float >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarEq<double>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
-
- template void cmpScalarNe<uchar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarNe<schar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarNe<ushort>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarNe<short >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarNe<int >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarNe<float >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarNe<double>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
-
- template void cmpScalarLt<uchar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarLt<schar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarLt<ushort>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarLt<short >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarLt<int >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarLt<float >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarLt<double>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
-
- template void cmpScalarLe<uchar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarLe<schar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarLe<ushort>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarLe<short >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarLe<int >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarLe<float >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarLe<double>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
-
- template void cmpScalarGt<uchar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarGt<schar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarGt<ushort>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarGt<short >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarGt<int >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarGt<float >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarGt<double>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
-
- template void cmpScalarGe<uchar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarGe<schar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarGe<ushort>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarGe<short >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarGe<int >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarGe<float >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
- template void cmpScalarGe<double>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////////////////
-// bitMat
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <typename T> struct TransformFunctorTraits< bit_not<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-
- template <typename T> struct TransformFunctorTraits< bit_and<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-
- template <typename T> struct TransformFunctorTraits< bit_or<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-
- template <typename T> struct TransformFunctorTraits< bit_xor<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-}}}
-
-namespace arithm
-{
- template <typename T> void bitMatNot(PtrStepSzb src, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream)
- {
- if (mask.data)
- cudev::transform((PtrStepSz<T>) src, (PtrStepSz<T>) dst, bit_not<T>(), mask, stream);
- else
- cudev::transform((PtrStepSz<T>) src, (PtrStepSz<T>) dst, bit_not<T>(), WithOutMask(), stream);
- }
-
- template <typename T> void bitMatAnd(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream)
- {
- if (mask.data)
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<T>) dst, bit_and<T>(), mask, stream);
- else
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<T>) dst, bit_and<T>(), WithOutMask(), stream);
- }
-
- template <typename T> void bitMatOr(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream)
- {
- if (mask.data)
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<T>) dst, bit_or<T>(), mask, stream);
- else
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<T>) dst, bit_or<T>(), WithOutMask(), stream);
- }
-
- template <typename T> void bitMatXor(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream)
- {
- if (mask.data)
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<T>) dst, bit_xor<T>(), mask, stream);
- else
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<T>) dst, bit_xor<T>(), WithOutMask(), stream);
- }
-
- template void bitMatNot<uchar>(PtrStepSzb src, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void bitMatNot<ushort>(PtrStepSzb src, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void bitMatNot<uint>(PtrStepSzb src, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- template void bitMatAnd<uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void bitMatAnd<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void bitMatAnd<uint>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- template void bitMatOr<uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void bitMatOr<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void bitMatOr<uint>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-
- template void bitMatXor<uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void bitMatXor<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
- template void bitMatXor<uint>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////////////////
-// bitScalar
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <typename T> struct TransformFunctorTraits< binder2nd< bit_and<T> > > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-
- template <typename T> struct TransformFunctorTraits< binder2nd< bit_or<T> > > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-
- template <typename T> struct TransformFunctorTraits< binder2nd< bit_xor<T> > > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-}}}
-
-namespace arithm
-{
- template <typename T> void bitScalarAnd(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream)
- {
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) dst, cv::gpu::cudev::bind2nd(bit_and<T>(), src2), WithOutMask(), stream);
- }
-
- template <typename T> void bitScalarOr(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream)
- {
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) dst, cv::gpu::cudev::bind2nd(bit_or<T>(), src2), WithOutMask(), stream);
- }
-
- template <typename T> void bitScalarXor(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream)
- {
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) dst, cv::gpu::cudev::bind2nd(bit_xor<T>(), src2), WithOutMask(), stream);
- }
-
- template void bitScalarAnd<uchar>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
- template void bitScalarAnd<ushort>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
- template void bitScalarAnd<int>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
- template void bitScalarAnd<unsigned int>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
-
- template void bitScalarOr<uchar>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
- template void bitScalarOr<ushort>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
- template void bitScalarOr<int>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
- template void bitScalarOr<unsigned int>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
-
- template void bitScalarXor<uchar>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
- template void bitScalarXor<ushort>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
- template void bitScalarXor<int>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
- template void bitScalarXor<unsigned int>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// min
-
-namespace arithm
-{
- struct VMin4 : binary_function<uint, uint, uint>
- {
- __device__ __forceinline__ uint operator ()(uint a, uint b) const
- {
- return vmin4(a, b);
- }
-
- __host__ __device__ __forceinline__ VMin4() {}
- __host__ __device__ __forceinline__ VMin4(const VMin4&) {}
- };
-
- ////////////////////////////////////
-
- struct VMin2 : binary_function<uint, uint, uint>
- {
- __device__ __forceinline__ uint operator ()(uint a, uint b) const
- {
- return vmin2(a, b);
- }
-
- __host__ __device__ __forceinline__ VMin2() {}
- __host__ __device__ __forceinline__ VMin2(const VMin2&) {}
- };
-}
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <> struct TransformFunctorTraits< arithm::VMin4 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
- {
- };
-
- ////////////////////////////////////
-
- template <> struct TransformFunctorTraits< arithm::VMin2 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
- {
- };
-
- ////////////////////////////////////
-
- template <typename T> struct TransformFunctorTraits< minimum<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-
- template <typename T> struct TransformFunctorTraits< binder2nd< minimum<T> > > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-}}}
-
-namespace arithm
-{
- void minMat_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
- {
- cudev::transform(src1, src2, dst, VMin4(), WithOutMask(), stream);
- }
-
- void minMat_v2(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
- {
- cudev::transform(src1, src2, dst, VMin2(), WithOutMask(), stream);
- }
-
- template <typename T> void minMat(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream)
- {
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<T>) dst, minimum<T>(), WithOutMask(), stream);
- }
-
- template void minMat<uchar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void minMat<schar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void minMat<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void minMat<short >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void minMat<int >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void minMat<float >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void minMat<double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
-
- template <typename T> void minScalar(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream)
- {
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) dst, cv::gpu::cudev::bind2nd(minimum<T>(), src2), WithOutMask(), stream);
- }
-
- template void minScalar<uchar >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
- template void minScalar<schar >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
- template void minScalar<ushort>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
- template void minScalar<short >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
- template void minScalar<int >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
- template void minScalar<float >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
- template void minScalar<double>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// max
-
-namespace arithm
-{
- struct VMax4 : binary_function<uint, uint, uint>
- {
- __device__ __forceinline__ uint operator ()(uint a, uint b) const
- {
- return vmax4(a, b);
- }
-
- __host__ __device__ __forceinline__ VMax4() {}
- __host__ __device__ __forceinline__ VMax4(const VMax4&) {}
- };
-
- ////////////////////////////////////
-
- struct VMax2 : binary_function<uint, uint, uint>
- {
- __device__ __forceinline__ uint operator ()(uint a, uint b) const
- {
- return vmax2(a, b);
- }
-
- __host__ __device__ __forceinline__ VMax2() {}
- __host__ __device__ __forceinline__ VMax2(const VMax2&) {}
- };
-}
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <> struct TransformFunctorTraits< arithm::VMax4 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
- {
- };
-
- ////////////////////////////////////
-
- template <> struct TransformFunctorTraits< arithm::VMax2 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
- {
- };
-
- ////////////////////////////////////
-
- template <typename T> struct TransformFunctorTraits< maximum<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-
- template <typename T> struct TransformFunctorTraits< binder2nd< maximum<T> > > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-}}}
-
-namespace arithm
-{
- void maxMat_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
- {
- cudev::transform(src1, src2, dst, VMax4(), WithOutMask(), stream);
- }
-
- void maxMat_v2(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
- {
- cudev::transform(src1, src2, dst, VMax2(), WithOutMask(), stream);
- }
-
- template <typename T> void maxMat(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream)
- {
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<T>) dst, maximum<T>(), WithOutMask(), stream);
- }
-
- template void maxMat<uchar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void maxMat<schar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void maxMat<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void maxMat<short >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void maxMat<int >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void maxMat<float >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
- template void maxMat<double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
-
- template <typename T> void maxScalar(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream)
- {
- cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) dst, cv::gpu::cudev::bind2nd(maximum<T>(), src2), WithOutMask(), stream);
- }
-
- template void maxScalar<uchar >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
- template void maxScalar<schar >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
- template void maxScalar<ushort>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
- template void maxScalar<short >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
- template void maxScalar<int >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
- template void maxScalar<float >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
- template void maxScalar<double>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// threshold
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <typename T> struct TransformFunctorTraits< thresh_binary_func<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-
- template <typename T> struct TransformFunctorTraits< thresh_binary_inv_func<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-
- template <typename T> struct TransformFunctorTraits< thresh_trunc_func<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-
- template <typename T> struct TransformFunctorTraits< thresh_to_zero_func<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-
- template <typename T> struct TransformFunctorTraits< thresh_to_zero_inv_func<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-}}}
-
-namespace arithm
-{
- template <template <typename> class Op, typename T>
- void threshold_caller(PtrStepSz<T> src, PtrStepSz<T> dst, T thresh, T maxVal, cudaStream_t stream)
- {
- Op<T> op(thresh, maxVal);
- cudev::transform(src, dst, op, WithOutMask(), stream);
- }
-
- template <typename T>
- void threshold(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream)
- {
- typedef void (*caller_t)(PtrStepSz<T> src, PtrStepSz<T> dst, T thresh, T maxVal, cudaStream_t stream);
-
- static const caller_t callers[] =
- {
- threshold_caller<thresh_binary_func, T>,
- threshold_caller<thresh_binary_inv_func, T>,
- threshold_caller<thresh_trunc_func, T>,
- threshold_caller<thresh_to_zero_func, T>,
- threshold_caller<thresh_to_zero_inv_func, T>
- };
-
- callers[type]((PtrStepSz<T>) src, (PtrStepSz<T>) dst, static_cast<T>(thresh), static_cast<T>(maxVal), stream);
- }
-
- template void threshold<uchar>(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream);
- template void threshold<schar>(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream);
- template void threshold<ushort>(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream);
- template void threshold<short>(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream);
- template void threshold<int>(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream);
- template void threshold<float>(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream);
- template void threshold<double>(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// pow
-
-namespace arithm
-{
- template<typename T, bool Signed = numeric_limits<T>::is_signed> struct PowOp : unary_function<T, T>
- {
- float power;
-
- PowOp(double power_) : power(static_cast<float>(power_)) {}
-
- __device__ __forceinline__ T operator()(T e) const
- {
- return saturate_cast<T>(__powf((float)e, power));
- }
- };
- template<typename T> struct PowOp<T, true> : unary_function<T, T>
- {
- float power;
-
- PowOp(double power_) : power(static_cast<float>(power_)) {}
-
- __device__ __forceinline__ T operator()(T e) const
- {
- T res = saturate_cast<T>(__powf((float)e, power));
-
- if ((e < 0) && (1 & static_cast<int>(power)))
- res *= -1;
-
- return res;
- }
- };
- template<> struct PowOp<float> : unary_function<float, float>
- {
- const float power;
-
- PowOp(double power_) : power(static_cast<float>(power_)) {}
-
- __device__ __forceinline__ float operator()(float e) const
- {
- return __powf(::fabs(e), power);
- }
- };
- template<> struct PowOp<double> : unary_function<double, double>
- {
- double power;
-
- PowOp(double power_) : power(power_) {}
-
- __device__ __forceinline__ double operator()(double e) const
- {
- return ::pow(::fabs(e), power);
- }
- };
-}
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <typename T> struct TransformFunctorTraits< arithm::PowOp<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
- {
- };
-}}}
-
-namespace arithm
-{
- template<typename T>
- void pow(PtrStepSzb src, double power, PtrStepSzb dst, cudaStream_t stream)
- {
- cudev::transform((PtrStepSz<T>) src, (PtrStepSz<T>) dst, PowOp<T>(power), WithOutMask(), stream);
- }
-
- template void pow<uchar>(PtrStepSzb src, double power, PtrStepSzb dst, cudaStream_t stream);
- template void pow<schar>(PtrStepSzb src, double power, PtrStepSzb dst, cudaStream_t stream);
- template void pow<short>(PtrStepSzb src, double power, PtrStepSzb dst, cudaStream_t stream);
- template void pow<ushort>(PtrStepSzb src, double power, PtrStepSzb dst, cudaStream_t stream);
- template void pow<int>(PtrStepSzb src, double power, PtrStepSzb dst, cudaStream_t stream);
- template void pow<float>(PtrStepSzb src, double power, PtrStepSzb dst, cudaStream_t stream);
- template void pow<double>(PtrStepSzb src, double power, PtrStepSzb dst, cudaStream_t stream);
-}
-
-//////////////////////////////////////////////////////////////////////////
-// addWeighted
-
-namespace arithm
-{
- template <typename T> struct UseDouble_
- {
- enum {value = 0};
- };
- template <> struct UseDouble_<double>
- {
- enum {value = 1};
- };
- template <typename T1, typename T2, typename D> struct UseDouble
- {
- enum {value = (UseDouble_<T1>::value || UseDouble_<T2>::value || UseDouble_<D>::value)};
- };
-
- template <typename T1, typename T2, typename D, bool useDouble> struct AddWeighted_;
- template <typename T1, typename T2, typename D> struct AddWeighted_<T1, T2, D, false> : binary_function<T1, T2, D>
- {
- float alpha;
- float beta;
- float gamma;
-
- AddWeighted_(double alpha_, double beta_, double gamma_) : alpha(static_cast<float>(alpha_)), beta(static_cast<float>(beta_)), gamma(static_cast<float>(gamma_)) {}
-
- __device__ __forceinline__ D operator ()(T1 a, T2 b) const
- {
- return saturate_cast<D>(a * alpha + b * beta + gamma);
- }
- };
- template <typename T1, typename T2, typename D> struct AddWeighted_<T1, T2, D, true> : binary_function<T1, T2, D>
- {
- double alpha;
- double beta;
- double gamma;
-
- AddWeighted_(double alpha_, double beta_, double gamma_) : alpha(alpha_), beta(beta_), gamma(gamma_) {}
-
- __device__ __forceinline__ D operator ()(T1 a, T2 b) const
- {
- return saturate_cast<D>(a * alpha + b * beta + gamma);
- }
- };
- template <typename T1, typename T2, typename D> struct AddWeighted : AddWeighted_<T1, T2, D, UseDouble<T1, T2, D>::value>
- {
- AddWeighted(double alpha_, double beta_, double gamma_) : AddWeighted_<T1, T2, D, UseDouble<T1, T2, D>::value>(alpha_, beta_, gamma_) {}
- };
-}
-
-namespace cv { namespace gpu { namespace cudev
-{
- template <typename T1, typename T2, typename D, size_t src1_size, size_t src2_size, size_t dst_size> struct AddWeightedTraits : DefaultTransformFunctorTraits< arithm::AddWeighted<T1, T2, D> >
- {
- };
- template <typename T1, typename T2, typename D, size_t src_size, size_t dst_size> struct AddWeightedTraits<T1, T2, D, src_size, src_size, dst_size> : arithm::ArithmFuncTraits<src_size, dst_size>
- {
- };
-
- template <typename T1, typename T2, typename D> struct TransformFunctorTraits< arithm::AddWeighted<T1, T2, D> > : AddWeightedTraits<T1, T2, D, sizeof(T1), sizeof(T2), sizeof(D)>
- {
- };
-}}}
-
-namespace arithm
-{
- template <typename T1, typename T2, typename D>
- void addWeighted(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream)
- {
- AddWeighted<T1, T2, D> op(alpha, beta, gamma);
-
- cudev::transform((PtrStepSz<T1>) src1, (PtrStepSz<T2>) src2, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
- }
-
- template void addWeighted<uchar, uchar, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, uchar, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, uchar, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, uchar, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, uchar, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, uchar, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, uchar, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<uchar, schar, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, schar, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, schar, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, schar, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, schar, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, schar, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, schar, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<uchar, ushort, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, ushort, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, ushort, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, ushort, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, ushort, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, ushort, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, ushort, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<uchar, short, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, short, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, short, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, short, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, short, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, short, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, short, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<uchar, int, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, int, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, int, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, int, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, int, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, int, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, int, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<uchar, float, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, float, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, float, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, float, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, float, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, float, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, float, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<uchar, double, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, double, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, double, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, double, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, double, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, double, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<uchar, double, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
-
-
- template void addWeighted<schar, schar, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, schar, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, schar, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, schar, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, schar, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, schar, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, schar, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<schar, ushort, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, ushort, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, ushort, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, ushort, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, ushort, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, ushort, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, ushort, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<schar, short, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, short, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, short, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, short, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, short, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, short, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, short, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<schar, int, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, int, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, int, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, int, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, int, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, int, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, int, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<schar, float, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, float, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, float, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, float, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, float, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, float, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, float, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<schar, double, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, double, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, double, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, double, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, double, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, double, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<schar, double, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
-
-
- template void addWeighted<ushort, ushort, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, ushort, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, ushort, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, ushort, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, ushort, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, ushort, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, ushort, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<ushort, short, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, short, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, short, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, short, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, short, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, short, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, short, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<ushort, int, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, int, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, int, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, int, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, int, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, int, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, int, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<ushort, float, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, float, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, float, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, float, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, float, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, float, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, float, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<ushort, double, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, double, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, double, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, double, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, double, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, double, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<ushort, double, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
-
-
- template void addWeighted<short, short, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, short, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, short, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, short, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, short, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, short, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, short, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<short, int, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, int, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, int, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, int, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, int, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, int, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, int, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<short, float, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, float, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, float, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, float, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, float, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, float, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, float, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<short, double, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, double, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, double, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, double, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, double, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, double, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<short, double, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
-
-
- template void addWeighted<int, int, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<int, int, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<int, int, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<int, int, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<int, int, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<int, int, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<int, int, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<int, float, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<int, float, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<int, float, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<int, float, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<int, float, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<int, float, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<int, float, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<int, double, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<int, double, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<int, double, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<int, double, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<int, double, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<int, double, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<int, double, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
-
-
- template void addWeighted<float, float, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<float, float, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<float, float, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<float, float, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<float, float, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<float, float, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<float, float, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
- template void addWeighted<float, double, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<float, double, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<float, double, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<float, double, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<float, double, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<float, double, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<float, double, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-
-
-
- template void addWeighted<double, double, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<double, double, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<double, double, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<double, double, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<double, double, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<double, double, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
- template void addWeighted<double, double, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
-}
-
-#endif /* CUDA_DISABLER */
+++ /dev/null
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-// By downloading, copying, installing or using the software you agree to this license.
-// If you do not agree to this license, do not download, install,
-// copy or use the software.
-//
-//
-// License Agreement
-// For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-// * Redistribution's of source code must retain the above copyright notice,
-// this list of conditions and the following disclaimer.
-//
-// * Redistribution's in binary form must reproduce the above copyright notice,
-// this list of conditions and the following disclaimer in the documentation
-// and/or other materials provided with the distribution.
-//
-// * The name of the copyright holders may not be used to endorse or promote products
-// derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#if !defined CUDA_DISABLER
-
-#include "opencv2/core/cuda/common.hpp"
-#include "opencv2/core/cuda/vec_traits.hpp"
-#include "opencv2/core/cuda/vec_math.hpp"
-#include "opencv2/core/cuda/saturate_cast.hpp"
-#include "opencv2/core/cuda/border_interpolate.hpp"
-#include "internal_shared.hpp"
-
-namespace cv { namespace gpu { namespace cudev
-{
- namespace imgproc
- {
- /////////////////////////////////// MeanShiftfiltering ///////////////////////////////////////////////
-
- texture<uchar4, 2> tex_meanshift;
-
- __device__ short2 do_mean_shift(int x0, int y0, unsigned char* out,
- size_t out_step, int cols, int rows,
- int sp, int sr, int maxIter, float eps)
- {
- int isr2 = sr*sr;
- uchar4 c = tex2D(tex_meanshift, x0, y0 );
-
- // iterate meanshift procedure
- for( int iter = 0; iter < maxIter; iter++ )
- {
- int count = 0;
- int s0 = 0, s1 = 0, s2 = 0, sx = 0, sy = 0;
- float icount;
-
- //mean shift: process pixels in window (p-sigmaSp)x(p+sigmaSp)
- int minx = x0-sp;
- int miny = y0-sp;
- int maxx = x0+sp;
- int maxy = y0+sp;
-
- for( int y = miny; y <= maxy; y++)
- {
- int rowCount = 0;
- for( int x = minx; x <= maxx; x++ )
- {
- uchar4 t = tex2D( tex_meanshift, x, y );
-
- int norm2 = (t.x - c.x) * (t.x - c.x) + (t.y - c.y) * (t.y - c.y) + (t.z - c.z) * (t.z - c.z);
- if( norm2 <= isr2 )
- {
- s0 += t.x; s1 += t.y; s2 += t.z;
- sx += x; rowCount++;
- }
- }
- count += rowCount;
- sy += y*rowCount;
- }
-
- if( count == 0 )
- break;
-
- icount = 1.f/count;
- int x1 = __float2int_rz(sx*icount);
- int y1 = __float2int_rz(sy*icount);
- s0 = __float2int_rz(s0*icount);
- s1 = __float2int_rz(s1*icount);
- s2 = __float2int_rz(s2*icount);
-
- int norm2 = (s0 - c.x) * (s0 - c.x) + (s1 - c.y) * (s1 - c.y) + (s2 - c.z) * (s2 - c.z);
-
- bool stopFlag = (x0 == x1 && y0 == y1) || (::abs(x1-x0) + ::abs(y1-y0) + norm2 <= eps);
-
- x0 = x1; y0 = y1;
- c.x = s0; c.y = s1; c.z = s2;
-
- if( stopFlag )
- break;
- }
-
- int base = (blockIdx.y * blockDim.y + threadIdx.y) * out_step + (blockIdx.x * blockDim.x + threadIdx.x) * 4 * sizeof(uchar);
- *(uchar4*)(out + base) = c;
-
- return make_short2((short)x0, (short)y0);
- }
-
- __global__ void meanshift_kernel(unsigned char* out, size_t out_step, int cols, int rows, int sp, int sr, int maxIter, float eps )
- {
- int x0 = blockIdx.x * blockDim.x + threadIdx.x;
- int y0 = blockIdx.y * blockDim.y + threadIdx.y;
-
- if( x0 < cols && y0 < rows )
- do_mean_shift(x0, y0, out, out_step, cols, rows, sp, sr, maxIter, eps);
- }
-
- __global__ void meanshiftproc_kernel(unsigned char* outr, size_t outrstep,
- unsigned char* outsp, size_t outspstep,
- int cols, int rows,
- int sp, int sr, int maxIter, float eps)
- {
- int x0 = blockIdx.x * blockDim.x + threadIdx.x;
- int y0 = blockIdx.y * blockDim.y + threadIdx.y;
-
- if( x0 < cols && y0 < rows )
- {
- int basesp = (blockIdx.y * blockDim.y + threadIdx.y) * outspstep + (blockIdx.x * blockDim.x + threadIdx.x) * 2 * sizeof(short);
- *(short2*)(outsp + basesp) = do_mean_shift(x0, y0, outr, outrstep, cols, rows, sp, sr, maxIter, eps);
- }
- }
-
- void meanShiftFiltering_gpu(const PtrStepSzb& src, PtrStepSzb dst, int sp, int sr, int maxIter, float eps, cudaStream_t stream)
- {
- dim3 grid(1, 1, 1);
- dim3 threads(32, 8, 1);
- grid.x = divUp(src.cols, threads.x);
- grid.y = divUp(src.rows, threads.y);
-
- cudaChannelFormatDesc desc = cudaCreateChannelDesc<uchar4>();
- cudaSafeCall( cudaBindTexture2D( 0, tex_meanshift, src.data, desc, src.cols, src.rows, src.step ) );
-
- meanshift_kernel<<< grid, threads, 0, stream >>>( dst.data, dst.step, dst.cols, dst.rows, sp, sr, maxIter, eps );
- cudaSafeCall( cudaGetLastError() );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
-
- //cudaSafeCall( cudaUnbindTexture( tex_meanshift ) );
- }
-
- void meanShiftProc_gpu(const PtrStepSzb& src, PtrStepSzb dstr, PtrStepSzb dstsp, int sp, int sr, int maxIter, float eps, cudaStream_t stream)
- {
- dim3 grid(1, 1, 1);
- dim3 threads(32, 8, 1);
- grid.x = divUp(src.cols, threads.x);
- grid.y = divUp(src.rows, threads.y);
-
- cudaChannelFormatDesc desc = cudaCreateChannelDesc<uchar4>();
- cudaSafeCall( cudaBindTexture2D( 0, tex_meanshift, src.data, desc, src.cols, src.rows, src.step ) );
-
- meanshiftproc_kernel<<< grid, threads, 0, stream >>>( dstr.data, dstr.step, dstsp.data, dstsp.step, dstr.cols, dstr.rows, sp, sr, maxIter, eps );
- cudaSafeCall( cudaGetLastError() );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
-
- //cudaSafeCall( cudaUnbindTexture( tex_meanshift ) );
- }
-
- /////////////////////////////////// drawColorDisp ///////////////////////////////////////////////
-
- template <typename T>
- __device__ unsigned int cvtPixel(T d, int ndisp, float S = 1, float V = 1)
- {
- unsigned int H = ((ndisp-d) * 240)/ndisp;
-
- unsigned int hi = (H/60) % 6;
- float f = H/60.f - H/60;
- float p = V * (1 - S);
- float q = V * (1 - f * S);
- float t = V * (1 - (1 - f) * S);
-
- float3 res;
-
- if (hi == 0) //R = V, G = t, B = p
- {
- res.x = p;
- res.y = t;
- res.z = V;
- }
-
- if (hi == 1) // R = q, G = V, B = p
- {
- res.x = p;
- res.y = V;
- res.z = q;
- }
-
- if (hi == 2) // R = p, G = V, B = t
- {
- res.x = t;
- res.y = V;
- res.z = p;
- }
-
- if (hi == 3) // R = p, G = q, B = V
- {
- res.x = V;
- res.y = q;
- res.z = p;
- }
-
- if (hi == 4) // R = t, G = p, B = V
- {
- res.x = V;
- res.y = p;
- res.z = t;
- }
-
- if (hi == 5) // R = V, G = p, B = q
- {
- res.x = q;
- res.y = p;
- res.z = V;
- }
- const unsigned int b = (unsigned int)(::max(0.f, ::min(res.x, 1.f)) * 255.f);
- const unsigned int g = (unsigned int)(::max(0.f, ::min(res.y, 1.f)) * 255.f);
- const unsigned int r = (unsigned int)(::max(0.f, ::min(res.z, 1.f)) * 255.f);
- const unsigned int a = 255U;
-
- return (a << 24) + (r << 16) + (g << 8) + b;
- }
-
- __global__ void drawColorDisp(uchar* disp, size_t disp_step, uchar* out_image, size_t out_step, int width, int height, int ndisp)
- {
- const int x = (blockIdx.x * blockDim.x + threadIdx.x) << 2;
- const int y = blockIdx.y * blockDim.y + threadIdx.y;
-
- if(x < width && y < height)
- {
- uchar4 d4 = *(uchar4*)(disp + y * disp_step + x);
-
- uint4 res;
- res.x = cvtPixel(d4.x, ndisp);
- res.y = cvtPixel(d4.y, ndisp);
- res.z = cvtPixel(d4.z, ndisp);
- res.w = cvtPixel(d4.w, ndisp);
-
- uint4* line = (uint4*)(out_image + y * out_step);
- line[x >> 2] = res;
- }
- }
-
- __global__ void drawColorDisp(short* disp, size_t disp_step, uchar* out_image, size_t out_step, int width, int height, int ndisp)
- {
- const int x = (blockIdx.x * blockDim.x + threadIdx.x) << 1;
- const int y = blockIdx.y * blockDim.y + threadIdx.y;
-
- if(x < width && y < height)
- {
- short2 d2 = *(short2*)(disp + y * disp_step + x);
-
- uint2 res;
- res.x = cvtPixel(d2.x, ndisp);
- res.y = cvtPixel(d2.y, ndisp);
-
- uint2* line = (uint2*)(out_image + y * out_step);
- line[x >> 1] = res;
- }
- }
-
-
- void drawColorDisp_gpu(const PtrStepSzb& src, const PtrStepSzb& dst, int ndisp, const cudaStream_t& stream)
- {
- dim3 threads(16, 16, 1);
- dim3 grid(1, 1, 1);
- grid.x = divUp(src.cols, threads.x << 2);
- grid.y = divUp(src.rows, threads.y);
-
- drawColorDisp<<<grid, threads, 0, stream>>>(src.data, src.step, dst.data, dst.step, src.cols, src.rows, ndisp);
- cudaSafeCall( cudaGetLastError() );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
-
- void drawColorDisp_gpu(const PtrStepSz<short>& src, const PtrStepSzb& dst, int ndisp, const cudaStream_t& stream)
- {
- dim3 threads(32, 8, 1);
- dim3 grid(1, 1, 1);
- grid.x = divUp(src.cols, threads.x << 1);
- grid.y = divUp(src.rows, threads.y);
-
- drawColorDisp<<<grid, threads, 0, stream>>>(src.data, src.step / sizeof(short), dst.data, dst.step, src.cols, src.rows, ndisp);
- cudaSafeCall( cudaGetLastError() );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
-
- /////////////////////////////////// reprojectImageTo3D ///////////////////////////////////////////////
-
- __constant__ float cq[16];
-
- template <typename T, typename D>
- __global__ void reprojectImageTo3D(const PtrStepSz<T> disp, PtrStep<D> xyz)
- {
- const int x = blockIdx.x * blockDim.x + threadIdx.x;
- const int y = blockIdx.y * blockDim.y + threadIdx.y;
-
- if (y >= disp.rows || x >= disp.cols)
- return;
-
- const float qx = x * cq[ 0] + y * cq[ 1] + cq[ 3];
- const float qy = x * cq[ 4] + y * cq[ 5] + cq[ 7];
- const float qz = x * cq[ 8] + y * cq[ 9] + cq[11];
- const float qw = x * cq[12] + y * cq[13] + cq[15];
-
- const T d = disp(y, x);
-
- const float iW = 1.f / (qw + cq[14] * d);
-
- D v = VecTraits<D>::all(1.0f);
- v.x = (qx + cq[2] * d) * iW;
- v.y = (qy + cq[6] * d) * iW;
- v.z = (qz + cq[10] * d) * iW;
-
- xyz(y, x) = v;
- }
-
- template <typename T, typename D>
- void reprojectImageTo3D_gpu(const PtrStepSzb disp, PtrStepSzb xyz, const float* q, cudaStream_t stream)
- {
- dim3 block(32, 8);
- dim3 grid(divUp(disp.cols, block.x), divUp(disp.rows, block.y));
-
- cudaSafeCall( cudaMemcpyToSymbol(cq, q, 16 * sizeof(float)) );
-
- reprojectImageTo3D<T, D><<<grid, block, 0, stream>>>((PtrStepSz<T>)disp, (PtrStepSz<D>)xyz);
- cudaSafeCall( cudaGetLastError() );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
-
- template void reprojectImageTo3D_gpu<uchar, float3>(const PtrStepSzb disp, PtrStepSzb xyz, const float* q, cudaStream_t stream);
- template void reprojectImageTo3D_gpu<uchar, float4>(const PtrStepSzb disp, PtrStepSzb xyz, const float* q, cudaStream_t stream);
- template void reprojectImageTo3D_gpu<short, float3>(const PtrStepSzb disp, PtrStepSzb xyz, const float* q, cudaStream_t stream);
- template void reprojectImageTo3D_gpu<short, float4>(const PtrStepSzb disp, PtrStepSzb xyz, const float* q, cudaStream_t stream);
-
- /////////////////////////////////////////// Corner Harris /////////////////////////////////////////////////
-
- texture<float, cudaTextureType2D, cudaReadModeElementType> harrisDxTex(0, cudaFilterModePoint, cudaAddressModeClamp);
- texture<float, cudaTextureType2D, cudaReadModeElementType> harrisDyTex(0, cudaFilterModePoint, cudaAddressModeClamp);
-
- __global__ void cornerHarris_kernel(const int block_size, const float k, PtrStepSzf dst)
- {
- const int x = blockIdx.x * blockDim.x + threadIdx.x;
- const int y = blockIdx.y * blockDim.y + threadIdx.y;
-
- if (x < dst.cols && y < dst.rows)
- {
- float a = 0.f;
- float b = 0.f;
- float c = 0.f;
-
- const int ibegin = y - (block_size / 2);
- const int jbegin = x - (block_size / 2);
- const int iend = ibegin + block_size;
- const int jend = jbegin + block_size;
-
- for (int i = ibegin; i < iend; ++i)
- {
- for (int j = jbegin; j < jend; ++j)
- {
- float dx = tex2D(harrisDxTex, j, i);
- float dy = tex2D(harrisDyTex, j, i);
-
- a += dx * dx;
- b += dx * dy;
- c += dy * dy;
- }
- }
-
- dst(y, x) = a * c - b * b - k * (a + c) * (a + c);
- }
- }
-
- template <typename BR, typename BC>
- __global__ void cornerHarris_kernel(const int block_size, const float k, PtrStepSzf dst, const BR border_row, const BC border_col)
- {
- const int x = blockIdx.x * blockDim.x + threadIdx.x;
- const int y = blockIdx.y * blockDim.y + threadIdx.y;
-
- if (x < dst.cols && y < dst.rows)
- {
- float a = 0.f;
- float b = 0.f;
- float c = 0.f;
-
- const int ibegin = y - (block_size / 2);
- const int jbegin = x - (block_size / 2);
- const int iend = ibegin + block_size;
- const int jend = jbegin + block_size;
-
- for (int i = ibegin; i < iend; ++i)
- {
- const int y = border_col.idx_row(i);
-
- for (int j = jbegin; j < jend; ++j)
- {
- const int x = border_row.idx_col(j);
-
- float dx = tex2D(harrisDxTex, x, y);
- float dy = tex2D(harrisDyTex, x, y);
-
- a += dx * dx;
- b += dx * dy;
- c += dy * dy;
- }
- }
-
- dst(y, x) = a * c - b * b - k * (a + c) * (a + c);
- }
- }
-
- void cornerHarris_gpu(int block_size, float k, PtrStepSzf Dx, PtrStepSzf Dy, PtrStepSzf dst, int border_type, cudaStream_t stream)
- {
- dim3 block(32, 8);
- dim3 grid(divUp(Dx.cols, block.x), divUp(Dx.rows, block.y));
-
- bindTexture(&harrisDxTex, Dx);
- bindTexture(&harrisDyTex, Dy);
-
- switch (border_type)
- {
- case BORDER_REFLECT101_GPU:
- cornerHarris_kernel<<<grid, block, 0, stream>>>(block_size, k, dst, BrdRowReflect101<void>(Dx.cols), BrdColReflect101<void>(Dx.rows));
- break;
-
- case BORDER_REFLECT_GPU:
- cornerHarris_kernel<<<grid, block, 0, stream>>>(block_size, k, dst, BrdRowReflect<void>(Dx.cols), BrdColReflect<void>(Dx.rows));
- break;
-
- case BORDER_REPLICATE_GPU:
- cornerHarris_kernel<<<grid, block, 0, stream>>>(block_size, k, dst);
- break;
- }
-
- cudaSafeCall( cudaGetLastError() );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
-
- /////////////////////////////////////////// Corner Min Eigen Val /////////////////////////////////////////////////
-
- texture<float, cudaTextureType2D, cudaReadModeElementType> minEigenValDxTex(0, cudaFilterModePoint, cudaAddressModeClamp);
- texture<float, cudaTextureType2D, cudaReadModeElementType> minEigenValDyTex(0, cudaFilterModePoint, cudaAddressModeClamp);
-
- __global__ void cornerMinEigenVal_kernel(const int block_size, PtrStepSzf dst)
- {
- const int x = blockIdx.x * blockDim.x + threadIdx.x;
- const int y = blockIdx.y * blockDim.y + threadIdx.y;
-
- if (x < dst.cols && y < dst.rows)
- {
- float a = 0.f;
- float b = 0.f;
- float c = 0.f;
-
- const int ibegin = y - (block_size / 2);
- const int jbegin = x - (block_size / 2);
- const int iend = ibegin + block_size;
- const int jend = jbegin + block_size;
-
- for (int i = ibegin; i < iend; ++i)
- {
- for (int j = jbegin; j < jend; ++j)
- {
- float dx = tex2D(minEigenValDxTex, j, i);
- float dy = tex2D(minEigenValDyTex, j, i);
-
- a += dx * dx;
- b += dx * dy;
- c += dy * dy;
- }
- }
-
- a *= 0.5f;
- c *= 0.5f;
-
- dst(y, x) = (a + c) - sqrtf((a - c) * (a - c) + b * b);
- }
- }
-
-
- template <typename BR, typename BC>
- __global__ void cornerMinEigenVal_kernel(const int block_size, PtrStepSzf dst, const BR border_row, const BC border_col)
- {
- const int x = blockIdx.x * blockDim.x + threadIdx.x;
- const int y = blockIdx.y * blockDim.y + threadIdx.y;
-
- if (x < dst.cols && y < dst.rows)
- {
- float a = 0.f;
- float b = 0.f;
- float c = 0.f;
-
- const int ibegin = y - (block_size / 2);
- const int jbegin = x - (block_size / 2);
- const int iend = ibegin + block_size;
- const int jend = jbegin + block_size;
-
- for (int i = ibegin; i < iend; ++i)
- {
- int y = border_col.idx_row(i);
-
- for (int j = jbegin; j < jend; ++j)
- {
- int x = border_row.idx_col(j);
-
- float dx = tex2D(minEigenValDxTex, x, y);
- float dy = tex2D(minEigenValDyTex, x, y);
-
- a += dx * dx;
- b += dx * dy;
- c += dy * dy;
- }
- }
-
- a *= 0.5f;
- c *= 0.5f;
-
- dst(y, x) = (a + c) - sqrtf((a - c) * (a - c) + b * b);
- }
- }
-
- void cornerMinEigenVal_gpu(int block_size, PtrStepSzf Dx, PtrStepSzf Dy, PtrStepSzf dst, int border_type, cudaStream_t stream)
- {
- dim3 block(32, 8);
- dim3 grid(divUp(Dx.cols, block.x), divUp(Dx.rows, block.y));
-
- bindTexture(&minEigenValDxTex, Dx);
- bindTexture(&minEigenValDyTex, Dy);
-
- switch (border_type)
- {
- case BORDER_REFLECT101_GPU:
- cornerMinEigenVal_kernel<<<grid, block, 0, stream>>>(block_size, dst, BrdRowReflect101<void>(Dx.cols), BrdColReflect101<void>(Dx.rows));
- break;
-
- case BORDER_REFLECT_GPU:
- cornerMinEigenVal_kernel<<<grid, block, 0, stream>>>(block_size, dst, BrdRowReflect<void>(Dx.cols), BrdColReflect<void>(Dx.rows));
- break;
-
- case BORDER_REPLICATE_GPU:
- cornerMinEigenVal_kernel<<<grid, block, 0, stream>>>(block_size, dst);
- break;
- }
-
- cudaSafeCall( cudaGetLastError() );
-
- if (stream == 0)
- cudaSafeCall(cudaDeviceSynchronize());
- }
-
- ////////////////////////////// Column Sum //////////////////////////////////////
-
- __global__ void column_sumKernel_32F(int cols, int rows, const PtrStepb src, const PtrStepb dst)
- {
- int x = blockIdx.x * blockDim.x + threadIdx.x;
-
- if (x < cols)
- {
- const unsigned char* src_data = src.data + x * sizeof(float);
- unsigned char* dst_data = dst.data + x * sizeof(float);
-
- float sum = 0.f;
- for (int y = 0; y < rows; ++y)
- {
- sum += *(const float*)src_data;
- *(float*)dst_data = sum;
- src_data += src.step;
- dst_data += dst.step;
- }
- }
- }
-
-
- void columnSum_32F(const PtrStepSzb src, const PtrStepSzb dst)
- {
- dim3 threads(256);
- dim3 grid(divUp(src.cols, threads.x));
-
- column_sumKernel_32F<<<grid, threads>>>(src.cols, src.rows, src, dst);
- cudaSafeCall( cudaGetLastError() );
-
- cudaSafeCall( cudaDeviceSynchronize() );
- }
-
-
- //////////////////////////////////////////////////////////////////////////
- // mulSpectrums
-
- __global__ void mulSpectrumsKernel(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, PtrStepSz<cufftComplex> c)
- {
- const int x = blockIdx.x * blockDim.x + threadIdx.x;
- const int y = blockIdx.y * blockDim.y + threadIdx.y;
-
- if (x < c.cols && y < c.rows)
- {
- c.ptr(y)[x] = cuCmulf(a.ptr(y)[x], b.ptr(y)[x]);
- }
- }
-
-
- void mulSpectrums(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, PtrStepSz<cufftComplex> c, cudaStream_t stream)
- {
- dim3 threads(256);
- dim3 grid(divUp(c.cols, threads.x), divUp(c.rows, threads.y));
-
- mulSpectrumsKernel<<<grid, threads, 0, stream>>>(a, b, c);
- cudaSafeCall( cudaGetLastError() );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
-
-
- //////////////////////////////////////////////////////////////////////////
- // mulSpectrums_CONJ
-
- __global__ void mulSpectrumsKernel_CONJ(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, PtrStepSz<cufftComplex> c)
- {
- const int x = blockIdx.x * blockDim.x + threadIdx.x;
- const int y = blockIdx.y * blockDim.y + threadIdx.y;
-
- if (x < c.cols && y < c.rows)
- {
- c.ptr(y)[x] = cuCmulf(a.ptr(y)[x], cuConjf(b.ptr(y)[x]));
- }
- }
-
-
- void mulSpectrums_CONJ(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, PtrStepSz<cufftComplex> c, cudaStream_t stream)
- {
- dim3 threads(256);
- dim3 grid(divUp(c.cols, threads.x), divUp(c.rows, threads.y));
-
- mulSpectrumsKernel_CONJ<<<grid, threads, 0, stream>>>(a, b, c);
- cudaSafeCall( cudaGetLastError() );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
-
-
- //////////////////////////////////////////////////////////////////////////
- // mulAndScaleSpectrums
-
- __global__ void mulAndScaleSpectrumsKernel(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, float scale, PtrStepSz<cufftComplex> c)
- {
- const int x = blockIdx.x * blockDim.x + threadIdx.x;
- const int y = blockIdx.y * blockDim.y + threadIdx.y;
-
- if (x < c.cols && y < c.rows)
- {
- cufftComplex v = cuCmulf(a.ptr(y)[x], b.ptr(y)[x]);
- c.ptr(y)[x] = make_cuFloatComplex(cuCrealf(v) * scale, cuCimagf(v) * scale);
- }
- }
-
-
- void mulAndScaleSpectrums(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, float scale, PtrStepSz<cufftComplex> c, cudaStream_t stream)
- {
- dim3 threads(256);
- dim3 grid(divUp(c.cols, threads.x), divUp(c.rows, threads.y));
-
- mulAndScaleSpectrumsKernel<<<grid, threads, 0, stream>>>(a, b, scale, c);
- cudaSafeCall( cudaGetLastError() );
-
- if (stream)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
-
-
- //////////////////////////////////////////////////////////////////////////
- // mulAndScaleSpectrums_CONJ
-
- __global__ void mulAndScaleSpectrumsKernel_CONJ(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, float scale, PtrStepSz<cufftComplex> c)
- {
- const int x = blockIdx.x * blockDim.x + threadIdx.x;
- const int y = blockIdx.y * blockDim.y + threadIdx.y;
-
- if (x < c.cols && y < c.rows)
- {
- cufftComplex v = cuCmulf(a.ptr(y)[x], cuConjf(b.ptr(y)[x]));
- c.ptr(y)[x] = make_cuFloatComplex(cuCrealf(v) * scale, cuCimagf(v) * scale);
- }
- }
-
-
- void mulAndScaleSpectrums_CONJ(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, float scale, PtrStepSz<cufftComplex> c, cudaStream_t stream)
- {
- dim3 threads(256);
- dim3 grid(divUp(c.cols, threads.x), divUp(c.rows, threads.y));
-
- mulAndScaleSpectrumsKernel_CONJ<<<grid, threads, 0, stream>>>(a, b, scale, c);
- cudaSafeCall( cudaGetLastError() );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
-
- //////////////////////////////////////////////////////////////////////////
- // buildWarpMaps
-
- // TODO use intrinsics like __sinf and so on
-
- namespace build_warp_maps
- {
-
- __constant__ float ck_rinv[9];
- __constant__ float cr_kinv[9];
- __constant__ float ct[3];
- __constant__ float cscale;
- }
-
-
- class PlaneMapper
- {
- public:
- static __device__ __forceinline__ void mapBackward(float u, float v, float &x, float &y)
- {
- using namespace build_warp_maps;
-
- float x_ = u / cscale - ct[0];
- float y_ = v / cscale - ct[1];
-
- float z;
- x = ck_rinv[0] * x_ + ck_rinv[1] * y_ + ck_rinv[2] * (1 - ct[2]);
- y = ck_rinv[3] * x_ + ck_rinv[4] * y_ + ck_rinv[5] * (1 - ct[2]);
- z = ck_rinv[6] * x_ + ck_rinv[7] * y_ + ck_rinv[8] * (1 - ct[2]);
-
- x /= z;
- y /= z;
- }
- };
-
-
- class CylindricalMapper
- {
- public:
- static __device__ __forceinline__ void mapBackward(float u, float v, float &x, float &y)
- {
- using namespace build_warp_maps;
-
- u /= cscale;
- float x_ = ::sinf(u);
- float y_ = v / cscale;
- float z_ = ::cosf(u);
-
- float z;
- x = ck_rinv[0] * x_ + ck_rinv[1] * y_ + ck_rinv[2] * z_;
- y = ck_rinv[3] * x_ + ck_rinv[4] * y_ + ck_rinv[5] * z_;
- z = ck_rinv[6] * x_ + ck_rinv[7] * y_ + ck_rinv[8] * z_;
-
- if (z > 0) { x /= z; y /= z; }
- else x = y = -1;
- }
- };
-
-
- class SphericalMapper
- {
- public:
- static __device__ __forceinline__ void mapBackward(float u, float v, float &x, float &y)
- {
- using namespace build_warp_maps;
-
- v /= cscale;
- u /= cscale;
-
- float sinv = ::sinf(v);
- float x_ = sinv * ::sinf(u);
- float y_ = -::cosf(v);
- float z_ = sinv * ::cosf(u);
-
- float z;
- x = ck_rinv[0] * x_ + ck_rinv[1] * y_ + ck_rinv[2] * z_;
- y = ck_rinv[3] * x_ + ck_rinv[4] * y_ + ck_rinv[5] * z_;
- z = ck_rinv[6] * x_ + ck_rinv[7] * y_ + ck_rinv[8] * z_;
-
- if (z > 0) { x /= z; y /= z; }
- else x = y = -1;
- }
- };
-
-
- template <typename Mapper>
- __global__ void buildWarpMapsKernel(int tl_u, int tl_v, int cols, int rows,
- PtrStepf map_x, PtrStepf map_y)
- {
- int du = blockIdx.x * blockDim.x + threadIdx.x;
- int dv = blockIdx.y * blockDim.y + threadIdx.y;
- if (du < cols && dv < rows)
- {
- float u = tl_u + du;
- float v = tl_v + dv;
- float x, y;
- Mapper::mapBackward(u, v, x, y);
- map_x.ptr(dv)[du] = x;
- map_y.ptr(dv)[du] = y;
- }
- }
-
-
- void buildWarpPlaneMaps(int tl_u, int tl_v, PtrStepSzf map_x, PtrStepSzf map_y,
- const float k_rinv[9], const float r_kinv[9], const float t[3],
- float scale, cudaStream_t stream)
- {
- cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::ck_rinv, k_rinv, 9*sizeof(float)));
- cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::cr_kinv, r_kinv, 9*sizeof(float)));
- cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::ct, t, 3*sizeof(float)));
- cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::cscale, &scale, sizeof(float)));
-
- int cols = map_x.cols;
- int rows = map_x.rows;
-
- dim3 threads(32, 8);
- dim3 grid(divUp(cols, threads.x), divUp(rows, threads.y));
-
- buildWarpMapsKernel<PlaneMapper><<<grid,threads>>>(tl_u, tl_v, cols, rows, map_x, map_y);
- cudaSafeCall(cudaGetLastError());
- if (stream == 0)
- cudaSafeCall(cudaDeviceSynchronize());
- }
-
-
- void buildWarpCylindricalMaps(int tl_u, int tl_v, PtrStepSzf map_x, PtrStepSzf map_y,
- const float k_rinv[9], const float r_kinv[9], float scale,
- cudaStream_t stream)
- {
- cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::ck_rinv, k_rinv, 9*sizeof(float)));
- cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::cr_kinv, r_kinv, 9*sizeof(float)));
- cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::cscale, &scale, sizeof(float)));
-
- int cols = map_x.cols;
- int rows = map_x.rows;
-
- dim3 threads(32, 8);
- dim3 grid(divUp(cols, threads.x), divUp(rows, threads.y));
-
- buildWarpMapsKernel<CylindricalMapper><<<grid,threads>>>(tl_u, tl_v, cols, rows, map_x, map_y);
- cudaSafeCall(cudaGetLastError());
- if (stream == 0)
- cudaSafeCall(cudaDeviceSynchronize());
- }
-
-
- void buildWarpSphericalMaps(int tl_u, int tl_v, PtrStepSzf map_x, PtrStepSzf map_y,
- const float k_rinv[9], const float r_kinv[9], float scale,
- cudaStream_t stream)
- {
- cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::ck_rinv, k_rinv, 9*sizeof(float)));
- cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::cr_kinv, r_kinv, 9*sizeof(float)));
- cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::cscale, &scale, sizeof(float)));
-
- int cols = map_x.cols;
- int rows = map_x.rows;
-
- dim3 threads(32, 8);
- dim3 grid(divUp(cols, threads.x), divUp(rows, threads.y));
-
- buildWarpMapsKernel<SphericalMapper><<<grid,threads>>>(tl_u, tl_v, cols, rows, map_x, map_y);
- cudaSafeCall(cudaGetLastError());
- if (stream == 0)
- cudaSafeCall(cudaDeviceSynchronize());
- }
-
- //////////////////////////////////////////////////////////////////////////
- // filter2D
-
- #define FILTER2D_MAX_KERNEL_SIZE 16
-
- __constant__ float c_filter2DKernel[FILTER2D_MAX_KERNEL_SIZE * FILTER2D_MAX_KERNEL_SIZE];
-
- template <class SrcT, typename D>
- __global__ void filter2D(const SrcT src, PtrStepSz<D> dst, const int kWidth, const int kHeight, const int anchorX, const int anchorY)
- {
- typedef typename TypeVec<float, VecTraits<D>::cn>::vec_type sum_t;
-
- const int x = blockIdx.x * blockDim.x + threadIdx.x;
- const int y = blockIdx.y * blockDim.y + threadIdx.y;
-
- if (x >= dst.cols || y >= dst.rows)
- return;
-
- sum_t res = VecTraits<sum_t>::all(0);
- int kInd = 0;
-
- for (int i = 0; i < kHeight; ++i)
- {
- for (int j = 0; j < kWidth; ++j)
- res = res + src(y - anchorY + i, x - anchorX + j) * c_filter2DKernel[kInd++];
- }
-
- dst(y, x) = saturate_cast<D>(res);
- }
-
- template <typename T, typename D, template <typename> class Brd> struct Filter2DCaller;
-
- #define IMPLEMENT_FILTER2D_TEX_READER(type) \
- texture< type , cudaTextureType2D, cudaReadModeElementType> tex_filter2D_ ## type (0, cudaFilterModePoint, cudaAddressModeClamp); \
- struct tex_filter2D_ ## type ## _reader \
- { \
- typedef type elem_type; \
- typedef int index_type; \
- const int xoff; \
- const int yoff; \
- tex_filter2D_ ## type ## _reader (int xoff_, int yoff_) : xoff(xoff_), yoff(yoff_) {} \
- __device__ __forceinline__ elem_type operator ()(index_type y, index_type x) const \
- { \
- return tex2D(tex_filter2D_ ## type , x + xoff, y + yoff); \
- } \
- }; \
- template <typename D, template <typename> class Brd> struct Filter2DCaller< type , D, Brd> \
- { \
- static void call(const PtrStepSz< type > srcWhole, int xoff, int yoff, PtrStepSz<D> dst, \
- int kWidth, int kHeight, int anchorX, int anchorY, const float* borderValue, cudaStream_t stream) \
- { \
- typedef typename TypeVec<float, VecTraits< type >::cn>::vec_type work_type; \
- dim3 block(16, 16); \
- dim3 grid(divUp(dst.cols, block.x), divUp(dst.rows, block.y)); \
- bindTexture(&tex_filter2D_ ## type , srcWhole); \
- tex_filter2D_ ## type ##_reader texSrc(xoff, yoff); \
- Brd<work_type> brd(dst.rows, dst.cols, VecTraits<work_type>::make(borderValue)); \
- BorderReader< tex_filter2D_ ## type ##_reader, Brd<work_type> > brdSrc(texSrc, brd); \
- filter2D<<<grid, block, 0, stream>>>(brdSrc, dst, kWidth, kHeight, anchorX, anchorY); \
- cudaSafeCall( cudaGetLastError() ); \
- if (stream == 0) \
- cudaSafeCall( cudaDeviceSynchronize() ); \
- } \
- };
-
- IMPLEMENT_FILTER2D_TEX_READER(uchar);
- IMPLEMENT_FILTER2D_TEX_READER(uchar4);
-
- IMPLEMENT_FILTER2D_TEX_READER(ushort);
- IMPLEMENT_FILTER2D_TEX_READER(ushort4);
-
- IMPLEMENT_FILTER2D_TEX_READER(float);
- IMPLEMENT_FILTER2D_TEX_READER(float4);
-
- #undef IMPLEMENT_FILTER2D_TEX_READER
-
- template <typename T, typename D>
- void filter2D_gpu(PtrStepSzb srcWhole, int ofsX, int ofsY, PtrStepSzb dst,
- int kWidth, int kHeight, int anchorX, int anchorY, const float* kernel,
- int borderMode, const float* borderValue, cudaStream_t stream)
- {
- typedef void (*func_t)(const PtrStepSz<T> srcWhole, int xoff, int yoff, PtrStepSz<D> dst, int kWidth, int kHeight, int anchorX, int anchorY, const float* borderValue, cudaStream_t stream);
- static const func_t funcs[] =
- {
- Filter2DCaller<T, D, BrdReflect101>::call,
- Filter2DCaller<T, D, BrdReplicate>::call,
- Filter2DCaller<T, D, BrdConstant>::call,
- Filter2DCaller<T, D, BrdReflect>::call,
- Filter2DCaller<T, D, BrdWrap>::call
- };
-
- if (stream == 0)
- cudaSafeCall( cudaMemcpyToSymbol(c_filter2DKernel, kernel, kWidth * kHeight * sizeof(float), 0, cudaMemcpyDeviceToDevice) );
- else
- cudaSafeCall( cudaMemcpyToSymbolAsync(c_filter2DKernel, kernel, kWidth * kHeight * sizeof(float), 0, cudaMemcpyDeviceToDevice, stream) );
-
- funcs[borderMode](static_cast< PtrStepSz<T> >(srcWhole), ofsX, ofsY, static_cast< PtrStepSz<D> >(dst), kWidth, kHeight, anchorX, anchorY, borderValue, stream);
- }
-
- template void filter2D_gpu<uchar, uchar>(PtrStepSzb srcWhole, int ofsX, int ofsY, PtrStepSzb dst, int kWidth, int kHeight, int anchorX, int anchorY, const float* kernel, int borderMode, const float* borderValue, cudaStream_t stream);
- template void filter2D_gpu<uchar4, uchar4>(PtrStepSzb srcWhole, int ofsX, int ofsY, PtrStepSzb dst, int kWidth, int kHeight, int anchorX, int anchorY, const float* kernel, int borderMode, const float* borderValue, cudaStream_t stream);
- template void filter2D_gpu<ushort, ushort>(PtrStepSzb srcWhole, int ofsX, int ofsY, PtrStepSzb dst, int kWidth, int kHeight, int anchorX, int anchorY, const float* kernel, int borderMode, const float* borderValue, cudaStream_t stream);
- template void filter2D_gpu<ushort4, ushort4>(PtrStepSzb srcWhole, int ofsX, int ofsY, PtrStepSzb dst, int kWidth, int kHeight, int anchorX, int anchorY, const float* kernel, int borderMode, const float* borderValue, cudaStream_t stream);
- template void filter2D_gpu<float, float>(PtrStepSzb srcWhole, int ofsX, int ofsY, PtrStepSzb dst, int kWidth, int kHeight, int anchorX, int anchorY, const float* kernel, int borderMode, const float* borderValue, cudaStream_t stream);
- template void filter2D_gpu<float4, float4>(PtrStepSzb srcWhole, int ofsX, int ofsY, PtrStepSzb dst, int kWidth, int kHeight, int anchorX, int anchorY, const float* kernel, int borderMode, const float* borderValue, cudaStream_t stream);
- } // namespace imgproc
-}}} // namespace cv { namespace gpu { namespace cudev {
-
-
-#endif /* CUDA_DISABLER */
+++ /dev/null
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-// By downloading, copying, installing or using the software you agree to this license.
-// If you do not agree to this license, do not download, install,
-// copy or use the software.
-//
-//
-// License Agreement
-// For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-// * Redistribution's of source code must retain the above copyright notice,
-// this list of conditions and the following disclaimer.
-//
-// * Redistribution's in binary form must reproduce the above copyright notice,
-// this list of conditions and the following disclaimer in the documentation
-// and/or other materials provided with the distribution.
-//
-// * The name of the copyright holders may not be used to endorse or promote products
-// derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#if !defined CUDA_DISABLER
-
-#include "opencv2/core/cuda/common.hpp"
-#include "opencv2/core/cuda/limits.hpp"
-#include "opencv2/core/cuda/saturate_cast.hpp"
-#include "opencv2/core/cuda/vec_traits.hpp"
-#include "opencv2/core/cuda/vec_math.hpp"
-#include "opencv2/core/cuda/reduce.hpp"
-#include "opencv2/core/cuda/functional.hpp"
-#include "opencv2/core/cuda/utility.hpp"
-#include "opencv2/core/cuda/type_traits.hpp"
-
-using namespace cv::gpu;
-using namespace cv::gpu::cudev;
-
-namespace detail
-{
- __device__ __forceinline__ int cvAtomicAdd(int* address, int val)
- {
- return ::atomicAdd(address, val);
- }
- __device__ __forceinline__ unsigned int cvAtomicAdd(unsigned int* address, unsigned int val)
- {
- return ::atomicAdd(address, val);
- }
- __device__ __forceinline__ float cvAtomicAdd(float* address, float val)
- {
- #if __CUDA_ARCH__ >= 200
- return ::atomicAdd(address, val);
- #else
- int* address_as_i = (int*) address;
- int old = *address_as_i, assumed;
- do {
- assumed = old;
- old = ::atomicCAS(address_as_i, assumed,
- __float_as_int(val + __int_as_float(assumed)));
- } while (assumed != old);
- return __int_as_float(old);
- #endif
- }
- __device__ __forceinline__ double cvAtomicAdd(double* address, double val)
- {
- #if __CUDA_ARCH__ >= 130
- unsigned long long int* address_as_ull = (unsigned long long int*) address;
- unsigned long long int old = *address_as_ull, assumed;
- do {
- assumed = old;
- old = ::atomicCAS(address_as_ull, assumed,
- __double_as_longlong(val + __longlong_as_double(assumed)));
- } while (assumed != old);
- return __longlong_as_double(old);
- #else
- (void) address;
- (void) val;
- return 0.0;
- #endif
- }
-
- __device__ __forceinline__ int cvAtomicMin(int* address, int val)
- {
- return ::atomicMin(address, val);
- }
- __device__ __forceinline__ float cvAtomicMin(float* address, float val)
- {
- #if __CUDA_ARCH__ >= 120
- int* address_as_i = (int*) address;
- int old = *address_as_i, assumed;
- do {
- assumed = old;
- old = ::atomicCAS(address_as_i, assumed,
- __float_as_int(::fminf(val, __int_as_float(assumed))));
- } while (assumed != old);
- return __int_as_float(old);
- #else
- (void) address;
- (void) val;
- return 0.0f;
- #endif
- }
- __device__ __forceinline__ double cvAtomicMin(double* address, double val)
- {
- #if __CUDA_ARCH__ >= 130
- unsigned long long int* address_as_ull = (unsigned long long int*) address;
- unsigned long long int old = *address_as_ull, assumed;
- do {
- assumed = old;
- old = ::atomicCAS(address_as_ull, assumed,
- __double_as_longlong(::fmin(val, __longlong_as_double(assumed))));
- } while (assumed != old);
- return __longlong_as_double(old);
- #else
- (void) address;
- (void) val;
- return 0.0;
- #endif
- }
-
- __device__ __forceinline__ int cvAtomicMax(int* address, int val)
- {
- return ::atomicMax(address, val);
- }
- __device__ __forceinline__ float cvAtomicMax(float* address, float val)
- {
- #if __CUDA_ARCH__ >= 120
- int* address_as_i = (int*) address;
- int old = *address_as_i, assumed;
- do {
- assumed = old;
- old = ::atomicCAS(address_as_i, assumed,
- __float_as_int(::fmaxf(val, __int_as_float(assumed))));
- } while (assumed != old);
- return __int_as_float(old);
- #else
- (void) address;
- (void) val;
- return 0.0f;
- #endif
- }
- __device__ __forceinline__ double cvAtomicMax(double* address, double val)
- {
- #if __CUDA_ARCH__ >= 130
- unsigned long long int* address_as_ull = (unsigned long long int*) address;
- unsigned long long int old = *address_as_ull, assumed;
- do {
- assumed = old;
- old = ::atomicCAS(address_as_ull, assumed,
- __double_as_longlong(::fmax(val, __longlong_as_double(assumed))));
- } while (assumed != old);
- return __longlong_as_double(old);
- #else
- (void) address;
- (void) val;
- return 0.0;
- #endif
- }
-}
-
-namespace detail
-{
- template <int cn> struct Unroll;
- template <> struct Unroll<1>
- {
- template <int BLOCK_SIZE, typename R>
- static __device__ __forceinline__ volatile R* smem_tuple(R* smem)
- {
- return smem;
- }
-
- template <typename R>
- static __device__ __forceinline__ R& tie(R& val)
- {
- return val;
- }
-
- template <class Op>
- static __device__ __forceinline__ const Op& op(const Op& op)
- {
- return op;
- }
- };
- template <> struct Unroll<2>
- {
- template <int BLOCK_SIZE, typename R>
- static __device__ __forceinline__ thrust::tuple<volatile R*, volatile R*> smem_tuple(R* smem)
- {
- return cv::gpu::cudev::smem_tuple(smem, smem + BLOCK_SIZE);
- }
-
- template <typename R>
- static __device__ __forceinline__ thrust::tuple<typename VecTraits<R>::elem_type&, typename VecTraits<R>::elem_type&> tie(R& val)
- {
- return thrust::tie(val.x, val.y);
- }
-
- template <class Op>
- static __device__ __forceinline__ const thrust::tuple<Op, Op> op(const Op& op)
- {
- return thrust::make_tuple(op, op);
- }
- };
- template <> struct Unroll<3>
- {
- template <int BLOCK_SIZE, typename R>
- static __device__ __forceinline__ thrust::tuple<volatile R*, volatile R*, volatile R*> smem_tuple(R* smem)
- {
- return cv::gpu::cudev::smem_tuple(smem, smem + BLOCK_SIZE, smem + 2 * BLOCK_SIZE);
- }
-
- template <typename R>
- static __device__ __forceinline__ thrust::tuple<typename VecTraits<R>::elem_type&, typename VecTraits<R>::elem_type&, typename VecTraits<R>::elem_type&> tie(R& val)
- {
- return thrust::tie(val.x, val.y, val.z);
- }
-
- template <class Op>
- static __device__ __forceinline__ const thrust::tuple<Op, Op, Op> op(const Op& op)
- {
- return thrust::make_tuple(op, op, op);
- }
- };
- template <> struct Unroll<4>
- {
- template <int BLOCK_SIZE, typename R>
- static __device__ __forceinline__ thrust::tuple<volatile R*, volatile R*, volatile R*, volatile R*> smem_tuple(R* smem)
- {
- return cv::gpu::cudev::smem_tuple(smem, smem + BLOCK_SIZE, smem + 2 * BLOCK_SIZE, smem + 3 * BLOCK_SIZE);
- }
-
- template <typename R>
- static __device__ __forceinline__ thrust::tuple<typename VecTraits<R>::elem_type&, typename VecTraits<R>::elem_type&, typename VecTraits<R>::elem_type&, typename VecTraits<R>::elem_type&> tie(R& val)
- {
- return thrust::tie(val.x, val.y, val.z, val.w);
- }
-
- template <class Op>
- static __device__ __forceinline__ const thrust::tuple<Op, Op, Op, Op> op(const Op& op)
- {
- return thrust::make_tuple(op, op, op, op);
- }
- };
-}
-
-/////////////////////////////////////////////////////////////
-// sum
-
-namespace sum
-{
- __device__ unsigned int blocks_finished = 0;
-
- template <typename R, int cn> struct AtomicAdd;
- template <typename R> struct AtomicAdd<R, 1>
- {
- static __device__ void run(R* ptr, R val)
- {
- detail::cvAtomicAdd(ptr, val);
- }
- };
- template <typename R> struct AtomicAdd<R, 2>
- {
- typedef typename TypeVec<R, 2>::vec_type val_type;
-
- static __device__ void run(R* ptr, val_type val)
- {
- detail::cvAtomicAdd(ptr, val.x);
- detail::cvAtomicAdd(ptr + 1, val.y);
- }
- };
- template <typename R> struct AtomicAdd<R, 3>
- {
- typedef typename TypeVec<R, 3>::vec_type val_type;
-
- static __device__ void run(R* ptr, val_type val)
- {
- detail::cvAtomicAdd(ptr, val.x);
- detail::cvAtomicAdd(ptr + 1, val.y);
- detail::cvAtomicAdd(ptr + 2, val.z);
- }
- };
- template <typename R> struct AtomicAdd<R, 4>
- {
- typedef typename TypeVec<R, 4>::vec_type val_type;
-
- static __device__ void run(R* ptr, val_type val)
- {
- detail::cvAtomicAdd(ptr, val.x);
- detail::cvAtomicAdd(ptr + 1, val.y);
- detail::cvAtomicAdd(ptr + 2, val.z);
- detail::cvAtomicAdd(ptr + 3, val.w);
- }
- };
-
- template <int BLOCK_SIZE, typename R, int cn>
- struct GlobalReduce
- {
- typedef typename TypeVec<R, cn>::vec_type result_type;
-
- static __device__ void run(result_type& sum, result_type* result, int tid, int bid, R* smem)
- {
- #if __CUDA_ARCH__ >= 200
- if (tid == 0)
- AtomicAdd<R, cn>::run((R*) result, sum);
- #else
- __shared__ bool is_last;
-
- if (tid == 0)
- {
- result[bid] = sum;
-
- __threadfence();
-
- unsigned int ticket = ::atomicAdd(&blocks_finished, 1);
- is_last = (ticket == gridDim.x * gridDim.y - 1);
- }
-
- __syncthreads();
-
- if (is_last)
- {
- sum = tid < gridDim.x * gridDim.y ? result[tid] : VecTraits<result_type>::all(0);
-
- cudev::reduce<BLOCK_SIZE>(detail::Unroll<cn>::template smem_tuple<BLOCK_SIZE>(smem), detail::Unroll<cn>::tie(sum), tid, detail::Unroll<cn>::op(plus<R>()));
-
- if (tid == 0)
- {
- result[0] = sum;
- blocks_finished = 0;
- }
- }
- #endif
- }
- };
-
- template <int BLOCK_SIZE, typename src_type, typename result_type, class Mask, class Op>
- __global__ void kernel(const PtrStepSz<src_type> src, result_type* result, const Mask mask, const Op op, const int twidth, const int theight)
- {
- typedef typename VecTraits<src_type>::elem_type T;
- typedef typename VecTraits<result_type>::elem_type R;
- const int cn = VecTraits<src_type>::cn;
-
- __shared__ R smem[BLOCK_SIZE * cn];
-
- const int x0 = blockIdx.x * blockDim.x * twidth + threadIdx.x;
- const int y0 = blockIdx.y * blockDim.y * theight + threadIdx.y;
-
- const int tid = threadIdx.y * blockDim.x + threadIdx.x;
- const int bid = blockIdx.y * gridDim.x + blockIdx.x;
-
- result_type sum = VecTraits<result_type>::all(0);
-
- for (int i = 0, y = y0; i < theight && y < src.rows; ++i, y += blockDim.y)
- {
- const src_type* ptr = src.ptr(y);
-
- for (int j = 0, x = x0; j < twidth && x < src.cols; ++j, x += blockDim.x)
- {
- if (mask(y, x))
- {
- const src_type srcVal = ptr[x];
- sum = sum + op(saturate_cast<result_type>(srcVal));
- }
- }
- }
-
- cudev::reduce<BLOCK_SIZE>(detail::Unroll<cn>::template smem_tuple<BLOCK_SIZE>(smem), detail::Unroll<cn>::tie(sum), tid, detail::Unroll<cn>::op(plus<R>()));
-
- GlobalReduce<BLOCK_SIZE, R, cn>::run(sum, result, tid, bid, smem);
- }
-
- const int threads_x = 32;
- const int threads_y = 8;
-
- void getLaunchCfg(int cols, int rows, dim3& block, dim3& grid)
- {
- block = dim3(threads_x, threads_y);
-
- grid = dim3(divUp(cols, block.x * block.y),
- divUp(rows, block.y * block.x));
-
- grid.x = ::min(grid.x, block.x);
- grid.y = ::min(grid.y, block.y);
- }
-
- void getBufSize(int cols, int rows, int cn, int& bufcols, int& bufrows)
- {
- dim3 block, grid;
- getLaunchCfg(cols, rows, block, grid);
-
- bufcols = grid.x * grid.y * sizeof(double) * cn;
- bufrows = 1;
- }
-
- template <typename T, typename R, int cn, template <typename> class Op>
- void caller(PtrStepSzb src_, void* buf_, double* out, PtrStepSzb mask)
- {
- typedef typename TypeVec<T, cn>::vec_type src_type;
- typedef typename TypeVec<R, cn>::vec_type result_type;
-
- PtrStepSz<src_type> src(src_);
- result_type* buf = (result_type*) buf_;
-
- dim3 block, grid;
- getLaunchCfg(src.cols, src.rows, block, grid);
-
- const int twidth = divUp(divUp(src.cols, grid.x), block.x);
- const int theight = divUp(divUp(src.rows, grid.y), block.y);
-
- Op<result_type> op;
-
- if (mask.data)
- kernel<threads_x * threads_y><<<grid, block>>>(src, buf, SingleMask(mask), op, twidth, theight);
- else
- kernel<threads_x * threads_y><<<grid, block>>>(src, buf, WithOutMask(), op, twidth, theight);
- cudaSafeCall( cudaGetLastError() );
-
- cudaSafeCall( cudaDeviceSynchronize() );
-
- R result[4] = {0, 0, 0, 0};
- cudaSafeCall( cudaMemcpy(&result, buf, sizeof(result_type), cudaMemcpyDeviceToHost) );
-
- out[0] = result[0];
- out[1] = result[1];
- out[2] = result[2];
- out[3] = result[3];
- }
-
- template <typename T> struct SumType;
- template <> struct SumType<uchar> { typedef unsigned int R; };
- template <> struct SumType<schar> { typedef int R; };
- template <> struct SumType<ushort> { typedef unsigned int R; };
- template <> struct SumType<short> { typedef int R; };
- template <> struct SumType<int> { typedef int R; };
- template <> struct SumType<float> { typedef float R; };
- template <> struct SumType<double> { typedef double R; };
-
- template <typename T, int cn>
- void run(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask)
- {
- typedef typename SumType<T>::R R;
- caller<T, R, cn, identity>(src, buf, out, mask);
- }
-
- template void run<uchar, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<uchar, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<uchar, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<uchar, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template void run<schar, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<schar, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<schar, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<schar, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template void run<ushort, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<ushort, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<ushort, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<ushort, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template void run<short, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<short, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<short, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<short, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template void run<int, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<int, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<int, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<int, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template void run<float, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<float, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<float, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<float, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template void run<double, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<double, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<double, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void run<double, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template <typename T, int cn>
- void runAbs(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask)
- {
- typedef typename SumType<T>::R R;
- caller<T, R, cn, abs_func>(src, buf, out, mask);
- }
-
- template void runAbs<uchar, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<uchar, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<uchar, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<uchar, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template void runAbs<schar, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<schar, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<schar, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<schar, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template void runAbs<ushort, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<ushort, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<ushort, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<ushort, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template void runAbs<short, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<short, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<short, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<short, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template void runAbs<int, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<int, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<int, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<int, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template void runAbs<float, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<float, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<float, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<float, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template void runAbs<double, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<double, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<double, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runAbs<double, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template <typename T> struct Sqr : unary_function<T, T>
- {
- __device__ __forceinline__ T operator ()(T x) const
- {
- return x * x;
- }
- };
-
- template <typename T, int cn>
- void runSqr(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask)
- {
- caller<T, double, cn, Sqr>(src, buf, out, mask);
- }
-
- template void runSqr<uchar, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<uchar, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<uchar, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<uchar, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template void runSqr<schar, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<schar, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<schar, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<schar, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template void runSqr<ushort, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<ushort, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<ushort, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<ushort, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template void runSqr<short, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<short, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<short, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<short, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template void runSqr<int, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<int, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<int, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<int, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template void runSqr<float, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<float, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<float, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<float, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-
- template void runSqr<double, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<double, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<double, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
- template void runSqr<double, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
-}
-
-/////////////////////////////////////////////////////////////
-// minMax
-
-namespace minMax
-{
- __device__ unsigned int blocks_finished = 0;
-
- // To avoid shared bank conflicts we convert each value into value of
- // appropriate type (32 bits minimum)
- template <typename T> struct MinMaxTypeTraits;
- template <> struct MinMaxTypeTraits<uchar> { typedef int best_type; };
- template <> struct MinMaxTypeTraits<schar> { typedef int best_type; };
- template <> struct MinMaxTypeTraits<ushort> { typedef int best_type; };
- template <> struct MinMaxTypeTraits<short> { typedef int best_type; };
- template <> struct MinMaxTypeTraits<int> { typedef int best_type; };
- template <> struct MinMaxTypeTraits<float> { typedef float best_type; };
- template <> struct MinMaxTypeTraits<double> { typedef double best_type; };
-
- template <int BLOCK_SIZE, typename R>
- struct GlobalReduce
- {
- static __device__ void run(R& mymin, R& mymax, R* minval, R* maxval, int tid, int bid, R* sminval, R* smaxval)
- {
- #if __CUDA_ARCH__ >= 200
- if (tid == 0)
- {
- detail::cvAtomicMin(minval, mymin);
- detail::cvAtomicMax(maxval, mymax);
- }
- #else
- __shared__ bool is_last;
-
- if (tid == 0)
- {
- minval[bid] = mymin;
- maxval[bid] = mymax;
-
- __threadfence();
-
- unsigned int ticket = ::atomicAdd(&blocks_finished, 1);
- is_last = (ticket == gridDim.x * gridDim.y - 1);
- }
-
- __syncthreads();
-
- if (is_last)
- {
- int idx = ::min(tid, gridDim.x * gridDim.y - 1);
-
- mymin = minval[idx];
- mymax = maxval[idx];
-
- const minimum<R> minOp;
- const maximum<R> maxOp;
- cudev::reduce<BLOCK_SIZE>(smem_tuple(sminval, smaxval), thrust::tie(mymin, mymax), tid, thrust::make_tuple(minOp, maxOp));
-
- if (tid == 0)
- {
- minval[0] = mymin;
- maxval[0] = mymax;
-
- blocks_finished = 0;
- }
- }
- #endif
- }
- };
-
- template <int BLOCK_SIZE, typename T, typename R, class Mask>
- __global__ void kernel(const PtrStepSz<T> src, const Mask mask, R* minval, R* maxval, const int twidth, const int theight)
- {
- __shared__ R sminval[BLOCK_SIZE];
- __shared__ R smaxval[BLOCK_SIZE];
-
- const int x0 = blockIdx.x * blockDim.x * twidth + threadIdx.x;
- const int y0 = blockIdx.y * blockDim.y * theight + threadIdx.y;
-
- const int tid = threadIdx.y * blockDim.x + threadIdx.x;
- const int bid = blockIdx.y * gridDim.x + blockIdx.x;
-
- R mymin = numeric_limits<R>::max();
- R mymax = -numeric_limits<R>::max();
-
- const minimum<R> minOp;
- const maximum<R> maxOp;
-
- for (int i = 0, y = y0; i < theight && y < src.rows; ++i, y += blockDim.y)
- {
- const T* ptr = src.ptr(y);
-
- for (int j = 0, x = x0; j < twidth && x < src.cols; ++j, x += blockDim.x)
- {
- if (mask(y, x))
- {
- const R srcVal = ptr[x];
-
- mymin = minOp(mymin, srcVal);
- mymax = maxOp(mymax, srcVal);
- }
- }
- }
-
- cudev::reduce<BLOCK_SIZE>(smem_tuple(sminval, smaxval), thrust::tie(mymin, mymax), tid, thrust::make_tuple(minOp, maxOp));
-
- GlobalReduce<BLOCK_SIZE, R>::run(mymin, mymax, minval, maxval, tid, bid, sminval, smaxval);
- }
-
- const int threads_x = 32;
- const int threads_y = 8;
-
- void getLaunchCfg(int cols, int rows, dim3& block, dim3& grid)
- {
- block = dim3(threads_x, threads_y);
-
- grid = dim3(divUp(cols, block.x * block.y),
- divUp(rows, block.y * block.x));
-
- grid.x = ::min(grid.x, block.x);
- grid.y = ::min(grid.y, block.y);
- }
-
- void getBufSize(int cols, int rows, int& bufcols, int& bufrows)
- {
- dim3 block, grid;
- getLaunchCfg(cols, rows, block, grid);
-
- bufcols = grid.x * grid.y * sizeof(double);
- bufrows = 2;
- }
-
- __global__ void setDefaultKernel(int* minval_buf, int* maxval_buf)
- {
- *minval_buf = numeric_limits<int>::max();
- *maxval_buf = numeric_limits<int>::min();
- }
- __global__ void setDefaultKernel(float* minval_buf, float* maxval_buf)
- {
- *minval_buf = numeric_limits<float>::max();
- *maxval_buf = -numeric_limits<float>::max();
- }
- __global__ void setDefaultKernel(double* minval_buf, double* maxval_buf)
- {
- *minval_buf = numeric_limits<double>::max();
- *maxval_buf = -numeric_limits<double>::max();
- }
-
- template <typename R>
- void setDefault(R* minval_buf, R* maxval_buf)
- {
- setDefaultKernel<<<1, 1>>>(minval_buf, maxval_buf);
- }
-
- template <typename T>
- void run(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf)
- {
- typedef typename MinMaxTypeTraits<T>::best_type R;
-
- dim3 block, grid;
- getLaunchCfg(src.cols, src.rows, block, grid);
-
- const int twidth = divUp(divUp(src.cols, grid.x), block.x);
- const int theight = divUp(divUp(src.rows, grid.y), block.y);
-
- R* minval_buf = (R*) buf.ptr(0);
- R* maxval_buf = (R*) buf.ptr(1);
-
- setDefault(minval_buf, maxval_buf);
-
- if (mask.data)
- kernel<threads_x * threads_y><<<grid, block>>>((PtrStepSz<T>) src, SingleMask(mask), minval_buf, maxval_buf, twidth, theight);
- else
- kernel<threads_x * threads_y><<<grid, block>>>((PtrStepSz<T>) src, WithOutMask(), minval_buf, maxval_buf, twidth, theight);
-
- cudaSafeCall( cudaGetLastError() );
-
- cudaSafeCall( cudaDeviceSynchronize() );
-
- R minval_, maxval_;
- cudaSafeCall( cudaMemcpy(&minval_, minval_buf, sizeof(R), cudaMemcpyDeviceToHost) );
- cudaSafeCall( cudaMemcpy(&maxval_, maxval_buf, sizeof(R), cudaMemcpyDeviceToHost) );
- *minval = minval_;
- *maxval = maxval_;
- }
-
- template void run<uchar >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf);
- template void run<schar >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf);
- template void run<ushort>(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf);
- template void run<short >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf);
- template void run<int >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf);
- template void run<float >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf);
- template void run<double>(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf);
-}
-
-/////////////////////////////////////////////////////////////
-// minMaxLoc
-
-namespace minMaxLoc
-{
- // To avoid shared bank conflicts we convert each value into value of
- // appropriate type (32 bits minimum)
- template <typename T> struct MinMaxTypeTraits;
- template <> struct MinMaxTypeTraits<unsigned char> { typedef int best_type; };
- template <> struct MinMaxTypeTraits<signed char> { typedef int best_type; };
- template <> struct MinMaxTypeTraits<unsigned short> { typedef int best_type; };
- template <> struct MinMaxTypeTraits<short> { typedef int best_type; };
- template <> struct MinMaxTypeTraits<int> { typedef int best_type; };
- template <> struct MinMaxTypeTraits<float> { typedef float best_type; };
- template <> struct MinMaxTypeTraits<double> { typedef double best_type; };
-
- template <int BLOCK_SIZE, typename T, class Mask>
- __global__ void kernel_pass_1(const PtrStepSz<T> src, const Mask mask, T* minval, T* maxval, unsigned int* minloc, unsigned int* maxloc, const int twidth, const int theight)
- {
- typedef typename MinMaxTypeTraits<T>::best_type work_type;
-
- __shared__ work_type sminval[BLOCK_SIZE];
- __shared__ work_type smaxval[BLOCK_SIZE];
- __shared__ unsigned int sminloc[BLOCK_SIZE];
- __shared__ unsigned int smaxloc[BLOCK_SIZE];
-
- const int x0 = blockIdx.x * blockDim.x * twidth + threadIdx.x;
- const int y0 = blockIdx.y * blockDim.y * theight + threadIdx.y;
-
- const int tid = threadIdx.y * blockDim.x + threadIdx.x;
- const int bid = blockIdx.y * gridDim.x + blockIdx.x;
-
- work_type mymin = numeric_limits<work_type>::max();
- work_type mymax = -numeric_limits<work_type>::max();
- unsigned int myminloc = 0;
- unsigned int mymaxloc = 0;
-
- for (int i = 0, y = y0; i < theight && y < src.rows; ++i, y += blockDim.y)
- {
- const T* ptr = src.ptr(y);
-
- for (int j = 0, x = x0; j < twidth && x < src.cols; ++j, x += blockDim.x)
- {
- if (mask(y, x))
- {
- const work_type srcVal = ptr[x];
-
- if (srcVal < mymin)
- {
- mymin = srcVal;
- myminloc = y * src.cols + x;
- }
-
- if (srcVal > mymax)
- {
- mymax = srcVal;
- mymaxloc = y * src.cols + x;
- }
- }
- }
- }
-
- reduceKeyVal<BLOCK_SIZE>(smem_tuple(sminval, smaxval), thrust::tie(mymin, mymax),
- smem_tuple(sminloc, smaxloc), thrust::tie(myminloc, mymaxloc),
- tid,
- thrust::make_tuple(less<work_type>(), greater<work_type>()));
-
- if (tid == 0)
- {
- minval[bid] = (T) mymin;
- maxval[bid] = (T) mymax;
- minloc[bid] = myminloc;
- maxloc[bid] = mymaxloc;
- }
- }
- template <int BLOCK_SIZE, typename T>
- __global__ void kernel_pass_2(T* minval, T* maxval, unsigned int* minloc, unsigned int* maxloc, int count)
- {
- typedef typename MinMaxTypeTraits<T>::best_type work_type;
-
- __shared__ work_type sminval[BLOCK_SIZE];
- __shared__ work_type smaxval[BLOCK_SIZE];
- __shared__ unsigned int sminloc[BLOCK_SIZE];
- __shared__ unsigned int smaxloc[BLOCK_SIZE];
-
- unsigned int idx = ::min(threadIdx.x, count - 1);
-
- work_type mymin = minval[idx];
- work_type mymax = maxval[idx];
- unsigned int myminloc = minloc[idx];
- unsigned int mymaxloc = maxloc[idx];
-
- reduceKeyVal<BLOCK_SIZE>(smem_tuple(sminval, smaxval), thrust::tie(mymin, mymax),
- smem_tuple(sminloc, smaxloc), thrust::tie(myminloc, mymaxloc),
- threadIdx.x,
- thrust::make_tuple(less<work_type>(), greater<work_type>()));
-
- if (threadIdx.x == 0)
- {
- minval[0] = (T) mymin;
- maxval[0] = (T) mymax;
- minloc[0] = myminloc;
- maxloc[0] = mymaxloc;
- }
- }
-
- const int threads_x = 32;
- const int threads_y = 8;
-
- void getLaunchCfg(int cols, int rows, dim3& block, dim3& grid)
- {
- block = dim3(threads_x, threads_y);
-
- grid = dim3(divUp(cols, block.x * block.y),
- divUp(rows, block.y * block.x));
-
- grid.x = ::min(grid.x, block.x);
- grid.y = ::min(grid.y, block.y);
- }
-
- void getBufSize(int cols, int rows, size_t elem_size, int& b1cols, int& b1rows, int& b2cols, int& b2rows)
- {
- dim3 block, grid;
- getLaunchCfg(cols, rows, block, grid);
-
- // For values
- b1cols = (int)(grid.x * grid.y * elem_size);
- b1rows = 2;
-
- // For locations
- b2cols = grid.x * grid.y * sizeof(int);
- b2rows = 2;
- }
-
- template <typename T>
- void run(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, int* minloc, int* maxloc, PtrStepb valbuf, PtrStep<unsigned int> locbuf)
- {
- dim3 block, grid;
- getLaunchCfg(src.cols, src.rows, block, grid);
-
- const int twidth = divUp(divUp(src.cols, grid.x), block.x);
- const int theight = divUp(divUp(src.rows, grid.y), block.y);
-
- T* minval_buf = (T*) valbuf.ptr(0);
- T* maxval_buf = (T*) valbuf.ptr(1);
- unsigned int* minloc_buf = locbuf.ptr(0);
- unsigned int* maxloc_buf = locbuf.ptr(1);
-
- if (mask.data)
- kernel_pass_1<threads_x * threads_y><<<grid, block>>>((PtrStepSz<T>) src, SingleMask(mask), minval_buf, maxval_buf, minloc_buf, maxloc_buf, twidth, theight);
- else
- kernel_pass_1<threads_x * threads_y><<<grid, block>>>((PtrStepSz<T>) src, WithOutMask(), minval_buf, maxval_buf, minloc_buf, maxloc_buf, twidth, theight);
-
- cudaSafeCall( cudaGetLastError() );
-
- kernel_pass_2<threads_x * threads_y><<<1, threads_x * threads_y>>>(minval_buf, maxval_buf, minloc_buf, maxloc_buf, grid.x * grid.y);
- cudaSafeCall( cudaGetLastError() );
-
- cudaSafeCall( cudaDeviceSynchronize() );
-
- T minval_, maxval_;
- cudaSafeCall( cudaMemcpy(&minval_, minval_buf, sizeof(T), cudaMemcpyDeviceToHost) );
- cudaSafeCall( cudaMemcpy(&maxval_, maxval_buf, sizeof(T), cudaMemcpyDeviceToHost) );
- *minval = minval_;
- *maxval = maxval_;
-
- unsigned int minloc_, maxloc_;
- cudaSafeCall( cudaMemcpy(&minloc_, minloc_buf, sizeof(unsigned int), cudaMemcpyDeviceToHost) );
- cudaSafeCall( cudaMemcpy(&maxloc_, maxloc_buf, sizeof(unsigned int), cudaMemcpyDeviceToHost) );
- minloc[1] = minloc_ / src.cols; minloc[0] = minloc_ - minloc[1] * src.cols;
- maxloc[1] = maxloc_ / src.cols; maxloc[0] = maxloc_ - maxloc[1] * src.cols;
- }
-
- template void run<unsigned char >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, int* minloc, int* maxloc, PtrStepb valbuf, PtrStep<unsigned int> locbuf);
- template void run<signed char >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, int* minloc, int* maxloc, PtrStepb valbuf, PtrStep<unsigned int> locbuf);
- template void run<unsigned short>(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, int* minloc, int* maxloc, PtrStepb valbuf, PtrStep<unsigned int> locbuf);
- template void run<short >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, int* minloc, int* maxloc, PtrStepb valbuf, PtrStep<unsigned int> locbuf);
- template void run<int >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, int* minloc, int* maxloc, PtrStepb valbuf, PtrStep<unsigned int> locbuf);
- template void run<float >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, int* minloc, int* maxloc, PtrStepb valbuf, PtrStep<unsigned int> locbuf);
- template void run<double>(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, int* minloc, int* maxloc, PtrStepb valbuf, PtrStep<unsigned int> locbuf);
-}
-
-/////////////////////////////////////////////////////////////
-// countNonZero
-
-namespace countNonZero
-{
- __device__ unsigned int blocks_finished = 0;
-
- template <int BLOCK_SIZE, typename T>
- __global__ void kernel(const PtrStepSz<T> src, unsigned int* count, const int twidth, const int theight)
- {
- __shared__ unsigned int scount[BLOCK_SIZE];
-
- const int x0 = blockIdx.x * blockDim.x * twidth + threadIdx.x;
- const int y0 = blockIdx.y * blockDim.y * theight + threadIdx.y;
-
- const int tid = threadIdx.y * blockDim.x + threadIdx.x;
-
- unsigned int mycount = 0;
-
- for (int i = 0, y = y0; i < theight && y < src.rows; ++i, y += blockDim.y)
- {
- const T* ptr = src.ptr(y);
-
- for (int j = 0, x = x0; j < twidth && x < src.cols; ++j, x += blockDim.x)
- {
- const T srcVal = ptr[x];
-
- mycount += (srcVal != 0);
- }
- }
-
- cudev::reduce<BLOCK_SIZE>(scount, mycount, tid, plus<unsigned int>());
-
- #if __CUDA_ARCH__ >= 200
- if (tid == 0)
- ::atomicAdd(count, mycount);
- #else
- __shared__ bool is_last;
- const int bid = blockIdx.y * gridDim.x + blockIdx.x;
-
- if (tid == 0)
- {
- count[bid] = mycount;
-
- __threadfence();
-
- unsigned int ticket = ::atomicInc(&blocks_finished, gridDim.x * gridDim.y);
- is_last = (ticket == gridDim.x * gridDim.y - 1);
- }
-
- __syncthreads();
-
- if (is_last)
- {
- mycount = tid < gridDim.x * gridDim.y ? count[tid] : 0;
-
- cudev::reduce<BLOCK_SIZE>(scount, mycount, tid, plus<unsigned int>());
-
- if (tid == 0)
- {
- count[0] = mycount;
-
- blocks_finished = 0;
- }
- }
- #endif
- }
-
- const int threads_x = 32;
- const int threads_y = 8;
-
- void getLaunchCfg(int cols, int rows, dim3& block, dim3& grid)
- {
- block = dim3(threads_x, threads_y);
-
- grid = dim3(divUp(cols, block.x * block.y),
- divUp(rows, block.y * block.x));
-
- grid.x = ::min(grid.x, block.x);
- grid.y = ::min(grid.y, block.y);
- }
-
- void getBufSize(int cols, int rows, int& bufcols, int& bufrows)
- {
- dim3 block, grid;
- getLaunchCfg(cols, rows, block, grid);
-
- bufcols = grid.x * grid.y * sizeof(int);
- bufrows = 1;
- }
-
- template <typename T>
- int run(const PtrStepSzb src, PtrStep<unsigned int> buf)
- {
- dim3 block, grid;
- getLaunchCfg(src.cols, src.rows, block, grid);
-
- const int twidth = divUp(divUp(src.cols, grid.x), block.x);
- const int theight = divUp(divUp(src.rows, grid.y), block.y);
-
- unsigned int* count_buf = buf.ptr(0);
-
- cudaSafeCall( cudaMemset(count_buf, 0, sizeof(unsigned int)) );
-
- kernel<threads_x * threads_y><<<grid, block>>>((PtrStepSz<T>) src, count_buf, twidth, theight);
- cudaSafeCall( cudaGetLastError() );
-
- cudaSafeCall( cudaDeviceSynchronize() );
-
- unsigned int count;
- cudaSafeCall(cudaMemcpy(&count, count_buf, sizeof(unsigned int), cudaMemcpyDeviceToHost));
-
- return count;
- }
-
- template int run<uchar >(const PtrStepSzb src, PtrStep<unsigned int> buf);
- template int run<schar >(const PtrStepSzb src, PtrStep<unsigned int> buf);
- template int run<ushort>(const PtrStepSzb src, PtrStep<unsigned int> buf);
- template int run<short >(const PtrStepSzb src, PtrStep<unsigned int> buf);
- template int run<int >(const PtrStepSzb src, PtrStep<unsigned int> buf);
- template int run<float >(const PtrStepSzb src, PtrStep<unsigned int> buf);
- template int run<double>(const PtrStepSzb src, PtrStep<unsigned int> buf);
-}
-
-//////////////////////////////////////////////////////////////////////////////
-// reduce
-
-namespace reduce
-{
- struct Sum
- {
- template <typename T>
- __device__ __forceinline__ T startValue() const
- {
- return VecTraits<T>::all(0);
- }
-
- template <typename T>
- __device__ __forceinline__ T operator ()(T a, T b) const
- {
- return a + b;
- }
-
- template <typename T>
- __device__ __forceinline__ T result(T r, double) const
- {
- return r;
- }
-
- __device__ __forceinline__ Sum() {}
- __device__ __forceinline__ Sum(const Sum&) {}
- };
-
- struct Avg
- {
- template <typename T>
- __device__ __forceinline__ T startValue() const
- {
- return VecTraits<T>::all(0);
- }
-
- template <typename T>
- __device__ __forceinline__ T operator ()(T a, T b) const
- {
- return a + b;
- }
-
- template <typename T>
- __device__ __forceinline__ typename TypeVec<double, VecTraits<T>::cn>::vec_type result(T r, double sz) const
- {
- return r / sz;
- }
-
- __device__ __forceinline__ Avg() {}
- __device__ __forceinline__ Avg(const Avg&) {}
- };
-
- struct Min
- {
- template <typename T>
- __device__ __forceinline__ T startValue() const
- {
- return VecTraits<T>::all(numeric_limits<typename VecTraits<T>::elem_type>::max());
- }
-
- template <typename T>
- __device__ __forceinline__ T operator ()(T a, T b) const
- {
- minimum<T> minOp;
- return minOp(a, b);
- }
-
- template <typename T>
- __device__ __forceinline__ T result(T r, double) const
- {
- return r;
- }
-
- __device__ __forceinline__ Min() {}
- __device__ __forceinline__ Min(const Min&) {}
- };
-
- struct Max
- {
- template <typename T>
- __device__ __forceinline__ T startValue() const
- {
- return VecTraits<T>::all(-numeric_limits<typename VecTraits<T>::elem_type>::max());
- }
-
- template <typename T>
- __device__ __forceinline__ T operator ()(T a, T b) const
- {
- maximum<T> maxOp;
- return maxOp(a, b);
- }
-
- template <typename T>
- __device__ __forceinline__ T result(T r, double) const
- {
- return r;
- }
-
- __device__ __forceinline__ Max() {}
- __device__ __forceinline__ Max(const Max&) {}
- };
-
- ///////////////////////////////////////////////////////////
-
- template <typename T, typename S, typename D, class Op>
- __global__ void rowsKernel(const PtrStepSz<T> src, D* dst, const Op op)
- {
- __shared__ S smem[16 * 16];
-
- const int x = blockIdx.x * 16 + threadIdx.x;
-
- S myVal = op.template startValue<S>();
-
- if (x < src.cols)
- {
- for (int y = threadIdx.y; y < src.rows; y += 16)
- {
- S srcVal = src(y, x);
- myVal = op(myVal, srcVal);
- }
- }
-
- smem[threadIdx.x * 16 + threadIdx.y] = myVal;
-
- __syncthreads();
-
- volatile S* srow = smem + threadIdx.y * 16;
-
- myVal = srow[threadIdx.x];
- cudev::reduce<16>(srow, myVal, threadIdx.x, op);
-
- if (threadIdx.x == 0)
- srow[0] = myVal;
-
- __syncthreads();
-
- if (threadIdx.y == 0 && x < src.cols)
- dst[x] = (D) op.result(smem[threadIdx.x * 16], src.rows);
- }
-
- template <typename T, typename S, typename D, class Op>
- void rowsCaller(PtrStepSz<T> src, D* dst, cudaStream_t stream)
- {
- const dim3 block(16, 16);
- const dim3 grid(divUp(src.cols, block.x));
-
- Op op;
- rowsKernel<T, S, D, Op><<<grid, block, 0, stream>>>(src, dst, op);
- cudaSafeCall( cudaGetLastError() );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
-
- template <typename T, typename S, typename D>
- void rows(PtrStepSzb src, void* dst, int op, cudaStream_t stream)
- {
- typedef void (*func_t)(PtrStepSz<T> src, D* dst, cudaStream_t stream);
- static const func_t funcs[] =
- {
- rowsCaller<T, S, D, Sum>,
- rowsCaller<T, S, D, Avg>,
- rowsCaller<T, S, D, Max>,
- rowsCaller<T, S, D, Min>
- };
-
- funcs[op]((PtrStepSz<T>) src, (D*) dst, stream);
- }
-
- template void rows<unsigned char, int, unsigned char>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
- template void rows<unsigned char, int, int>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
- template void rows<unsigned char, float, float>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
- template void rows<unsigned char, double, double>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
-
- template void rows<unsigned short, int, unsigned short>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
- template void rows<unsigned short, int, int>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
- template void rows<unsigned short, float, float>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
- template void rows<unsigned short, double, double>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
-
- template void rows<short, int, short>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
- template void rows<short, int, int>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
- template void rows<short, float, float>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
- template void rows<short, double, double>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
-
- template void rows<int, int, int>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
- template void rows<int, float, float>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
- template void rows<int, double, double>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
-
- template void rows<float, float, float>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
- template void rows<float, double, double>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
-
- template void rows<double, double, double>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
-
- ///////////////////////////////////////////////////////////
-
- template <int BLOCK_SIZE, typename T, typename S, typename D, int cn, class Op>
- __global__ void colsKernel(const PtrStepSz<typename TypeVec<T, cn>::vec_type> src, typename TypeVec<D, cn>::vec_type* dst, const Op op)
- {
- typedef typename TypeVec<T, cn>::vec_type src_type;
- typedef typename TypeVec<S, cn>::vec_type work_type;
- typedef typename TypeVec<D, cn>::vec_type dst_type;
-
- __shared__ S smem[BLOCK_SIZE * cn];
-
- const int y = blockIdx.x;
-
- const src_type* srcRow = src.ptr(y);
-
- work_type myVal = op.template startValue<work_type>();
-
- for (int x = threadIdx.x; x < src.cols; x += BLOCK_SIZE)
- myVal = op(myVal, saturate_cast<work_type>(srcRow[x]));
-
- cudev::reduce<BLOCK_SIZE>(detail::Unroll<cn>::template smem_tuple<BLOCK_SIZE>(smem), detail::Unroll<cn>::tie(myVal), threadIdx.x, detail::Unroll<cn>::op(op));
-
- if (threadIdx.x == 0)
- dst[y] = saturate_cast<dst_type>(op.result(myVal, src.cols));
- }
-
- template <typename T, typename S, typename D, int cn, class Op> void colsCaller(PtrStepSzb src, void* dst, cudaStream_t stream)
- {
- const int BLOCK_SIZE = 256;
-
- const dim3 block(BLOCK_SIZE);
- const dim3 grid(src.rows);
-
- Op op;
- colsKernel<BLOCK_SIZE, T, S, D, cn, Op><<<grid, block, 0, stream>>>((PtrStepSz<typename TypeVec<T, cn>::vec_type>) src, (typename TypeVec<D, cn>::vec_type*) dst, op);
- cudaSafeCall( cudaGetLastError() );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
-
- }
-
- template <typename T, typename S, typename D> void cols(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream)
- {
- typedef void (*func_t)(PtrStepSzb src, void* dst, cudaStream_t stream);
- static const func_t funcs[5][4] =
- {
- {0,0,0,0},
- {colsCaller<T, S, D, 1, Sum>, colsCaller<T, S, D, 1, Avg>, colsCaller<T, S, D, 1, Max>, colsCaller<T, S, D, 1, Min>},
- {colsCaller<T, S, D, 2, Sum>, colsCaller<T, S, D, 2, Avg>, colsCaller<T, S, D, 2, Max>, colsCaller<T, S, D, 2, Min>},
- {colsCaller<T, S, D, 3, Sum>, colsCaller<T, S, D, 3, Avg>, colsCaller<T, S, D, 3, Max>, colsCaller<T, S, D, 3, Min>},
- {colsCaller<T, S, D, 4, Sum>, colsCaller<T, S, D, 4, Avg>, colsCaller<T, S, D, 4, Max>, colsCaller<T, S, D, 4, Min>},
- };
-
- funcs[cn][op](src, dst, stream);
- }
-
- template void cols<unsigned char, int, unsigned char>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
- template void cols<unsigned char, int, int>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
- template void cols<unsigned char, float, float>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
- template void cols<unsigned char, double, double>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
-
- template void cols<unsigned short, int, unsigned short>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
- template void cols<unsigned short, int, int>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
- template void cols<unsigned short, float, float>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
- template void cols<unsigned short, double, double>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
-
- template void cols<short, int, short>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
- template void cols<short, int, int>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
- template void cols<short, float, float>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
- template void cols<short, double, double>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
-
- template void cols<int, int, int>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
- template void cols<int, float, float>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
- template void cols<int, double, double>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
-
- template void cols<float, float, float>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
- template void cols<float, double, double>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
-
- template void cols<double, double, double>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
-}
-
-#endif /* CUDA_DISABLER */
+++ /dev/null
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-// By downloading, copying, installing or using the software you agree to this license.
-// If you do not agree to this license, do not download, install,
-// copy or use the software.
-//
-//
-// License Agreement
-// For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-// * Redistribution's of source code must retain the above copyright notice,
-// this list of conditions and the following disclaimer.
-//
-// * Redistribution's in binary form must reproduce the above copyright notice,
-// this list of conditions and the following disclaimer in the documentation
-// and/or other materials provided with the distribution.
-//
-// * The name of the copyright holders may not be used to endorse or promote products
-// derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#if !defined CUDA_DISABLER
-
-#include "opencv2/core/cuda/common.hpp"
-#include "opencv2/core/cuda/vec_traits.hpp"
-
-namespace cv { namespace gpu { namespace cudev
-{
- namespace video_encoding
- {
- __device__ __forceinline__ void rgbtoy(const uchar b, const uchar g, const uchar r, uchar& y)
- {
- y = static_cast<uchar>(((int)(30 * r) + (int)(59 * g) + (int)(11 * b)) / 100);
- }
-
- __device__ __forceinline__ void rgbtoyuv(const uchar b, const uchar g, const uchar r, uchar& y, uchar& u, uchar& v)
- {
- rgbtoy(b, g, r, y);
- u = static_cast<uchar>(((int)(-17 * r) - (int)(33 * g) + (int)(50 * b) + 12800) / 100);
- v = static_cast<uchar>(((int)(50 * r) - (int)(42 * g) - (int)(8 * b) + 12800) / 100);
- }
-
- __global__ void Gray_to_YV12(const PtrStepSzb src, PtrStepb dst)
- {
- const int x = (blockIdx.x * blockDim.x + threadIdx.x) * 2;
- const int y = (blockIdx.y * blockDim.y + threadIdx.y) * 2;
-
- if (x + 1 >= src.cols || y + 1 >= src.rows)
- return;
-
- // get pointers to the data
- const size_t planeSize = src.rows * dst.step;
- PtrStepb y_plane(dst.data, dst.step);
- PtrStepb u_plane(y_plane.data + planeSize, dst.step / 2);
- PtrStepb v_plane(u_plane.data + (planeSize / 4), dst.step / 2);
-
- uchar pix;
- uchar y_val, u_val, v_val;
-
- pix = src(y, x);
- rgbtoy(pix, pix, pix, y_val);
- y_plane(y, x) = y_val;
-
- pix = src(y, x + 1);
- rgbtoy(pix, pix, pix, y_val);
- y_plane(y, x + 1) = y_val;
-
- pix = src(y + 1, x);
- rgbtoy(pix, pix, pix, y_val);
- y_plane(y + 1, x) = y_val;
-
- pix = src(y + 1, x + 1);
- rgbtoyuv(pix, pix, pix, y_val, u_val, v_val);
- y_plane(y + 1, x + 1) = y_val;
- u_plane(y / 2, x / 2) = u_val;
- v_plane(y / 2, x / 2) = v_val;
- }
-
- template <typename T>
- __global__ void BGR_to_YV12(const PtrStepSz<T> src, PtrStepb dst)
- {
- const int x = (blockIdx.x * blockDim.x + threadIdx.x) * 2;
- const int y = (blockIdx.y * blockDim.y + threadIdx.y) * 2;
-
- if (x + 1 >= src.cols || y + 1 >= src.rows)
- return;
-
- // get pointers to the data
- const size_t planeSize = src.rows * dst.step;
- PtrStepb y_plane(dst.data, dst.step);
- PtrStepb u_plane(y_plane.data + planeSize, dst.step / 2);
- PtrStepb v_plane(u_plane.data + (planeSize / 4), dst.step / 2);
-
- T pix;
- uchar y_val, u_val, v_val;
-
- pix = src(y, x);
- rgbtoy(pix.z, pix.y, pix.x, y_val);
- y_plane(y, x) = y_val;
-
- pix = src(y, x + 1);
- rgbtoy(pix.z, pix.y, pix.x, y_val);
- y_plane(y, x + 1) = y_val;
-
- pix = src(y + 1, x);
- rgbtoy(pix.z, pix.y, pix.x, y_val);
- y_plane(y + 1, x) = y_val;
-
- pix = src(y + 1, x + 1);
- rgbtoyuv(pix.z, pix.y, pix.x, y_val, u_val, v_val);
- y_plane(y + 1, x + 1) = y_val;
- u_plane(y / 2, x / 2) = u_val;
- v_plane(y / 2, x / 2) = v_val;
- }
-
- void Gray_to_YV12_caller(const PtrStepSzb src, PtrStepb dst)
- {
- dim3 block(32, 8);
- dim3 grid(divUp(src.cols, block.x * 2), divUp(src.rows, block.y * 2));
-
- Gray_to_YV12<<<grid, block>>>(src, dst);
- cudaSafeCall( cudaGetLastError() );
-
- cudaSafeCall( cudaDeviceSynchronize() );
- }
- template <int cn>
- void BGR_to_YV12_caller(const PtrStepSzb src, PtrStepb dst)
- {
- typedef typename TypeVec<uchar, cn>::vec_type src_t;
-
- dim3 block(32, 8);
- dim3 grid(divUp(src.cols, block.x * 2), divUp(src.rows, block.y * 2));
-
- BGR_to_YV12<<<grid, block>>>(static_cast< PtrStepSz<src_t> >(src), dst);
- cudaSafeCall( cudaGetLastError() );
-
- cudaSafeCall( cudaDeviceSynchronize() );
- }
-
- void YV12_gpu(const PtrStepSzb src, int cn, PtrStepSzb dst)
- {
- typedef void (*func_t)(const PtrStepSzb src, PtrStepb dst);
-
- static const func_t funcs[] =
- {
- 0, Gray_to_YV12_caller, 0, BGR_to_YV12_caller<3>, BGR_to_YV12_caller<4>
- };
-
- funcs[cn](src, dst);
- }
- }
-}}}
-
-#endif /* CUDA_DISABLER */
+++ /dev/null
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-// By downloading, copying, installing or using the software you agree to this license.
-// If you do not agree to this license, do not download, install,
-// copy or use the software.
-//
-//
-// License Agreement
-// For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-// * Redistribution's of source code must retain the above copyright notice,
-// this list of conditions and the following disclaimer.
-//
-// * Redistribution's in binary form must reproduce the above copyright notice,
-// this list of conditions and the following disclaimer in the documentation
-// and/or other materials provided with the distribution.
-//
-// * The name of the copyright holders may not be used to endorse or promote products
-// derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#ifndef OPENCV_GPU_TEXTURE_BINDER_HPP_
-#define OPENCV_GPU_TEXTURE_BINDER_HPP_
-
-#include "opencv2/gpu/devmem2d.hpp"
-#include <safe_call.hpp>
-
-namespace cv
-{
- namespace gpu
- {
- class TextureBinder
- {
- public:
- template<class T, enum cudaTextureReadMode readMode>
- TextureBinder(const PtrStepSz<T>& arr, const struct texture<T, 2, readMode>& tex) : texref(&tex)
- {
- cudaChannelFormatDesc desc = cudaCreateChannelDesc<T>();
- cudaSafeCall( cudaBindTexture2D(0, tex, arr.data, desc, arr.cols, arr.rows, arr.step) );
- }
-
- template<class T, enum cudaTextureReadMode readMode>
- TextureBinder(const PtrSz<T>& arr, const struct texture<T, 1, readMode> &tex) : texref(&tex)
- {
- cudaChannelFormatDesc desc = cudaCreateChannelDesc<T>();
- cudaSafeCall( cudaBindTexture(0, tex, arr.data, desc, arr.size * arr.elemSize()) );
- }
-
- template<class A, class T, enum cudaTextureReadMode readMode>
- TextureBinder(const A& arr, const struct texture<T, 2, readMode>& tex, const cudaChannelFormatDesc& desc) : texref(&tex)
- {
- cudaSafeCall( cudaBindTexture2D(0, tex, arr.data, desc, arr.cols, arr.rows, arr.step) );
- }
-
-
- ~TextureBinder()
- {
- cudaSafeCall( cudaUnbindTexture(texref) );
- }
- private:
- const struct textureReference *texref;
- };
- }
-
- namespace cuda
- {
- using cv::gpu::TextureBinder;
- }
-}
-
-#endif /* OPENCV_GPU_TEXTURE_BINDER_HPP_*/
+++ /dev/null
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-// By downloading, copying, installing or using the software you agree to this license.
-// If you do not agree to this license, do not download, install,
-// copy or use the software.
-//
-//
-// License Agreement
-// For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-// * Redistribution's of source code must retain the above copyright notice,
-// this list of conditions and the following disclaimer.
-//
-// * Redistribution's in binary form must reproduce the above copyright notice,
-// this list of conditions and the following disclaimer in the documentation
-// and/or other materials provided with the distribution.
-//
-// * The name of the copyright holders may not be used to endorse or promote products
-// derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#ifndef __CUVUD_VIDEO_SOURCE_H__
-#define __CUVUD_VIDEO_SOURCE_H__
-
-#include "precomp.hpp"
-
-#if defined(HAVE_CUDA) && defined(HAVE_NVCUVID)
-
-namespace cv { namespace gpu
-{
- namespace detail
- {
- class CuvidVideoSource : public VideoReader_GPU::VideoSource
- {
- public:
- explicit CuvidVideoSource(const String& fname);
- ~CuvidVideoSource() { cuvidDestroyVideoSource(videoSource_); }
-
- VideoReader_GPU::FormatInfo format() const;
- void start();
- void stop();
- bool isStarted() const;
- bool hasError() const;
-
- private:
- CuvidVideoSource(const CuvidVideoSource&);
- CuvidVideoSource& operator =(const CuvidVideoSource&);
-
- // Callback for handling packages of demuxed video data.
- //
- // Parameters:
- // pUserData - Pointer to user data. We must pass a pointer to a
- // VideoSourceData struct here, that contains a valid CUvideoparser
- // and FrameQueue.
- // pPacket - video-source data packet.
- //
- // NOTE: called from a different thread that doesn't not have a cuda context
- //
- static int CUDAAPI HandleVideoData(void* pUserData, CUVIDSOURCEDATAPACKET* pPacket);
-
- CUvideosource videoSource_;
- VideoReader_GPU::FormatInfo format_;
- };
- }
-}}
-
-#endif // defined(HAVE_CUDA) && !defined(__APPLE__)
-
-#endif // __CUVUD_VIDEO_SOURCE_H__
+++ /dev/null
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-// By downloading, copying, installing or using the software you agree to this license.
-// If you do not agree to this license, do not download, install,
-// copy or use the software.
-//
-//
-// License Agreement
-// For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-// * Redistribution's of source code must retain the above copyright notice,
-// this list of conditions and the following disclaimer.
-//
-// * Redistribution's in binary form must reproduce the above copyright notice,
-// this list of conditions and the following disclaimer in the documentation
-// and/or other materials provided with the distribution.
-//
-// * The name of the copyright holders may not be used to endorse or promote products
-// derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-#include "precomp.hpp"
-
-using namespace cv;
-using namespace cv::gpu;
-
-#ifdef HAVE_CUDA
-
-namespace
-{
- #define error_entry(entry) { entry, #entry }
-
- struct ErrorEntry
- {
- int code;
- const char* str;
- };
-
- struct ErrorEntryComparer
- {
- int code;
- ErrorEntryComparer(int code_) : code(code_) {}
- bool operator()(const ErrorEntry& e) const { return e.code == code; }
- };
-
- String getErrorString(int code, const ErrorEntry* errors, size_t n)
- {
- size_t idx = std::find_if(errors, errors + n, ErrorEntryComparer(code)) - errors;
-
- const char* msg = (idx != n) ? errors[idx].str : "Unknown error code";
- String str = cv::format("%s [Code = %d]", msg, code);
-
- return str;
- }
-
- //////////////////////////////////////////////////////////////////////////
- // NCV errors
-
- const ErrorEntry ncv_errors [] =
- {
- error_entry( NCV_SUCCESS ),
- error_entry( NCV_UNKNOWN_ERROR ),
- error_entry( NCV_CUDA_ERROR ),
- error_entry( NCV_NPP_ERROR ),
- error_entry( NCV_FILE_ERROR ),
- error_entry( NCV_NULL_PTR ),
- error_entry( NCV_INCONSISTENT_INPUT ),
- error_entry( NCV_TEXTURE_BIND_ERROR ),
- error_entry( NCV_DIMENSIONS_INVALID ),
- error_entry( NCV_INVALID_ROI ),
- error_entry( NCV_INVALID_STEP ),
- error_entry( NCV_INVALID_SCALE ),
- error_entry( NCV_INVALID_SCALE ),
- error_entry( NCV_ALLOCATOR_NOT_INITIALIZED ),
- error_entry( NCV_ALLOCATOR_BAD_ALLOC ),
- error_entry( NCV_ALLOCATOR_BAD_DEALLOC ),
- error_entry( NCV_ALLOCATOR_INSUFFICIENT_CAPACITY ),
- error_entry( NCV_ALLOCATOR_DEALLOC_ORDER ),
- error_entry( NCV_ALLOCATOR_BAD_REUSE ),
- error_entry( NCV_MEM_COPY_ERROR ),
- error_entry( NCV_MEM_RESIDENCE_ERROR ),
- error_entry( NCV_MEM_INSUFFICIENT_CAPACITY ),
- error_entry( NCV_HAAR_INVALID_PIXEL_STEP ),
- error_entry( NCV_HAAR_TOO_MANY_FEATURES_IN_CLASSIFIER ),
- error_entry( NCV_HAAR_TOO_MANY_FEATURES_IN_CASCADE ),
- error_entry( NCV_HAAR_TOO_LARGE_FEATURES ),
- error_entry( NCV_HAAR_XML_LOADING_EXCEPTION ),
- error_entry( NCV_NOIMPL_HAAR_TILTED_FEATURES ),
- error_entry( NCV_WARNING_HAAR_DETECTIONS_VECTOR_OVERFLOW ),
- error_entry( NPPST_SUCCESS ),
- error_entry( NPPST_ERROR ),
- error_entry( NPPST_CUDA_KERNEL_EXECUTION_ERROR ),
- error_entry( NPPST_NULL_POINTER_ERROR ),
- error_entry( NPPST_TEXTURE_BIND_ERROR ),
- error_entry( NPPST_MEMCPY_ERROR ),
- error_entry( NPPST_MEM_ALLOC_ERR ),
- error_entry( NPPST_MEMFREE_ERR ),
- error_entry( NPPST_INVALID_ROI ),
- error_entry( NPPST_INVALID_STEP ),
- error_entry( NPPST_INVALID_SCALE ),
- error_entry( NPPST_MEM_INSUFFICIENT_BUFFER ),
- error_entry( NPPST_MEM_RESIDENCE_ERROR ),
- error_entry( NPPST_MEM_INTERNAL_ERROR )
- };
-
- const size_t ncv_error_num = sizeof(ncv_errors) / sizeof(ncv_errors[0]);
-
- //////////////////////////////////////////////////////////////////////////
- // CUFFT errors
-
- const ErrorEntry cufft_errors[] =
- {
- error_entry( CUFFT_INVALID_PLAN ),
- error_entry( CUFFT_ALLOC_FAILED ),
- error_entry( CUFFT_INVALID_TYPE ),
- error_entry( CUFFT_INVALID_VALUE ),
- error_entry( CUFFT_INTERNAL_ERROR ),
- error_entry( CUFFT_EXEC_FAILED ),
- error_entry( CUFFT_SETUP_FAILED ),
- error_entry( CUFFT_INVALID_SIZE ),
- error_entry( CUFFT_UNALIGNED_DATA )
- };
-
- const int cufft_error_num = sizeof(cufft_errors) / sizeof(cufft_errors[0]);
-
- //////////////////////////////////////////////////////////////////////////
- // CUBLAS errors
-
- const ErrorEntry cublas_errors[] =
- {
- error_entry( CUBLAS_STATUS_SUCCESS ),
- error_entry( CUBLAS_STATUS_NOT_INITIALIZED ),
- error_entry( CUBLAS_STATUS_ALLOC_FAILED ),
- error_entry( CUBLAS_STATUS_INVALID_VALUE ),
- error_entry( CUBLAS_STATUS_ARCH_MISMATCH ),
- error_entry( CUBLAS_STATUS_MAPPING_ERROR ),
- error_entry( CUBLAS_STATUS_EXECUTION_FAILED ),
- error_entry( CUBLAS_STATUS_INTERNAL_ERROR )
- };
-
- const int cublas_error_num = sizeof(cublas_errors) / sizeof(cublas_errors[0]);
-}
-
-namespace cv
-{
- namespace gpu
- {
- void ncvError(int code, const char* file, const int line, const char* func)
- {
- String msg = getErrorString(code, ncv_errors, ncv_error_num);
- cv::error(cv::Error::GpuApiCallError, msg, func, file, line);
- }
-
- void cufftError(int code, const char* file, const int line, const char* func)
- {
- String msg = getErrorString(code, cufft_errors, cufft_error_num);
- cv::error(cv::Error::GpuApiCallError, msg, func, file, line);
- }
-
- void cublasError(int code, const char* file, const int line, const char* func)
- {
- String msg = getErrorString(code, cublas_errors, cublas_error_num);
- cv::error(cv::Error::GpuApiCallError, msg, func, file, line);
- }
- }
-}
-
-#endif
+++ /dev/null
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-// By downloading, copying, installing or using the software you agree to this license.
-// If you do not agree to this license, do not download, install,
-// copy or use the software.
-//
-//
-// License Agreement
-// For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-// * Redistribution's of source code must retain the above copyright notice,
-// this list of conditions and the following disclaimer.
-//
-// * Redistribution's in binary form must reproduce the above copyright notice,
-// this list of conditions and the following disclaimer in the documentation
-// and/or other materials provided with the distribution.
-//
-// * The name of the copyright holders may not be used to endorse or promote products
-// derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#include "precomp.hpp"
-
-using namespace cv;
-using namespace cv::gpu;
-
-#if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
-
-void cv::gpu::meanShiftFiltering(const GpuMat&, GpuMat&, int, int, TermCriteria, Stream&) { throw_no_cuda(); }
-void cv::gpu::meanShiftProc(const GpuMat&, GpuMat&, GpuMat&, int, int, TermCriteria, Stream&) { throw_no_cuda(); }
-void cv::gpu::drawColorDisp(const GpuMat&, GpuMat&, int, Stream&) { throw_no_cuda(); }
-void cv::gpu::reprojectImageTo3D(const GpuMat&, GpuMat&, const Mat&, int, Stream&) { throw_no_cuda(); }
-void cv::gpu::copyMakeBorder(const GpuMat&, GpuMat&, int, int, int, int, int, const Scalar&, Stream&) { throw_no_cuda(); }
-void cv::gpu::buildWarpPlaneMaps(Size, Rect, const Mat&, const Mat&, const Mat&, float, GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::buildWarpCylindricalMaps(Size, Rect, const Mat&, const Mat&, float, GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::buildWarpSphericalMaps(Size, Rect, const Mat&, const Mat&, float, GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::rotate(const GpuMat&, GpuMat&, Size, double, double, double, int, Stream&) { throw_no_cuda(); }
-void cv::gpu::integral(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::integralBuffered(const GpuMat&, GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::sqrIntegral(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::columnSum(const GpuMat&, GpuMat&) { throw_no_cuda(); }
-void cv::gpu::rectStdDev(const GpuMat&, const GpuMat&, GpuMat&, const Rect&, Stream&) { throw_no_cuda(); }
-void cv::gpu::evenLevels(GpuMat&, int, int, int) { throw_no_cuda(); }
-void cv::gpu::histEven(const GpuMat&, GpuMat&, int, int, int, Stream&) { throw_no_cuda(); }
-void cv::gpu::histEven(const GpuMat&, GpuMat&, GpuMat&, int, int, int, Stream&) { throw_no_cuda(); }
-void cv::gpu::histEven(const GpuMat&, GpuMat*, int*, int*, int*, Stream&) { throw_no_cuda(); }
-void cv::gpu::histEven(const GpuMat&, GpuMat*, GpuMat&, int*, int*, int*, Stream&) { throw_no_cuda(); }
-void cv::gpu::histRange(const GpuMat&, GpuMat&, const GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::histRange(const GpuMat&, GpuMat&, const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::histRange(const GpuMat&, GpuMat*, const GpuMat*, Stream&) { throw_no_cuda(); }
-void cv::gpu::histRange(const GpuMat&, GpuMat*, const GpuMat*, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::calcHist(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::equalizeHist(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::equalizeHist(const GpuMat&, GpuMat&, GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::cornerHarris(const GpuMat&, GpuMat&, int, int, double, int) { throw_no_cuda(); }
-void cv::gpu::cornerHarris(const GpuMat&, GpuMat&, GpuMat&, GpuMat&, int, int, double, int) { throw_no_cuda(); }
-void cv::gpu::cornerHarris(const GpuMat&, GpuMat&, GpuMat&, GpuMat&, GpuMat&, int, int, double, int, Stream&) { throw_no_cuda(); }
-void cv::gpu::cornerMinEigenVal(const GpuMat&, GpuMat&, int, int, int) { throw_no_cuda(); }
-void cv::gpu::cornerMinEigenVal(const GpuMat&, GpuMat&, GpuMat&, GpuMat&, int, int, int) { throw_no_cuda(); }
-void cv::gpu::cornerMinEigenVal(const GpuMat&, GpuMat&, GpuMat&, GpuMat&, GpuMat&, int, int, int, Stream&) { throw_no_cuda(); }
-void cv::gpu::mulSpectrums(const GpuMat&, const GpuMat&, GpuMat&, int, bool, Stream&) { throw_no_cuda(); }
-void cv::gpu::mulAndScaleSpectrums(const GpuMat&, const GpuMat&, GpuMat&, int, float, bool, Stream&) { throw_no_cuda(); }
-void cv::gpu::dft(const GpuMat&, GpuMat&, Size, int, Stream&) { throw_no_cuda(); }
-void cv::gpu::ConvolveBuf::create(Size, Size) { throw_no_cuda(); }
-void cv::gpu::convolve(const GpuMat&, const GpuMat&, GpuMat&, bool) { throw_no_cuda(); }
-void cv::gpu::convolve(const GpuMat&, const GpuMat&, GpuMat&, bool, ConvolveBuf&, Stream&) { throw_no_cuda(); }
-void cv::gpu::Canny(const GpuMat&, GpuMat&, double, double, int, bool) { throw_no_cuda(); }
-void cv::gpu::Canny(const GpuMat&, CannyBuf&, GpuMat&, double, double, int, bool) { throw_no_cuda(); }
-void cv::gpu::Canny(const GpuMat&, const GpuMat&, GpuMat&, double, double, bool) { throw_no_cuda(); }
-void cv::gpu::Canny(const GpuMat&, const GpuMat&, CannyBuf&, GpuMat&, double, double, bool) { throw_no_cuda(); }
-void cv::gpu::CannyBuf::create(const Size&, int) { throw_no_cuda(); }
-void cv::gpu::CannyBuf::release() { throw_no_cuda(); }
-cv::Ptr<cv::gpu::CLAHE> cv::gpu::createCLAHE(double, cv::Size) { throw_no_cuda(); return cv::Ptr<cv::gpu::CLAHE>(); }
-
-#else /* !defined (HAVE_CUDA) */
-
-////////////////////////////////////////////////////////////////////////
-// meanShiftFiltering_GPU
-
-namespace cv { namespace gpu { namespace cudev
-{
- namespace imgproc
- {
- void meanShiftFiltering_gpu(const PtrStepSzb& src, PtrStepSzb dst, int sp, int sr, int maxIter, float eps, cudaStream_t stream);
- }
-}}}
-
-void cv::gpu::meanShiftFiltering(const GpuMat& src, GpuMat& dst, int sp, int sr, TermCriteria criteria, Stream& stream)
-{
- using namespace ::cv::gpu::cudev::imgproc;
-
- if( src.empty() )
- CV_Error( cv::Error::StsBadArg, "The input image is empty" );
-
- if( src.depth() != CV_8U || src.channels() != 4 )
- CV_Error( cv::Error::StsUnsupportedFormat, "Only 8-bit, 4-channel images are supported" );
-
- dst.create( src.size(), CV_8UC4 );
-
- if( !(criteria.type & TermCriteria::MAX_ITER) )
- criteria.maxCount = 5;
-
- int maxIter = std::min(std::max(criteria.maxCount, 1), 100);
-
- float eps;
- if( !(criteria.type & TermCriteria::EPS) )
- eps = 1.f;
- eps = (float)std::max(criteria.epsilon, 0.0);
-
- meanShiftFiltering_gpu(src, dst, sp, sr, maxIter, eps, StreamAccessor::getStream(stream));
-}
-
-////////////////////////////////////////////////////////////////////////
-// meanShiftProc_GPU
-
-namespace cv { namespace gpu { namespace cudev
-{
- namespace imgproc
- {
- void meanShiftProc_gpu(const PtrStepSzb& src, PtrStepSzb dstr, PtrStepSzb dstsp, int sp, int sr, int maxIter, float eps, cudaStream_t stream);
- }
-}}}
-
-void cv::gpu::meanShiftProc(const GpuMat& src, GpuMat& dstr, GpuMat& dstsp, int sp, int sr, TermCriteria criteria, Stream& stream)
-{
- using namespace ::cv::gpu::cudev::imgproc;
-
- if( src.empty() )
- CV_Error( cv::Error::StsBadArg, "The input image is empty" );
-
- if( src.depth() != CV_8U || src.channels() != 4 )
- CV_Error( cv::Error::StsUnsupportedFormat, "Only 8-bit, 4-channel images are supported" );
-
- dstr.create( src.size(), CV_8UC4 );
- dstsp.create( src.size(), CV_16SC2 );
-
- if( !(criteria.type & TermCriteria::MAX_ITER) )
- criteria.maxCount = 5;
-
- int maxIter = std::min(std::max(criteria.maxCount, 1), 100);
-
- float eps;
- if( !(criteria.type & TermCriteria::EPS) )
- eps = 1.f;
- eps = (float)std::max(criteria.epsilon, 0.0);
-
- meanShiftProc_gpu(src, dstr, dstsp, sp, sr, maxIter, eps, StreamAccessor::getStream(stream));
-}
-
-////////////////////////////////////////////////////////////////////////
-// drawColorDisp
-
-namespace cv { namespace gpu { namespace cudev
-{
- namespace imgproc
- {
- void drawColorDisp_gpu(const PtrStepSzb& src, const PtrStepSzb& dst, int ndisp, const cudaStream_t& stream);
- void drawColorDisp_gpu(const PtrStepSz<short>& src, const PtrStepSzb& dst, int ndisp, const cudaStream_t& stream);
- }
-}}}
-
-namespace
-{
- template <typename T>
- void drawColorDisp_caller(const GpuMat& src, GpuMat& dst, int ndisp, const cudaStream_t& stream)
- {
- using namespace ::cv::gpu::cudev::imgproc;
-
- dst.create(src.size(), CV_8UC4);
-
- drawColorDisp_gpu((PtrStepSz<T>)src, dst, ndisp, stream);
- }
-
- typedef void (*drawColorDisp_caller_t)(const GpuMat& src, GpuMat& dst, int ndisp, const cudaStream_t& stream);
-
- const drawColorDisp_caller_t drawColorDisp_callers[] = {drawColorDisp_caller<unsigned char>, 0, 0, drawColorDisp_caller<short>, 0, 0, 0, 0};
-}
-
-void cv::gpu::drawColorDisp(const GpuMat& src, GpuMat& dst, int ndisp, Stream& stream)
-{
- CV_Assert(src.type() == CV_8U || src.type() == CV_16S);
-
- drawColorDisp_callers[src.type()](src, dst, ndisp, StreamAccessor::getStream(stream));
-}
-
-////////////////////////////////////////////////////////////////////////
-// reprojectImageTo3D
-
-namespace cv { namespace gpu { namespace cudev
-{
- namespace imgproc
- {
- template <typename T, typename D>
- void reprojectImageTo3D_gpu(const PtrStepSzb disp, PtrStepSzb xyz, const float* q, cudaStream_t stream);
- }
-}}}
-
-void cv::gpu::reprojectImageTo3D(const GpuMat& disp, GpuMat& xyz, const Mat& Q, int dst_cn, Stream& stream)
-{
- using namespace cv::gpu::cudev::imgproc;
-
- typedef void (*func_t)(const PtrStepSzb disp, PtrStepSzb xyz, const float* q, cudaStream_t stream);
- static const func_t funcs[2][4] =
- {
- {reprojectImageTo3D_gpu<uchar, float3>, 0, 0, reprojectImageTo3D_gpu<short, float3>},
- {reprojectImageTo3D_gpu<uchar, float4>, 0, 0, reprojectImageTo3D_gpu<short, float4>}
- };
-
- CV_Assert(disp.type() == CV_8U || disp.type() == CV_16S);
- CV_Assert(Q.type() == CV_32F && Q.rows == 4 && Q.cols == 4 && Q.isContinuous());
- CV_Assert(dst_cn == 3 || dst_cn == 4);
-
- xyz.create(disp.size(), CV_MAKE_TYPE(CV_32F, dst_cn));
-
- funcs[dst_cn == 4][disp.type()](disp, xyz, Q.ptr<float>(), StreamAccessor::getStream(stream));
-}
-
-////////////////////////////////////////////////////////////////////////
-// copyMakeBorder
-
-namespace cv { namespace gpu { namespace cudev
-{
- namespace imgproc
- {
- template <typename T, int cn> void copyMakeBorder_gpu(const PtrStepSzb& src, const PtrStepSzb& dst, int top, int left, int borderMode, const T* borderValue, cudaStream_t stream);
- }
-}}}
-
-namespace
-{
- template <typename T, int cn> void copyMakeBorder_caller(const PtrStepSzb& src, const PtrStepSzb& dst, int top, int left, int borderType, const Scalar& value, cudaStream_t stream)
- {
- using namespace ::cv::gpu::cudev::imgproc;
-
- Scalar_<T> val(saturate_cast<T>(value[0]), saturate_cast<T>(value[1]), saturate_cast<T>(value[2]), saturate_cast<T>(value[3]));
-
- copyMakeBorder_gpu<T, cn>(src, dst, top, left, borderType, val.val, stream);
- }
-}
-
-#if defined __GNUC__ && __GNUC__ > 2 && __GNUC_MINOR__ > 4
-typedef Npp32s __attribute__((__may_alias__)) Npp32s_a;
-#else
-typedef Npp32s Npp32s_a;
-#endif
-
-void cv::gpu::copyMakeBorder(const GpuMat& src, GpuMat& dst, int top, int bottom, int left, int right, int borderType, const Scalar& value, Stream& s)
-{
- CV_Assert(src.depth() <= CV_32F && src.channels() <= 4);
- CV_Assert(borderType == BORDER_REFLECT101 || borderType == BORDER_REPLICATE || borderType == BORDER_CONSTANT || borderType == BORDER_REFLECT || borderType == BORDER_WRAP);
-
- dst.create(src.rows + top + bottom, src.cols + left + right, src.type());
-
- cudaStream_t stream = StreamAccessor::getStream(s);
-
- if (borderType == BORDER_CONSTANT && (src.type() == CV_8UC1 || src.type() == CV_8UC4 || src.type() == CV_32SC1 || src.type() == CV_32FC1))
- {
- NppiSize srcsz;
- srcsz.width = src.cols;
- srcsz.height = src.rows;
-
- NppiSize dstsz;
- dstsz.width = dst.cols;
- dstsz.height = dst.rows;
-
- NppStreamHandler h(stream);
-
- switch (src.type())
- {
- case CV_8UC1:
- {
- Npp8u nVal = saturate_cast<Npp8u>(value[0]);
- nppSafeCall( nppiCopyConstBorder_8u_C1R(src.ptr<Npp8u>(), static_cast<int>(src.step), srcsz,
- dst.ptr<Npp8u>(), static_cast<int>(dst.step), dstsz, top, left, nVal) );
- break;
- }
- case CV_8UC4:
- {
- Npp8u nVal[] = {saturate_cast<Npp8u>(value[0]), saturate_cast<Npp8u>(value[1]), saturate_cast<Npp8u>(value[2]), saturate_cast<Npp8u>(value[3])};
- nppSafeCall( nppiCopyConstBorder_8u_C4R(src.ptr<Npp8u>(), static_cast<int>(src.step), srcsz,
- dst.ptr<Npp8u>(), static_cast<int>(dst.step), dstsz, top, left, nVal) );
- break;
- }
- case CV_32SC1:
- {
- Npp32s nVal = saturate_cast<Npp32s>(value[0]);
- nppSafeCall( nppiCopyConstBorder_32s_C1R(src.ptr<Npp32s>(), static_cast<int>(src.step), srcsz,
- dst.ptr<Npp32s>(), static_cast<int>(dst.step), dstsz, top, left, nVal) );
- break;
- }
- case CV_32FC1:
- {
- Npp32f val = saturate_cast<Npp32f>(value[0]);
- Npp32s nVal = *(reinterpret_cast<Npp32s_a*>(&val));
- nppSafeCall( nppiCopyConstBorder_32s_C1R(src.ptr<Npp32s>(), static_cast<int>(src.step), srcsz,
- dst.ptr<Npp32s>(), static_cast<int>(dst.step), dstsz, top, left, nVal) );
- break;
- }
- }
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
- else
- {
- typedef void (*caller_t)(const PtrStepSzb& src, const PtrStepSzb& dst, int top, int left, int borderType, const Scalar& value, cudaStream_t stream);
- static const caller_t callers[6][4] =
- {
- { copyMakeBorder_caller<uchar, 1> , copyMakeBorder_caller<uchar, 2> , copyMakeBorder_caller<uchar, 3> , copyMakeBorder_caller<uchar, 4>},
- {0/*copyMakeBorder_caller<schar, 1>*/, 0/*copyMakeBorder_caller<schar, 2>*/ , 0/*copyMakeBorder_caller<schar, 3>*/, 0/*copyMakeBorder_caller<schar, 4>*/},
- { copyMakeBorder_caller<ushort, 1> , 0/*copyMakeBorder_caller<ushort, 2>*/, copyMakeBorder_caller<ushort, 3> , copyMakeBorder_caller<ushort, 4>},
- { copyMakeBorder_caller<short, 1> , 0/*copyMakeBorder_caller<short, 2>*/ , copyMakeBorder_caller<short, 3> , copyMakeBorder_caller<short, 4>},
- {0/*copyMakeBorder_caller<int, 1>*/, 0/*copyMakeBorder_caller<int, 2>*/ , 0/*copyMakeBorder_caller<int, 3>*/, 0/*copyMakeBorder_caller<int , 4>*/},
- { copyMakeBorder_caller<float, 1> , 0/*copyMakeBorder_caller<float, 2>*/ , copyMakeBorder_caller<float, 3> , copyMakeBorder_caller<float ,4>}
- };
-
- caller_t func = callers[src.depth()][src.channels() - 1];
- CV_Assert(func != 0);
-
- int gpuBorderType;
- CV_Assert(tryConvertToGpuBorderType(borderType, gpuBorderType));
-
- func(src, dst, top, left, gpuBorderType, value, stream);
- }
-}
-
-//////////////////////////////////////////////////////////////////////////////
-// buildWarpPlaneMaps
-
-namespace cv { namespace gpu { namespace cudev
-{
- namespace imgproc
- {
- void buildWarpPlaneMaps(int tl_u, int tl_v, PtrStepSzf map_x, PtrStepSzf map_y,
- const float k_rinv[9], const float r_kinv[9], const float t[3], float scale,
- cudaStream_t stream);
- }
-}}}
-
-void cv::gpu::buildWarpPlaneMaps(Size src_size, Rect dst_roi, const Mat &K, const Mat& R, const Mat &T,
- float scale, GpuMat& map_x, GpuMat& map_y, Stream& stream)
-{
- (void)src_size;
- using namespace ::cv::gpu::cudev::imgproc;
-
- CV_Assert(K.size() == Size(3,3) && K.type() == CV_32F);
- CV_Assert(R.size() == Size(3,3) && R.type() == CV_32F);
- CV_Assert((T.size() == Size(3,1) || T.size() == Size(1,3)) && T.type() == CV_32F && T.isContinuous());
-
- Mat K_Rinv = K * R.t();
- Mat R_Kinv = R * K.inv();
- CV_Assert(K_Rinv.isContinuous());
- CV_Assert(R_Kinv.isContinuous());
-
- map_x.create(dst_roi.size(), CV_32F);
- map_y.create(dst_roi.size(), CV_32F);
- cudev::imgproc::buildWarpPlaneMaps(dst_roi.tl().x, dst_roi.tl().y, map_x, map_y, K_Rinv.ptr<float>(), R_Kinv.ptr<float>(),
- T.ptr<float>(), scale, StreamAccessor::getStream(stream));
-}
-
-//////////////////////////////////////////////////////////////////////////////
-// buildWarpCylyndricalMaps
-
-namespace cv { namespace gpu { namespace cudev
-{
- namespace imgproc
- {
- void buildWarpCylindricalMaps(int tl_u, int tl_v, PtrStepSzf map_x, PtrStepSzf map_y,
- const float k_rinv[9], const float r_kinv[9], float scale,
- cudaStream_t stream);
- }
-}}}
-
-void cv::gpu::buildWarpCylindricalMaps(Size src_size, Rect dst_roi, const Mat &K, const Mat& R, float scale,
- GpuMat& map_x, GpuMat& map_y, Stream& stream)
-{
- (void)src_size;
- using namespace ::cv::gpu::cudev::imgproc;
-
- CV_Assert(K.size() == Size(3,3) && K.type() == CV_32F);
- CV_Assert(R.size() == Size(3,3) && R.type() == CV_32F);
-
- Mat K_Rinv = K * R.t();
- Mat R_Kinv = R * K.inv();
- CV_Assert(K_Rinv.isContinuous());
- CV_Assert(R_Kinv.isContinuous());
-
- map_x.create(dst_roi.size(), CV_32F);
- map_y.create(dst_roi.size(), CV_32F);
- cudev::imgproc::buildWarpCylindricalMaps(dst_roi.tl().x, dst_roi.tl().y, map_x, map_y, K_Rinv.ptr<float>(), R_Kinv.ptr<float>(), scale, StreamAccessor::getStream(stream));
-}
-
-
-//////////////////////////////////////////////////////////////////////////////
-// buildWarpSphericalMaps
-
-namespace cv { namespace gpu { namespace cudev
-{
- namespace imgproc
- {
- void buildWarpSphericalMaps(int tl_u, int tl_v, PtrStepSzf map_x, PtrStepSzf map_y,
- const float k_rinv[9], const float r_kinv[9], float scale,
- cudaStream_t stream);
- }
-}}}
-
-void cv::gpu::buildWarpSphericalMaps(Size src_size, Rect dst_roi, const Mat &K, const Mat& R, float scale,
- GpuMat& map_x, GpuMat& map_y, Stream& stream)
-{
- (void)src_size;
- using namespace ::cv::gpu::cudev::imgproc;
-
- CV_Assert(K.size() == Size(3,3) && K.type() == CV_32F);
- CV_Assert(R.size() == Size(3,3) && R.type() == CV_32F);
-
- Mat K_Rinv = K * R.t();
- Mat R_Kinv = R * K.inv();
- CV_Assert(K_Rinv.isContinuous());
- CV_Assert(R_Kinv.isContinuous());
-
- map_x.create(dst_roi.size(), CV_32F);
- map_y.create(dst_roi.size(), CV_32F);
- cudev::imgproc::buildWarpSphericalMaps(dst_roi.tl().x, dst_roi.tl().y, map_x, map_y, K_Rinv.ptr<float>(), R_Kinv.ptr<float>(), scale, StreamAccessor::getStream(stream));
-}
-
-////////////////////////////////////////////////////////////////////////
-// rotate
-
-namespace
-{
- template<int DEPTH> struct NppTypeTraits;
- template<> struct NppTypeTraits<CV_8U> { typedef Npp8u npp_t; };
- template<> struct NppTypeTraits<CV_8S> { typedef Npp8s npp_t; };
- template<> struct NppTypeTraits<CV_16U> { typedef Npp16u npp_t; };
- template<> struct NppTypeTraits<CV_16S> { typedef Npp16s npp_t; };
- template<> struct NppTypeTraits<CV_32S> { typedef Npp32s npp_t; };
- template<> struct NppTypeTraits<CV_32F> { typedef Npp32f npp_t; };
- template<> struct NppTypeTraits<CV_64F> { typedef Npp64f npp_t; };
-
- template <int DEPTH> struct NppRotateFunc
- {
- typedef typename NppTypeTraits<DEPTH>::npp_t npp_t;
-
- typedef NppStatus (*func_t)(const npp_t* pSrc, NppiSize oSrcSize, int nSrcStep, NppiRect oSrcROI,
- npp_t* pDst, int nDstStep, NppiRect oDstROI,
- double nAngle, double nShiftX, double nShiftY, int eInterpolation);
- };
-
- template <int DEPTH, typename NppRotateFunc<DEPTH>::func_t func> struct NppRotate
- {
- typedef typename NppRotateFunc<DEPTH>::npp_t npp_t;
-
- static void call(const GpuMat& src, GpuMat& dst, Size dsize, double angle, double xShift, double yShift, int interpolation, cudaStream_t stream)
- {
- (void)dsize;
- static const int npp_inter[] = {NPPI_INTER_NN, NPPI_INTER_LINEAR, NPPI_INTER_CUBIC};
-
- NppStreamHandler h(stream);
-
- NppiSize srcsz;
- srcsz.height = src.rows;
- srcsz.width = src.cols;
- NppiRect srcroi;
- srcroi.x = srcroi.y = 0;
- srcroi.height = src.rows;
- srcroi.width = src.cols;
- NppiRect dstroi;
- dstroi.x = dstroi.y = 0;
- dstroi.height = dst.rows;
- dstroi.width = dst.cols;
-
- nppSafeCall( func(src.ptr<npp_t>(), srcsz, static_cast<int>(src.step), srcroi,
- dst.ptr<npp_t>(), static_cast<int>(dst.step), dstroi, angle, xShift, yShift, npp_inter[interpolation]) );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
- };
-}
-
-void cv::gpu::rotate(const GpuMat& src, GpuMat& dst, Size dsize, double angle, double xShift, double yShift, int interpolation, Stream& stream)
-{
- typedef void (*func_t)(const GpuMat& src, GpuMat& dst, Size dsize, double angle, double xShift, double yShift, int interpolation, cudaStream_t stream);
-
- static const func_t funcs[6][4] =
- {
- {NppRotate<CV_8U, nppiRotate_8u_C1R>::call, 0, NppRotate<CV_8U, nppiRotate_8u_C3R>::call, NppRotate<CV_8U, nppiRotate_8u_C4R>::call},
- {0,0,0,0},
- {NppRotate<CV_16U, nppiRotate_16u_C1R>::call, 0, NppRotate<CV_16U, nppiRotate_16u_C3R>::call, NppRotate<CV_16U, nppiRotate_16u_C4R>::call},
- {0,0,0,0},
- {0,0,0,0},
- {NppRotate<CV_32F, nppiRotate_32f_C1R>::call, 0, NppRotate<CV_32F, nppiRotate_32f_C3R>::call, NppRotate<CV_32F, nppiRotate_32f_C4R>::call}
- };
-
- CV_Assert(src.depth() == CV_8U || src.depth() == CV_16U || src.depth() == CV_32F);
- CV_Assert(src.channels() == 1 || src.channels() == 3 || src.channels() == 4);
- CV_Assert(interpolation == INTER_NEAREST || interpolation == INTER_LINEAR || interpolation == INTER_CUBIC);
-
- dst.create(dsize, src.type());
- dst.setTo(Scalar::all(0));
-
- funcs[src.depth()][src.channels() - 1](src, dst, dsize, angle, xShift, yShift, interpolation, StreamAccessor::getStream(stream));
-}
-
-////////////////////////////////////////////////////////////////////////
-// integral
-
-void cv::gpu::integral(const GpuMat& src, GpuMat& sum, Stream& s)
-{
- GpuMat buffer;
- integralBuffered(src, sum, buffer, s);
-}
-
-namespace cv { namespace gpu { namespace cudev
-{
- namespace imgproc
- {
- void shfl_integral_gpu(const PtrStepSzb& img, PtrStepSz<unsigned int> integral, cudaStream_t stream);
- }
-}}}
-
-void cv::gpu::integralBuffered(const GpuMat& src, GpuMat& sum, GpuMat& buffer, Stream& s)
-{
- CV_Assert(src.type() == CV_8UC1);
-
- cudaStream_t stream = StreamAccessor::getStream(s);
-
- cv::Size whole;
- cv::Point offset;
-
- src.locateROI(whole, offset);
-
- if (deviceSupports(WARP_SHUFFLE_FUNCTIONS) && src.cols <= 2048
- && offset.x % 16 == 0 && ((src.cols + 63) / 64) * 64 <= (static_cast<int>(src.step) - offset.x))
- {
- ensureSizeIsEnough(((src.rows + 7) / 8) * 8, ((src.cols + 63) / 64) * 64, CV_32SC1, buffer);
-
- cv::gpu::cudev::imgproc::shfl_integral_gpu(src, buffer, stream);
-
- sum.create(src.rows + 1, src.cols + 1, CV_32SC1);
- if (s)
- s.enqueueMemSet(sum, Scalar::all(0));
- else
- sum.setTo(Scalar::all(0));
-
- GpuMat inner = sum(Rect(1, 1, src.cols, src.rows));
- GpuMat res = buffer(Rect(0, 0, src.cols, src.rows));
-
- if (s)
- s.enqueueCopy(res, inner);
- else
- res.copyTo(inner);
- }
- else
- {
- sum.create(src.rows + 1, src.cols + 1, CV_32SC1);
-
- NcvSize32u roiSize;
- roiSize.width = src.cols;
- roiSize.height = src.rows;
-
- cudaDeviceProp prop;
- cudaSafeCall( cudaGetDeviceProperties(&prop, cv::gpu::getDevice()) );
-
- Ncv32u bufSize;
- ncvSafeCall( nppiStIntegralGetSize_8u32u(roiSize, &bufSize, prop) );
- ensureSizeIsEnough(1, bufSize, CV_8UC1, buffer);
-
-
- NppStStreamHandler h(stream);
-
- ncvSafeCall( nppiStIntegral_8u32u_C1R(const_cast<Ncv8u*>(src.ptr<Ncv8u>()), static_cast<int>(src.step),
- sum.ptr<Ncv32u>(), static_cast<int>(sum.step), roiSize, buffer.ptr<Ncv8u>(), bufSize, prop) );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
-}
-
-//////////////////////////////////////////////////////////////////////////////
-// sqrIntegral
-
-void cv::gpu::sqrIntegral(const GpuMat& src, GpuMat& sqsum, Stream& s)
-{
- CV_Assert(src.type() == CV_8U);
-
- NcvSize32u roiSize;
- roiSize.width = src.cols;
- roiSize.height = src.rows;
-
- cudaDeviceProp prop;
- cudaSafeCall( cudaGetDeviceProperties(&prop, cv::gpu::getDevice()) );
-
- Ncv32u bufSize;
- ncvSafeCall(nppiStSqrIntegralGetSize_8u64u(roiSize, &bufSize, prop));
- GpuMat buf(1, bufSize, CV_8U);
-
- cudaStream_t stream = StreamAccessor::getStream(s);
-
- NppStStreamHandler h(stream);
-
- sqsum.create(src.rows + 1, src.cols + 1, CV_64F);
- ncvSafeCall(nppiStSqrIntegral_8u64u_C1R(const_cast<Ncv8u*>(src.ptr<Ncv8u>(0)), static_cast<int>(src.step),
- sqsum.ptr<Ncv64u>(0), static_cast<int>(sqsum.step), roiSize, buf.ptr<Ncv8u>(0), bufSize, prop));
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
-}
-
-//////////////////////////////////////////////////////////////////////////////
-// columnSum
-
-namespace cv { namespace gpu { namespace cudev
-{
- namespace imgproc
- {
- void columnSum_32F(const PtrStepSzb src, const PtrStepSzb dst);
- }
-}}}
-
-void cv::gpu::columnSum(const GpuMat& src, GpuMat& dst)
-{
- using namespace ::cv::gpu::cudev::imgproc;
-
- CV_Assert(src.type() == CV_32F);
-
- dst.create(src.size(), CV_32F);
-
- cudev::imgproc::columnSum_32F(src, dst);
-}
-
-void cv::gpu::rectStdDev(const GpuMat& src, const GpuMat& sqr, GpuMat& dst, const Rect& rect, Stream& s)
-{
- CV_Assert(src.type() == CV_32SC1 && sqr.type() == CV_64FC1);
-
- dst.create(src.size(), CV_32FC1);
-
- NppiSize sz;
- sz.width = src.cols;
- sz.height = src.rows;
-
- NppiRect nppRect;
- nppRect.height = rect.height;
- nppRect.width = rect.width;
- nppRect.x = rect.x;
- nppRect.y = rect.y;
-
- cudaStream_t stream = StreamAccessor::getStream(s);
-
- NppStreamHandler h(stream);
-
- nppSafeCall( nppiRectStdDev_32s32f_C1R(src.ptr<Npp32s>(), static_cast<int>(src.step), sqr.ptr<Npp64f>(), static_cast<int>(sqr.step),
- dst.ptr<Npp32f>(), static_cast<int>(dst.step), sz, nppRect) );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
-}
-
-
-////////////////////////////////////////////////////////////////////////
-// Histogram
-
-namespace
-{
- typedef NppStatus (*get_buf_size_c1_t)(NppiSize oSizeROI, int nLevels, int* hpBufferSize);
- typedef NppStatus (*get_buf_size_c4_t)(NppiSize oSizeROI, int nLevels[], int* hpBufferSize);
-
- template<int SDEPTH> struct NppHistogramEvenFuncC1
- {
- typedef typename NppTypeTraits<SDEPTH>::npp_t src_t;
-
- typedef NppStatus (*func_ptr)(const src_t* pSrc, int nSrcStep, NppiSize oSizeROI, Npp32s * pHist,
- int nLevels, Npp32s nLowerLevel, Npp32s nUpperLevel, Npp8u * pBuffer);
- };
- template<int SDEPTH> struct NppHistogramEvenFuncC4
- {
- typedef typename NppTypeTraits<SDEPTH>::npp_t src_t;
-
- typedef NppStatus (*func_ptr)(const src_t* pSrc, int nSrcStep, NppiSize oSizeROI,
- Npp32s * pHist[4], int nLevels[4], Npp32s nLowerLevel[4], Npp32s nUpperLevel[4], Npp8u * pBuffer);
- };
-
- template<int SDEPTH, typename NppHistogramEvenFuncC1<SDEPTH>::func_ptr func, get_buf_size_c1_t get_buf_size>
- struct NppHistogramEvenC1
- {
- typedef typename NppHistogramEvenFuncC1<SDEPTH>::src_t src_t;
-
- static void hist(const GpuMat& src, GpuMat& hist, GpuMat& buffer, int histSize, int lowerLevel, int upperLevel, cudaStream_t stream)
- {
- int levels = histSize + 1;
- hist.create(1, histSize, CV_32S);
-
- NppiSize sz;
- sz.width = src.cols;
- sz.height = src.rows;
-
- int buf_size;
- get_buf_size(sz, levels, &buf_size);
-
- ensureSizeIsEnough(1, buf_size, CV_8U, buffer);
-
- NppStreamHandler h(stream);
-
- nppSafeCall( func(src.ptr<src_t>(), static_cast<int>(src.step), sz, hist.ptr<Npp32s>(), levels,
- lowerLevel, upperLevel, buffer.ptr<Npp8u>()) );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
- };
- template<int SDEPTH, typename NppHistogramEvenFuncC4<SDEPTH>::func_ptr func, get_buf_size_c4_t get_buf_size>
- struct NppHistogramEvenC4
- {
- typedef typename NppHistogramEvenFuncC4<SDEPTH>::src_t src_t;
-
- static void hist(const GpuMat& src, GpuMat hist[4], GpuMat& buffer, int histSize[4], int lowerLevel[4], int upperLevel[4], cudaStream_t stream)
- {
- int levels[] = {histSize[0] + 1, histSize[1] + 1, histSize[2] + 1, histSize[3] + 1};
- hist[0].create(1, histSize[0], CV_32S);
- hist[1].create(1, histSize[1], CV_32S);
- hist[2].create(1, histSize[2], CV_32S);
- hist[3].create(1, histSize[3], CV_32S);
-
- NppiSize sz;
- sz.width = src.cols;
- sz.height = src.rows;
-
- Npp32s* pHist[] = {hist[0].ptr<Npp32s>(), hist[1].ptr<Npp32s>(), hist[2].ptr<Npp32s>(), hist[3].ptr<Npp32s>()};
-
- int buf_size;
- get_buf_size(sz, levels, &buf_size);
-
- ensureSizeIsEnough(1, buf_size, CV_8U, buffer);
-
- NppStreamHandler h(stream);
-
- nppSafeCall( func(src.ptr<src_t>(), static_cast<int>(src.step), sz, pHist, levels, lowerLevel, upperLevel, buffer.ptr<Npp8u>()) );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
- };
-
- template<int SDEPTH> struct NppHistogramRangeFuncC1
- {
- typedef typename NppTypeTraits<SDEPTH>::npp_t src_t;
- typedef Npp32s level_t;
- enum {LEVEL_TYPE_CODE=CV_32SC1};
-
- typedef NppStatus (*func_ptr)(const src_t* pSrc, int nSrcStep, NppiSize oSizeROI, Npp32s* pHist,
- const Npp32s* pLevels, int nLevels, Npp8u* pBuffer);
- };
- template<> struct NppHistogramRangeFuncC1<CV_32F>
- {
- typedef Npp32f src_t;
- typedef Npp32f level_t;
- enum {LEVEL_TYPE_CODE=CV_32FC1};
-
- typedef NppStatus (*func_ptr)(const Npp32f* pSrc, int nSrcStep, NppiSize oSizeROI, Npp32s* pHist,
- const Npp32f* pLevels, int nLevels, Npp8u* pBuffer);
- };
- template<int SDEPTH> struct NppHistogramRangeFuncC4
- {
- typedef typename NppTypeTraits<SDEPTH>::npp_t src_t;
- typedef Npp32s level_t;
- enum {LEVEL_TYPE_CODE=CV_32SC1};
-
- typedef NppStatus (*func_ptr)(const src_t* pSrc, int nSrcStep, NppiSize oSizeROI, Npp32s* pHist[4],
- const Npp32s* pLevels[4], int nLevels[4], Npp8u* pBuffer);
- };
- template<> struct NppHistogramRangeFuncC4<CV_32F>
- {
- typedef Npp32f src_t;
- typedef Npp32f level_t;
- enum {LEVEL_TYPE_CODE=CV_32FC1};
-
- typedef NppStatus (*func_ptr)(const Npp32f* pSrc, int nSrcStep, NppiSize oSizeROI, Npp32s* pHist[4],
- const Npp32f* pLevels[4], int nLevels[4], Npp8u* pBuffer);
- };
-
- template<int SDEPTH, typename NppHistogramRangeFuncC1<SDEPTH>::func_ptr func, get_buf_size_c1_t get_buf_size>
- struct NppHistogramRangeC1
- {
- typedef typename NppHistogramRangeFuncC1<SDEPTH>::src_t src_t;
- typedef typename NppHistogramRangeFuncC1<SDEPTH>::level_t level_t;
- enum {LEVEL_TYPE_CODE=NppHistogramRangeFuncC1<SDEPTH>::LEVEL_TYPE_CODE};
-
- static void hist(const GpuMat& src, GpuMat& hist, const GpuMat& levels, GpuMat& buffer, cudaStream_t stream)
- {
- CV_Assert(levels.type() == LEVEL_TYPE_CODE && levels.rows == 1);
-
- hist.create(1, levels.cols - 1, CV_32S);
-
- NppiSize sz;
- sz.width = src.cols;
- sz.height = src.rows;
-
- int buf_size;
- get_buf_size(sz, levels.cols, &buf_size);
-
- ensureSizeIsEnough(1, buf_size, CV_8U, buffer);
-
- NppStreamHandler h(stream);
-
- nppSafeCall( func(src.ptr<src_t>(), static_cast<int>(src.step), sz, hist.ptr<Npp32s>(), levels.ptr<level_t>(), levels.cols, buffer.ptr<Npp8u>()) );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
- };
- template<int SDEPTH, typename NppHistogramRangeFuncC4<SDEPTH>::func_ptr func, get_buf_size_c4_t get_buf_size>
- struct NppHistogramRangeC4
- {
- typedef typename NppHistogramRangeFuncC4<SDEPTH>::src_t src_t;
- typedef typename NppHistogramRangeFuncC1<SDEPTH>::level_t level_t;
- enum {LEVEL_TYPE_CODE=NppHistogramRangeFuncC1<SDEPTH>::LEVEL_TYPE_CODE};
-
- static void hist(const GpuMat& src, GpuMat hist[4], const GpuMat levels[4], GpuMat& buffer, cudaStream_t stream)
- {
- CV_Assert(levels[0].type() == LEVEL_TYPE_CODE && levels[0].rows == 1);
- CV_Assert(levels[1].type() == LEVEL_TYPE_CODE && levels[1].rows == 1);
- CV_Assert(levels[2].type() == LEVEL_TYPE_CODE && levels[2].rows == 1);
- CV_Assert(levels[3].type() == LEVEL_TYPE_CODE && levels[3].rows == 1);
-
- hist[0].create(1, levels[0].cols - 1, CV_32S);
- hist[1].create(1, levels[1].cols - 1, CV_32S);
- hist[2].create(1, levels[2].cols - 1, CV_32S);
- hist[3].create(1, levels[3].cols - 1, CV_32S);
-
- Npp32s* pHist[] = {hist[0].ptr<Npp32s>(), hist[1].ptr<Npp32s>(), hist[2].ptr<Npp32s>(), hist[3].ptr<Npp32s>()};
- int nLevels[] = {levels[0].cols, levels[1].cols, levels[2].cols, levels[3].cols};
- const level_t* pLevels[] = {levels[0].ptr<level_t>(), levels[1].ptr<level_t>(), levels[2].ptr<level_t>(), levels[3].ptr<level_t>()};
-
- NppiSize sz;
- sz.width = src.cols;
- sz.height = src.rows;
-
- int buf_size;
- get_buf_size(sz, nLevels, &buf_size);
-
- ensureSizeIsEnough(1, buf_size, CV_8U, buffer);
-
- NppStreamHandler h(stream);
-
- nppSafeCall( func(src.ptr<src_t>(), static_cast<int>(src.step), sz, pHist, pLevels, nLevels, buffer.ptr<Npp8u>()) );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
- };
-}
-
-void cv::gpu::evenLevels(GpuMat& levels, int nLevels, int lowerLevel, int upperLevel)
-{
- Mat host_levels(1, nLevels, CV_32SC1);
- nppSafeCall( nppiEvenLevelsHost_32s(host_levels.ptr<Npp32s>(), nLevels, lowerLevel, upperLevel) );
- levels.upload(host_levels);
-}
-
-void cv::gpu::histEven(const GpuMat& src, GpuMat& hist, int histSize, int lowerLevel, int upperLevel, Stream& stream)
-{
- GpuMat buf;
- histEven(src, hist, buf, histSize, lowerLevel, upperLevel, stream);
-}
-
-void cv::gpu::histEven(const GpuMat& src, GpuMat& hist, GpuMat& buf, int histSize, int lowerLevel, int upperLevel, Stream& stream)
-{
- CV_Assert(src.type() == CV_8UC1 || src.type() == CV_16UC1 || src.type() == CV_16SC1 );
-
- typedef void (*hist_t)(const GpuMat& src, GpuMat& hist, GpuMat& buf, int levels, int lowerLevel, int upperLevel, cudaStream_t stream);
- static const hist_t hist_callers[] =
- {
- NppHistogramEvenC1<CV_8U , nppiHistogramEven_8u_C1R , nppiHistogramEvenGetBufferSize_8u_C1R >::hist,
- 0,
- NppHistogramEvenC1<CV_16U, nppiHistogramEven_16u_C1R, nppiHistogramEvenGetBufferSize_16u_C1R>::hist,
- NppHistogramEvenC1<CV_16S, nppiHistogramEven_16s_C1R, nppiHistogramEvenGetBufferSize_16s_C1R>::hist
- };
-
- hist_callers[src.depth()](src, hist, buf, histSize, lowerLevel, upperLevel, StreamAccessor::getStream(stream));
-}
-
-void cv::gpu::histEven(const GpuMat& src, GpuMat hist[4], int histSize[4], int lowerLevel[4], int upperLevel[4], Stream& stream)
-{
- GpuMat buf;
- histEven(src, hist, buf, histSize, lowerLevel, upperLevel, stream);
-}
-
-void cv::gpu::histEven(const GpuMat& src, GpuMat hist[4], GpuMat& buf, int histSize[4], int lowerLevel[4], int upperLevel[4], Stream& stream)
-{
- CV_Assert(src.type() == CV_8UC4 || src.type() == CV_16UC4 || src.type() == CV_16SC4 );
-
- typedef void (*hist_t)(const GpuMat& src, GpuMat hist[4], GpuMat& buf, int levels[4], int lowerLevel[4], int upperLevel[4], cudaStream_t stream);
- static const hist_t hist_callers[] =
- {
- NppHistogramEvenC4<CV_8U , nppiHistogramEven_8u_C4R , nppiHistogramEvenGetBufferSize_8u_C4R >::hist,
- 0,
- NppHistogramEvenC4<CV_16U, nppiHistogramEven_16u_C4R, nppiHistogramEvenGetBufferSize_16u_C4R>::hist,
- NppHistogramEvenC4<CV_16S, nppiHistogramEven_16s_C4R, nppiHistogramEvenGetBufferSize_16s_C4R>::hist
- };
-
- hist_callers[src.depth()](src, hist, buf, histSize, lowerLevel, upperLevel, StreamAccessor::getStream(stream));
-}
-
-void cv::gpu::histRange(const GpuMat& src, GpuMat& hist, const GpuMat& levels, Stream& stream)
-{
- GpuMat buf;
- histRange(src, hist, levels, buf, stream);
-}
-
-void cv::gpu::histRange(const GpuMat& src, GpuMat& hist, const GpuMat& levels, GpuMat& buf, Stream& stream)
-{
- CV_Assert(src.type() == CV_8UC1 || src.type() == CV_16UC1 || src.type() == CV_16SC1 || src.type() == CV_32FC1);
-
- typedef void (*hist_t)(const GpuMat& src, GpuMat& hist, const GpuMat& levels, GpuMat& buf, cudaStream_t stream);
- static const hist_t hist_callers[] =
- {
- NppHistogramRangeC1<CV_8U , nppiHistogramRange_8u_C1R , nppiHistogramRangeGetBufferSize_8u_C1R >::hist,
- 0,
- NppHistogramRangeC1<CV_16U, nppiHistogramRange_16u_C1R, nppiHistogramRangeGetBufferSize_16u_C1R>::hist,
- NppHistogramRangeC1<CV_16S, nppiHistogramRange_16s_C1R, nppiHistogramRangeGetBufferSize_16s_C1R>::hist,
- 0,
- NppHistogramRangeC1<CV_32F, nppiHistogramRange_32f_C1R, nppiHistogramRangeGetBufferSize_32f_C1R>::hist
- };
-
- hist_callers[src.depth()](src, hist, levels, buf, StreamAccessor::getStream(stream));
-}
-
-void cv::gpu::histRange(const GpuMat& src, GpuMat hist[4], const GpuMat levels[4], Stream& stream)
-{
- GpuMat buf;
- histRange(src, hist, levels, buf, stream);
-}
-
-void cv::gpu::histRange(const GpuMat& src, GpuMat hist[4], const GpuMat levels[4], GpuMat& buf, Stream& stream)
-{
- CV_Assert(src.type() == CV_8UC4 || src.type() == CV_16UC4 || src.type() == CV_16SC4 || src.type() == CV_32FC4);
-
- typedef void (*hist_t)(const GpuMat& src, GpuMat hist[4], const GpuMat levels[4], GpuMat& buf, cudaStream_t stream);
- static const hist_t hist_callers[] =
- {
- NppHistogramRangeC4<CV_8U , nppiHistogramRange_8u_C4R , nppiHistogramRangeGetBufferSize_8u_C4R >::hist,
- 0,
- NppHistogramRangeC4<CV_16U, nppiHistogramRange_16u_C4R, nppiHistogramRangeGetBufferSize_16u_C4R>::hist,
- NppHistogramRangeC4<CV_16S, nppiHistogramRange_16s_C4R, nppiHistogramRangeGetBufferSize_16s_C4R>::hist,
- 0,
- NppHistogramRangeC4<CV_32F, nppiHistogramRange_32f_C4R, nppiHistogramRangeGetBufferSize_32f_C4R>::hist
- };
-
- hist_callers[src.depth()](src, hist, levels, buf, StreamAccessor::getStream(stream));
-}
-
-namespace hist
-{
- void histogram256(PtrStepSzb src, int* hist, cudaStream_t stream);
- void equalizeHist(PtrStepSzb src, PtrStepSzb dst, const int* lut, cudaStream_t stream);
-}
-
-void cv::gpu::calcHist(const GpuMat& src, GpuMat& hist, Stream& stream)
-{
- CV_Assert(src.type() == CV_8UC1);
-
- hist.create(1, 256, CV_32SC1);
- hist.setTo(Scalar::all(0));
-
- hist::histogram256(src, hist.ptr<int>(), StreamAccessor::getStream(stream));
-}
-
-void cv::gpu::equalizeHist(const GpuMat& src, GpuMat& dst, Stream& stream)
-{
- GpuMat hist;
- GpuMat buf;
- equalizeHist(src, dst, hist, buf, stream);
-}
-
-void cv::gpu::equalizeHist(const GpuMat& src, GpuMat& dst, GpuMat& hist, GpuMat& buf, Stream& s)
-{
- CV_Assert(src.type() == CV_8UC1);
-
- dst.create(src.size(), src.type());
-
- int intBufSize;
- nppSafeCall( nppsIntegralGetBufferSize_32s(256, &intBufSize) );
-
- ensureSizeIsEnough(1, intBufSize + 256 * sizeof(int), CV_8UC1, buf);
-
- GpuMat intBuf(1, intBufSize, CV_8UC1, buf.ptr());
- GpuMat lut(1, 256, CV_32S, buf.ptr() + intBufSize);
-
- calcHist(src, hist, s);
-
- cudaStream_t stream = StreamAccessor::getStream(s);
-
- NppStreamHandler h(stream);
-
- nppSafeCall( nppsIntegral_32s(hist.ptr<Npp32s>(), lut.ptr<Npp32s>(), 256, intBuf.ptr<Npp8u>()) );
-
- hist::equalizeHist(src, dst, lut.ptr<int>(), stream);
-}
-
-////////////////////////////////////////////////////////////////////////
-// cornerHarris & minEgenVal
-
-namespace cv { namespace gpu { namespace cudev
-{
- namespace imgproc
- {
- void cornerHarris_gpu(int block_size, float k, PtrStepSzf Dx, PtrStepSzf Dy, PtrStepSzf dst, int border_type, cudaStream_t stream);
- void cornerMinEigenVal_gpu(int block_size, PtrStepSzf Dx, PtrStepSzf Dy, PtrStepSzf dst, int border_type, cudaStream_t stream);
- }
-}}}
-
-namespace
-{
- void extractCovData(const GpuMat& src, GpuMat& Dx, GpuMat& Dy, GpuMat& buf, int blockSize, int ksize, int borderType, Stream& stream)
- {
- double scale = static_cast<double>(1 << ((ksize > 0 ? ksize : 3) - 1)) * blockSize;
-
- if (ksize < 0)
- scale *= 2.;
-
- if (src.depth() == CV_8U)
- scale *= 255.;
-
- scale = 1./scale;
-
- Dx.create(src.size(), CV_32F);
- Dy.create(src.size(), CV_32F);
-
- if (ksize > 0)
- {
- Sobel(src, Dx, CV_32F, 1, 0, buf, ksize, scale, borderType, -1, stream);
- Sobel(src, Dy, CV_32F, 0, 1, buf, ksize, scale, borderType, -1, stream);
- }
- else
- {
- Scharr(src, Dx, CV_32F, 1, 0, buf, scale, borderType, -1, stream);
- Scharr(src, Dy, CV_32F, 0, 1, buf, scale, borderType, -1, stream);
- }
- }
-}
-
-void cv::gpu::cornerHarris(const GpuMat& src, GpuMat& dst, int blockSize, int ksize, double k, int borderType)
-{
- GpuMat Dx, Dy;
- cornerHarris(src, dst, Dx, Dy, blockSize, ksize, k, borderType);
-}
-
-void cv::gpu::cornerHarris(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, int blockSize, int ksize, double k, int borderType)
-{
- GpuMat buf;
- cornerHarris(src, dst, Dx, Dy, buf, blockSize, ksize, k, borderType);
-}
-
-void cv::gpu::cornerHarris(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, GpuMat& buf, int blockSize, int ksize, double k, int borderType, Stream& stream)
-{
- using namespace cv::gpu::cudev::imgproc;
-
- CV_Assert(borderType == cv::BORDER_REFLECT101 || borderType == cv::BORDER_REPLICATE || borderType == cv::BORDER_REFLECT);
-
- int gpuBorderType;
- CV_Assert(tryConvertToGpuBorderType(borderType, gpuBorderType));
-
- extractCovData(src, Dx, Dy, buf, blockSize, ksize, borderType, stream);
-
- dst.create(src.size(), CV_32F);
-
- cornerHarris_gpu(blockSize, static_cast<float>(k), Dx, Dy, dst, gpuBorderType, StreamAccessor::getStream(stream));
-}
-
-void cv::gpu::cornerMinEigenVal(const GpuMat& src, GpuMat& dst, int blockSize, int ksize, int borderType)
-{
- GpuMat Dx, Dy;
- cornerMinEigenVal(src, dst, Dx, Dy, blockSize, ksize, borderType);
-}
-
-void cv::gpu::cornerMinEigenVal(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, int blockSize, int ksize, int borderType)
-{
- GpuMat buf;
- cornerMinEigenVal(src, dst, Dx, Dy, buf, blockSize, ksize, borderType);
-}
-
-void cv::gpu::cornerMinEigenVal(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, GpuMat& buf, int blockSize, int ksize, int borderType, Stream& stream)
-{
- using namespace ::cv::gpu::cudev::imgproc;
-
- CV_Assert(borderType == cv::BORDER_REFLECT101 || borderType == cv::BORDER_REPLICATE || borderType == cv::BORDER_REFLECT);
-
- int gpuBorderType;
- CV_Assert(tryConvertToGpuBorderType(borderType, gpuBorderType));
-
- extractCovData(src, Dx, Dy, buf, blockSize, ksize, borderType, stream);
-
- dst.create(src.size(), CV_32F);
-
- cornerMinEigenVal_gpu(blockSize, Dx, Dy, dst, gpuBorderType, StreamAccessor::getStream(stream));
-}
-
-//////////////////////////////////////////////////////////////////////////////
-// mulSpectrums
-
-namespace cv { namespace gpu { namespace cudev
-{
- namespace imgproc
- {
- void mulSpectrums(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, PtrStepSz<cufftComplex> c, cudaStream_t stream);
-
- void mulSpectrums_CONJ(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, PtrStepSz<cufftComplex> c, cudaStream_t stream);
- }
-}}}
-
-void cv::gpu::mulSpectrums(const GpuMat& a, const GpuMat& b, GpuMat& c, int flags, bool conjB, Stream& stream)
-{
- (void)flags;
- using namespace ::cv::gpu::cudev::imgproc;
-
- typedef void (*Caller)(const PtrStep<cufftComplex>, const PtrStep<cufftComplex>, PtrStepSz<cufftComplex>, cudaStream_t stream);
-
- static Caller callers[] = { cudev::imgproc::mulSpectrums, cudev::imgproc::mulSpectrums_CONJ };
-
- CV_Assert(a.type() == b.type() && a.type() == CV_32FC2);
- CV_Assert(a.size() == b.size());
-
- c.create(a.size(), CV_32FC2);
-
- Caller caller = callers[(int)conjB];
- caller(a, b, c, StreamAccessor::getStream(stream));
-}
-
-//////////////////////////////////////////////////////////////////////////////
-// mulAndScaleSpectrums
-
-namespace cv { namespace gpu { namespace cudev
-{
- namespace imgproc
- {
- void mulAndScaleSpectrums(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, float scale, PtrStepSz<cufftComplex> c, cudaStream_t stream);
-
- void mulAndScaleSpectrums_CONJ(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, float scale, PtrStepSz<cufftComplex> c, cudaStream_t stream);
- }
-}}}
-
-void cv::gpu::mulAndScaleSpectrums(const GpuMat& a, const GpuMat& b, GpuMat& c, int flags, float scale, bool conjB, Stream& stream)
-{
- (void)flags;
- using namespace ::cv::gpu::cudev::imgproc;
-
- typedef void (*Caller)(const PtrStep<cufftComplex>, const PtrStep<cufftComplex>, float scale, PtrStepSz<cufftComplex>, cudaStream_t stream);
- static Caller callers[] = { cudev::imgproc::mulAndScaleSpectrums, cudev::imgproc::mulAndScaleSpectrums_CONJ };
-
- CV_Assert(a.type() == b.type() && a.type() == CV_32FC2);
- CV_Assert(a.size() == b.size());
-
- c.create(a.size(), CV_32FC2);
-
- Caller caller = callers[(int)conjB];
- caller(a, b, scale, c, StreamAccessor::getStream(stream));
-}
-
-//////////////////////////////////////////////////////////////////////////////
-// dft
-
-void cv::gpu::dft(const GpuMat& src, GpuMat& dst, Size dft_size, int flags, Stream& stream)
-{
-#ifndef HAVE_CUFFT
-
- OPENCV_GPU_UNUSED(src);
- OPENCV_GPU_UNUSED(dst);
- OPENCV_GPU_UNUSED(dft_size);
- OPENCV_GPU_UNUSED(flags);
- OPENCV_GPU_UNUSED(stream);
-
- throw_no_cuda();
-
-#else
-
- CV_Assert(src.type() == CV_32F || src.type() == CV_32FC2);
-
- // We don't support unpacked output (in the case of real input)
- CV_Assert(!(flags & DFT_COMPLEX_OUTPUT));
-
- bool is_1d_input = (dft_size.height == 1) || (dft_size.width == 1);
- int is_row_dft = flags & DFT_ROWS;
- int is_scaled_dft = flags & DFT_SCALE;
- int is_inverse = flags & DFT_INVERSE;
- bool is_complex_input = src.channels() == 2;
- bool is_complex_output = !(flags & DFT_REAL_OUTPUT);
-
- // We don't support real-to-real transform
- CV_Assert(is_complex_input || is_complex_output);
-
- GpuMat src_data;
-
- // Make sure here we work with the continuous input,
- // as CUFFT can't handle gaps
- src_data = src;
- createContinuous(src.rows, src.cols, src.type(), src_data);
- if (src_data.data != src.data)
- src.copyTo(src_data);
-
- Size dft_size_opt = dft_size;
- if (is_1d_input && !is_row_dft)
- {
- // If the source matrix is single column handle it as single row
- dft_size_opt.width = std::max(dft_size.width, dft_size.height);
- dft_size_opt.height = std::min(dft_size.width, dft_size.height);
- }
-
- cufftType dft_type = CUFFT_R2C;
- if (is_complex_input)
- dft_type = is_complex_output ? CUFFT_C2C : CUFFT_C2R;
-
- CV_Assert(dft_size_opt.width > 1);
-
- cufftHandle plan;
- if (is_1d_input || is_row_dft)
- cufftPlan1d(&plan, dft_size_opt.width, dft_type, dft_size_opt.height);
- else
- cufftPlan2d(&plan, dft_size_opt.height, dft_size_opt.width, dft_type);
-
- cufftSafeCall( cufftSetStream(plan, StreamAccessor::getStream(stream)) );
-
- if (is_complex_input)
- {
- if (is_complex_output)
- {
- createContinuous(dft_size, CV_32FC2, dst);
- cufftSafeCall(cufftExecC2C(
- plan, src_data.ptr<cufftComplex>(), dst.ptr<cufftComplex>(),
- is_inverse ? CUFFT_INVERSE : CUFFT_FORWARD));
- }
- else
- {
- createContinuous(dft_size, CV_32F, dst);
- cufftSafeCall(cufftExecC2R(
- plan, src_data.ptr<cufftComplex>(), dst.ptr<cufftReal>()));
- }
- }
- else
- {
- // We could swap dft_size for efficiency. Here we must reflect it
- if (dft_size == dft_size_opt)
- createContinuous(Size(dft_size.width / 2 + 1, dft_size.height), CV_32FC2, dst);
- else
- createContinuous(Size(dft_size.width, dft_size.height / 2 + 1), CV_32FC2, dst);
-
- cufftSafeCall(cufftExecR2C(
- plan, src_data.ptr<cufftReal>(), dst.ptr<cufftComplex>()));
- }
-
- cufftSafeCall(cufftDestroy(plan));
-
- if (is_scaled_dft)
- multiply(dst, Scalar::all(1. / dft_size.area()), dst, 1, -1, stream);
-
-#endif
-}
-
-//////////////////////////////////////////////////////////////////////////////
-// convolve
-
-void cv::gpu::ConvolveBuf::create(Size image_size, Size templ_size)
-{
- result_size = Size(image_size.width - templ_size.width + 1,
- image_size.height - templ_size.height + 1);
-
- block_size = user_block_size;
- if (user_block_size.width == 0 || user_block_size.height == 0)
- block_size = estimateBlockSize(result_size, templ_size);
-
- dft_size.width = 1 << int(ceil(std::log(block_size.width + templ_size.width - 1.) / std::log(2.)));
- dft_size.height = 1 << int(ceil(std::log(block_size.height + templ_size.height - 1.) / std::log(2.)));
-
- // CUFFT has hard-coded kernels for power-of-2 sizes (up to 8192),
- // see CUDA Toolkit 4.1 CUFFT Library Programming Guide
- if (dft_size.width > 8192)
- dft_size.width = getOptimalDFTSize(block_size.width + templ_size.width - 1);
- if (dft_size.height > 8192)
- dft_size.height = getOptimalDFTSize(block_size.height + templ_size.height - 1);
-
- // To avoid wasting time doing small DFTs
- dft_size.width = std::max(dft_size.width, 512);
- dft_size.height = std::max(dft_size.height, 512);
-
- createContinuous(dft_size, CV_32F, image_block);
- createContinuous(dft_size, CV_32F, templ_block);
- createContinuous(dft_size, CV_32F, result_data);
-
- spect_len = dft_size.height * (dft_size.width / 2 + 1);
- createContinuous(1, spect_len, CV_32FC2, image_spect);
- createContinuous(1, spect_len, CV_32FC2, templ_spect);
- createContinuous(1, spect_len, CV_32FC2, result_spect);
-
- // Use maximum result matrix block size for the estimated DFT block size
- block_size.width = std::min(dft_size.width - templ_size.width + 1, result_size.width);
- block_size.height = std::min(dft_size.height - templ_size.height + 1, result_size.height);
-}
-
-
-Size cv::gpu::ConvolveBuf::estimateBlockSize(Size result_size, Size /*templ_size*/)
-{
- int width = (result_size.width + 2) / 3;
- int height = (result_size.height + 2) / 3;
- width = std::min(width, result_size.width);
- height = std::min(height, result_size.height);
- return Size(width, height);
-}
-
-
-void cv::gpu::convolve(const GpuMat& image, const GpuMat& templ, GpuMat& result, bool ccorr)
-{
- ConvolveBuf buf;
- convolve(image, templ, result, ccorr, buf);
-}
-
-void cv::gpu::convolve(const GpuMat& image, const GpuMat& templ, GpuMat& result, bool ccorr, ConvolveBuf& buf, Stream& stream)
-{
- using namespace ::cv::gpu::cudev::imgproc;
-
-#ifndef HAVE_CUFFT
- throw_no_cuda();
-#else
- CV_Assert(image.type() == CV_32F);
- CV_Assert(templ.type() == CV_32F);
-
- buf.create(image.size(), templ.size());
- result.create(buf.result_size, CV_32F);
-
- Size& block_size = buf.block_size;
- Size& dft_size = buf.dft_size;
-
- GpuMat& image_block = buf.image_block;
- GpuMat& templ_block = buf.templ_block;
- GpuMat& result_data = buf.result_data;
-
- GpuMat& image_spect = buf.image_spect;
- GpuMat& templ_spect = buf.templ_spect;
- GpuMat& result_spect = buf.result_spect;
-
- cufftHandle planR2C, planC2R;
- cufftSafeCall(cufftPlan2d(&planC2R, dft_size.height, dft_size.width, CUFFT_C2R));
- cufftSafeCall(cufftPlan2d(&planR2C, dft_size.height, dft_size.width, CUFFT_R2C));
-
- cufftSafeCall( cufftSetStream(planR2C, StreamAccessor::getStream(stream)) );
- cufftSafeCall( cufftSetStream(planC2R, StreamAccessor::getStream(stream)) );
-
- GpuMat templ_roi(templ.size(), CV_32F, templ.data, templ.step);
- copyMakeBorder(templ_roi, templ_block, 0, templ_block.rows - templ_roi.rows, 0,
- templ_block.cols - templ_roi.cols, 0, Scalar(), stream);
-
- cufftSafeCall(cufftExecR2C(planR2C, templ_block.ptr<cufftReal>(),
- templ_spect.ptr<cufftComplex>()));
-
- // Process all blocks of the result matrix
- for (int y = 0; y < result.rows; y += block_size.height)
- {
- for (int x = 0; x < result.cols; x += block_size.width)
- {
- Size image_roi_size(std::min(x + dft_size.width, image.cols) - x,
- std::min(y + dft_size.height, image.rows) - y);
- GpuMat image_roi(image_roi_size, CV_32F, (void*)(image.ptr<float>(y) + x),
- image.step);
- copyMakeBorder(image_roi, image_block, 0, image_block.rows - image_roi.rows,
- 0, image_block.cols - image_roi.cols, 0, Scalar(), stream);
-
- cufftSafeCall(cufftExecR2C(planR2C, image_block.ptr<cufftReal>(),
- image_spect.ptr<cufftComplex>()));
- mulAndScaleSpectrums(image_spect, templ_spect, result_spect, 0,
- 1.f / dft_size.area(), ccorr, stream);
- cufftSafeCall(cufftExecC2R(planC2R, result_spect.ptr<cufftComplex>(),
- result_data.ptr<cufftReal>()));
-
- Size result_roi_size(std::min(x + block_size.width, result.cols) - x,
- std::min(y + block_size.height, result.rows) - y);
- GpuMat result_roi(result_roi_size, result.type(),
- (void*)(result.ptr<float>(y) + x), result.step);
- GpuMat result_block(result_roi_size, result_data.type(),
- result_data.ptr(), result_data.step);
-
- if (stream)
- stream.enqueueCopy(result_block, result_roi);
- else
- result_block.copyTo(result_roi);
- }
- }
-
- cufftSafeCall(cufftDestroy(planR2C));
- cufftSafeCall(cufftDestroy(planC2R));
-#endif
-}
-
-
-//////////////////////////////////////////////////////////////////////////////
-// Canny
-
-void cv::gpu::CannyBuf::create(const Size& image_size, int apperture_size)
-{
- if (apperture_size > 0)
- {
- ensureSizeIsEnough(image_size, CV_32SC1, dx);
- ensureSizeIsEnough(image_size, CV_32SC1, dy);
-
- if (apperture_size != 3)
- {
- filterDX = createDerivFilter_GPU(CV_8UC1, CV_32S, 1, 0, apperture_size, BORDER_REPLICATE);
- filterDY = createDerivFilter_GPU(CV_8UC1, CV_32S, 0, 1, apperture_size, BORDER_REPLICATE);
- }
- }
-
- ensureSizeIsEnough(image_size, CV_32FC1, mag);
- ensureSizeIsEnough(image_size, CV_32SC1, map);
-
- ensureSizeIsEnough(1, image_size.area(), CV_16UC2, st1);
- ensureSizeIsEnough(1, image_size.area(), CV_16UC2, st2);
-}
-
-void cv::gpu::CannyBuf::release()
-{
- dx.release();
- dy.release();
- mag.release();
- map.release();
- st1.release();
- st2.release();
-}
-
-namespace canny
-{
- void calcMagnitude(PtrStepSzb srcWhole, int xoff, int yoff, PtrStepSzi dx, PtrStepSzi dy, PtrStepSzf mag, bool L2Grad);
- void calcMagnitude(PtrStepSzi dx, PtrStepSzi dy, PtrStepSzf mag, bool L2Grad);
-
- void calcMap(PtrStepSzi dx, PtrStepSzi dy, PtrStepSzf mag, PtrStepSzi map, float low_thresh, float high_thresh);
-
- void edgesHysteresisLocal(PtrStepSzi map, ushort2* st1);
-
- void edgesHysteresisGlobal(PtrStepSzi map, ushort2* st1, ushort2* st2);
-
- void getEdges(PtrStepSzi map, PtrStepSzb dst);
-}
-
-namespace
-{
- void CannyCaller(const GpuMat& dx, const GpuMat& dy, CannyBuf& buf, GpuMat& dst, float low_thresh, float high_thresh)
- {
- using namespace canny;
-
- buf.map.setTo(Scalar::all(0));
- calcMap(dx, dy, buf.mag, buf.map, low_thresh, high_thresh);
-
- edgesHysteresisLocal(buf.map, buf.st1.ptr<ushort2>());
-
- edgesHysteresisGlobal(buf.map, buf.st1.ptr<ushort2>(), buf.st2.ptr<ushort2>());
-
- getEdges(buf.map, dst);
- }
-}
-
-void cv::gpu::Canny(const GpuMat& src, GpuMat& dst, double low_thresh, double high_thresh, int apperture_size, bool L2gradient)
-{
- CannyBuf buf;
- Canny(src, buf, dst, low_thresh, high_thresh, apperture_size, L2gradient);
-}
-
-void cv::gpu::Canny(const GpuMat& src, CannyBuf& buf, GpuMat& dst, double low_thresh, double high_thresh, int apperture_size, bool L2gradient)
-{
- using namespace canny;
-
- CV_Assert(src.type() == CV_8UC1);
-
- if (!deviceSupports(SHARED_ATOMICS))
- CV_Error(cv::Error::StsNotImplemented, "The device doesn't support shared atomics");
-
- if( low_thresh > high_thresh )
- std::swap( low_thresh, high_thresh);
-
- dst.create(src.size(), CV_8U);
- buf.create(src.size(), apperture_size);
-
- if (apperture_size == 3)
- {
- Size wholeSize;
- Point ofs;
- src.locateROI(wholeSize, ofs);
- GpuMat srcWhole(wholeSize, src.type(), src.datastart, src.step);
-
- calcMagnitude(srcWhole, ofs.x, ofs.y, buf.dx, buf.dy, buf.mag, L2gradient);
- }
- else
- {
- buf.filterDX->apply(src, buf.dx, Rect(0, 0, src.cols, src.rows));
- buf.filterDY->apply(src, buf.dy, Rect(0, 0, src.cols, src.rows));
-
- calcMagnitude(buf.dx, buf.dy, buf.mag, L2gradient);
- }
-
- CannyCaller(buf.dx, buf.dy, buf, dst, static_cast<float>(low_thresh), static_cast<float>(high_thresh));
-}
-
-void cv::gpu::Canny(const GpuMat& dx, const GpuMat& dy, GpuMat& dst, double low_thresh, double high_thresh, bool L2gradient)
-{
- CannyBuf buf;
- Canny(dx, dy, buf, dst, low_thresh, high_thresh, L2gradient);
-}
-
-void cv::gpu::Canny(const GpuMat& dx, const GpuMat& dy, CannyBuf& buf, GpuMat& dst, double low_thresh, double high_thresh, bool L2gradient)
-{
- using namespace canny;
-
- CV_Assert(TargetArchs::builtWith(SHARED_ATOMICS) && DeviceInfo().supports(SHARED_ATOMICS));
- CV_Assert(dx.type() == CV_32SC1 && dy.type() == CV_32SC1 && dx.size() == dy.size());
-
- if( low_thresh > high_thresh )
- std::swap( low_thresh, high_thresh);
-
- dst.create(dx.size(), CV_8U);
- buf.create(dx.size(), -1);
-
- calcMagnitude(dx, dy, buf.mag, L2gradient);
-
- CannyCaller(dx, dy, buf, dst, static_cast<float>(low_thresh), static_cast<float>(high_thresh));
-}
-
-////////////////////////////////////////////////////////////////////////
-// CLAHE
-
-namespace clahe
-{
- void calcLut(PtrStepSzb src, PtrStepb lut, int tilesX, int tilesY, int2 tileSize, int clipLimit, float lutScale, cudaStream_t stream);
- void transform(PtrStepSzb src, PtrStepSzb dst, PtrStepb lut, int tilesX, int tilesY, int2 tileSize, cudaStream_t stream);
-}
-
-namespace
-{
- class CLAHE_Impl : public cv::gpu::CLAHE
- {
- public:
- CLAHE_Impl(double clipLimit = 40.0, int tilesX = 8, int tilesY = 8);
-
- cv::AlgorithmInfo* info() const;
-
- void apply(cv::InputArray src, cv::OutputArray dst);
- void apply(InputArray src, OutputArray dst, Stream& stream);
-
- void setClipLimit(double clipLimit);
- double getClipLimit() const;
-
- void setTilesGridSize(cv::Size tileGridSize);
- cv::Size getTilesGridSize() const;
-
- void collectGarbage();
-
- private:
- double clipLimit_;
- int tilesX_;
- int tilesY_;
-
- GpuMat srcExt_;
- GpuMat lut_;
- };
-
- CLAHE_Impl::CLAHE_Impl(double clipLimit, int tilesX, int tilesY) :
- clipLimit_(clipLimit), tilesX_(tilesX), tilesY_(tilesY)
- {
- }
-
- CV_INIT_ALGORITHM(CLAHE_Impl, "CLAHE_GPU",
- obj.info()->addParam(obj, "clipLimit", obj.clipLimit_);
- obj.info()->addParam(obj, "tilesX", obj.tilesX_);
- obj.info()->addParam(obj, "tilesY", obj.tilesY_))
-
- void CLAHE_Impl::apply(cv::InputArray _src, cv::OutputArray _dst)
- {
- apply(_src, _dst, Stream::Null());
- }
-
- void CLAHE_Impl::apply(InputArray _src, OutputArray _dst, Stream& s)
- {
- GpuMat src = _src.getGpuMat();
-
- CV_Assert( src.type() == CV_8UC1 );
-
- _dst.create( src.size(), src.type() );
- GpuMat dst = _dst.getGpuMat();
-
- const int histSize = 256;
-
- ensureSizeIsEnough(tilesX_ * tilesY_, histSize, CV_8UC1, lut_);
-
- cudaStream_t stream = StreamAccessor::getStream(s);
-
- cv::Size tileSize;
- GpuMat srcForLut;
-
- if (src.cols % tilesX_ == 0 && src.rows % tilesY_ == 0)
- {
- tileSize = cv::Size(src.cols / tilesX_, src.rows / tilesY_);
- srcForLut = src;
- }
- else
- {
- cv::gpu::copyMakeBorder(src, srcExt_, 0, tilesY_ - (src.rows % tilesY_), 0, tilesX_ - (src.cols % tilesX_), cv::BORDER_REFLECT_101, cv::Scalar(), s);
-
- tileSize = cv::Size(srcExt_.cols / tilesX_, srcExt_.rows / tilesY_);
- srcForLut = srcExt_;
- }
-
- const int tileSizeTotal = tileSize.area();
- const float lutScale = static_cast<float>(histSize - 1) / tileSizeTotal;
-
- int clipLimit = 0;
- if (clipLimit_ > 0.0)
- {
- clipLimit = static_cast<int>(clipLimit_ * tileSizeTotal / histSize);
- clipLimit = std::max(clipLimit, 1);
- }
-
- clahe::calcLut(srcForLut, lut_, tilesX_, tilesY_, make_int2(tileSize.width, tileSize.height), clipLimit, lutScale, stream);
-
- clahe::transform(src, dst, lut_, tilesX_, tilesY_, make_int2(tileSize.width, tileSize.height), stream);
- }
-
- void CLAHE_Impl::setClipLimit(double clipLimit)
- {
- clipLimit_ = clipLimit;
- }
-
- double CLAHE_Impl::getClipLimit() const
- {
- return clipLimit_;
- }
-
- void CLAHE_Impl::setTilesGridSize(cv::Size tileGridSize)
- {
- tilesX_ = tileGridSize.width;
- tilesY_ = tileGridSize.height;
- }
-
- cv::Size CLAHE_Impl::getTilesGridSize() const
- {
- return cv::Size(tilesX_, tilesY_);
- }
-
- void CLAHE_Impl::collectGarbage()
- {
- srcExt_.release();
- lut_.release();
- }
-}
-
-cv::Ptr<cv::gpu::CLAHE> cv::gpu::createCLAHE(double clipLimit, cv::Size tileGridSize)
-{
- return new CLAHE_Impl(clipLimit, tileGridSize.width, tileGridSize.height);
-}
-
-#endif /* !defined (HAVE_CUDA) */
#ifndef __OPENCV_PRECOMP_H__
#define __OPENCV_PRECOMP_H__
-#if defined _MSC_VER && _MSC_VER >= 1200
- #pragma warning( disable: 4251 4710 4711 4514 4996 )
-#endif
-
-#include <cstring>
-#include <iostream>
-#include <limits>
-#include <vector>
-#include <algorithm>
-#include <sstream>
-#include <exception>
-#include <iterator>
-#include <functional>
-#include <utility>
-#include <deque>
-#include <stdexcept>
-#include <memory>
-
-#include "opencv2/core.hpp"
-#include "opencv2/core/utility.hpp"
#include "opencv2/gpu.hpp"
-#include "opencv2/imgproc.hpp"
+#include "opencv2/gpuarithm.hpp"
+#include "opencv2/gpuwarping.hpp"
#include "opencv2/calib3d.hpp"
-#include "opencv2/video.hpp"
+#include "opencv2/objdetect.hpp"
-#include "opencv2/core/private.hpp"
#include "opencv2/core/gpu_private.hpp"
-#ifdef HAVE_CUDA
- #ifdef HAVE_CUFFT
- #include <cufft.h>
- #endif
-
- #ifdef HAVE_CUBLAS
- #include <cublas.h>
- #endif
-
- #ifdef HAVE_NVCUVID
- #include <nvcuvid.h>
+#include "opencv2/opencv_modules.hpp"
- #ifdef WIN32
- #include <windows.h>
- #undef small
- #undef min
- #undef max
- #undef abs
-
- #include <NVEncoderAPI.h>
- #endif
- #endif
-
- #include "internal_shared.hpp"
- #include "opencv2/core/stream_accessor.hpp"
-
- #include "nvidia/core/NCV.hpp"
- #include "nvidia/NPP_staging/NPP_staging.hpp"
- #include "nvidia/NCVHaarObjectDetection.hpp"
- #include "nvidia/NCVBroxOpticalFlow.hpp"
-#endif /* defined(HAVE_CUDA) */
+#ifdef HAVE_OPENCV_GPULEGACY
+# include "opencv2/gpulegacy/private.hpp"
+#endif
#endif /* __OPENCV_PRECOMP_H__ */
+++ /dev/null
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-// By downloading, copying, installing or using the software you agree to this license.
-// If you do not agree to this license, do not download, install,
-// copy or use the software.
-//
-//
-// License Agreement
-// For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-// * Redistribution's of source code must retain the above copyright notice,
-// this list of conditions and the following disclaimer.
-//
-// * Redistribution's in binary form must reproduce the above copyright notice,
-// this list of conditions and the following disclaimer in the documentation
-// and/or other materials provided with the distribution.
-//
-// * The name of the copyright holders may not be used to endorse or promote products
-// derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#include "precomp.hpp"
-
-using namespace cv;
-
-namespace cv {
-// TODO: conflicts with calib3d.hpp : filterSpeckles, should be removed ?
-
-//! Speckle filtering - filters small connected components on diparity image.
-//! It sets pixel (x,y) to newVal if it coresponds to small CC with size < maxSpeckleSize.
-//! Threshold for border between CC is diffThreshold;
- CV_EXPORTS void filterSpeckles(Mat& img, uchar newVal, int maxSpeckleSize, uchar diffThreshold, Mat& buf);
-}
-
-typedef Point_<short> Point2s;
-
-void cv::filterSpeckles( Mat& img, uchar newVal, int maxSpeckleSize, uchar maxDiff, Mat& _buf)
-{
- int MaxD = 1024;
- int WinSz = 64;
-
- int bufSize0 = (MaxD + 2)*sizeof(int) + (img.rows+WinSz+2)*MaxD*sizeof(int) +
- (img.rows + WinSz + 2)*sizeof(int) +
- (img.rows+WinSz+2)*MaxD*(WinSz+1)*sizeof(uchar) + 256;
- int bufSize1 = (img.cols + 9 + 2) * sizeof(int) + 256;
- int bufSz = std::max(bufSize0 * 1, bufSize1 * 2);
-
- _buf.create(1, bufSz, CV_8U);
-
- CV_Assert( img.type() == CV_8U );
-
- int width = img.cols, height = img.rows, npixels = width*height;
- size_t bufSize = npixels*(int)(sizeof(Point2s) + sizeof(int) + sizeof(uchar));
- if( !_buf.isContinuous() || !_buf.data || _buf.cols*_buf.rows*_buf.elemSize() < bufSize )
- _buf.create(1, (int)bufSize, CV_8U);
-
- uchar* buf = _buf.data;
- int i, j, dstep = (int)(img.step/sizeof(uchar));
- int* labels = (int*)buf;
- buf += npixels*sizeof(labels[0]);
- Point2s* wbuf = (Point2s*)buf;
- buf += npixels*sizeof(wbuf[0]);
- uchar* rtype = (uchar*)buf;
- int curlabel = 0;
-
- // clear out label assignments
- memset(labels, 0, npixels*sizeof(labels[0]));
-
- for( i = 0; i < height; i++ )
- {
- uchar* ds = img.ptr<uchar>(i);
- int* ls = labels + width*i;
-
- for( j = 0; j < width; j++ )
- {
- if( ds[j] != newVal ) // not a bad disparity
- {
- if( ls[j] ) // has a label, check for bad label
- {
- if( rtype[ls[j]] ) // small region, zero out disparity
- ds[j] = (uchar)newVal;
- }
- // no label, assign and propagate
- else
- {
- Point2s* ws = wbuf; // initialize wavefront
- Point2s p((short)j, (short)i); // current pixel
- curlabel++; // next label
- int count = 0; // current region size
- ls[j] = curlabel;
-
- // wavefront propagation
- while( ws >= wbuf ) // wavefront not empty
- {
- count++;
- // put neighbors onto wavefront
- uchar* dpp = &img.at<uchar>(p.y, p.x);
- uchar dp = *dpp;
- int* lpp = labels + width*p.y + p.x;
-
- if( p.x < width-1 && !lpp[+1] && dpp[+1] != newVal && std::abs(dp - dpp[+1]) <= maxDiff )
- {
- lpp[+1] = curlabel;
- *ws++ = Point2s(p.x+1, p.y);
- }
-
- if( p.x > 0 && !lpp[-1] && dpp[-1] != newVal && std::abs(dp - dpp[-1]) <= maxDiff )
- {
- lpp[-1] = curlabel;
- *ws++ = Point2s(p.x-1, p.y);
- }
-
- if( p.y < height-1 && !lpp[+width] && dpp[+dstep] != newVal && std::abs(dp - dpp[+dstep]) <= maxDiff )
- {
- lpp[+width] = curlabel;
- *ws++ = Point2s(p.x, p.y+1);
- }
-
- if( p.y > 0 && !lpp[-width] && dpp[-dstep] != newVal && std::abs(dp - dpp[-dstep]) <= maxDiff )
- {
- lpp[-width] = curlabel;
- *ws++ = Point2s(p.x, p.y-1);
- }
-
- // pop most recent and propagate
- // NB: could try least recent, maybe better convergence
- p = *--ws;
- }
-
- // assign label type
- if( count <= maxSpeckleSize ) // speckle region
- {
- //printf("count = %d\n", count);
- rtype[ls[j]] = 1; // small region label
- ds[j] = (uchar)newVal;
- }
- else
- {
- //printf("count = %d\n", count);
- rtype[ls[j]] = 0; // large region label
- }
- }
- }
- }
- }
-}
-
+++ /dev/null
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-// By downloading, copying, installing or using the software you agree to this license.
-// If you do not agree to this license, do not download, install,
-// copy or use the software.
-//
-//
-// License Agreement
-// For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-// * Redistribution's of source code must retain the above copyright notice,
-// this list of conditions and the following disclaimer.
-//
-// * Redistribution's in binary form must reproduce the above copyright notice,
-// this list of conditions and the following disclaimer in the documentation
-// and/or other materials provided with the distribution.
-//
-// * The name of the copyright holders may not be used to endorse or promote products
-// derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#include "precomp.hpp"
-
-using namespace cv;
-using namespace cv::gpu;
-
-#if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
-
-void cv::gpu::merge(const GpuMat* /*src*/, size_t /*count*/, GpuMat& /*dst*/, Stream& /*stream*/) { throw_no_cuda(); }
-void cv::gpu::merge(const std::vector<GpuMat>& /*src*/, GpuMat& /*dst*/, Stream& /*stream*/) { throw_no_cuda(); }
-void cv::gpu::split(const GpuMat& /*src*/, GpuMat* /*dst*/, Stream& /*stream*/) { throw_no_cuda(); }
-void cv::gpu::split(const GpuMat& /*src*/, std::vector<GpuMat>& /*dst*/, Stream& /*stream*/) { throw_no_cuda(); }
-
-#else /* !defined (HAVE_CUDA) */
-
-namespace cv { namespace gpu { namespace cudev
-{
- namespace split_merge
- {
- void merge_caller(const PtrStepSzb* src, PtrStepSzb& dst, int total_channels, size_t elem_size, const cudaStream_t& stream);
- void split_caller(const PtrStepSzb& src, PtrStepSzb* dst, int num_channels, size_t elem_size1, const cudaStream_t& stream);
- }
-}}}
-
-namespace
-{
- void merge(const GpuMat* src, size_t n, GpuMat& dst, const cudaStream_t& stream)
- {
- using namespace ::cv::gpu::cudev::split_merge;
-
- CV_Assert(src);
- CV_Assert(n > 0);
-
- int depth = src[0].depth();
- Size size = src[0].size();
-
- if (depth == CV_64F)
- {
- if (!deviceSupports(NATIVE_DOUBLE))
- CV_Error(cv::Error::StsUnsupportedFormat, "The device doesn't support double");
- }
-
- bool single_channel_only = true;
- int total_channels = 0;
-
- for (size_t i = 0; i < n; ++i)
- {
- CV_Assert(src[i].size() == size);
- CV_Assert(src[i].depth() == depth);
- single_channel_only = single_channel_only && src[i].channels() == 1;
- total_channels += src[i].channels();
- }
-
- CV_Assert(single_channel_only);
- CV_Assert(total_channels <= 4);
-
- if (total_channels == 1)
- src[0].copyTo(dst);
- else
- {
- dst.create(size, CV_MAKETYPE(depth, total_channels));
-
- PtrStepSzb src_as_devmem[4];
- for(size_t i = 0; i < n; ++i)
- src_as_devmem[i] = src[i];
-
- PtrStepSzb dst_as_devmem(dst);
- merge_caller(src_as_devmem, dst_as_devmem, total_channels, CV_ELEM_SIZE(depth), stream);
- }
- }
-
- void split(const GpuMat& src, GpuMat* dst, const cudaStream_t& stream)
- {
- using namespace ::cv::gpu::cudev::split_merge;
-
- CV_Assert(dst);
-
- int depth = src.depth();
- int num_channels = src.channels();
-
- if (depth == CV_64F)
- {
- if (!deviceSupports(NATIVE_DOUBLE))
- CV_Error(cv::Error::StsUnsupportedFormat, "The device doesn't support double");
- }
-
- if (num_channels == 1)
- {
- src.copyTo(dst[0]);
- return;
- }
-
- for (int i = 0; i < num_channels; ++i)
- dst[i].create(src.size(), depth);
-
- CV_Assert(num_channels <= 4);
-
- PtrStepSzb dst_as_devmem[4];
- for (int i = 0; i < num_channels; ++i)
- dst_as_devmem[i] = dst[i];
-
- PtrStepSzb src_as_devmem(src);
- split_caller(src_as_devmem, dst_as_devmem, num_channels, src.elemSize1(), stream);
- }
-}
-
-void cv::gpu::merge(const GpuMat* src, size_t n, GpuMat& dst, Stream& stream)
-{
- ::merge(src, n, dst, StreamAccessor::getStream(stream));
-}
-
-
-void cv::gpu::merge(const std::vector<GpuMat>& src, GpuMat& dst, Stream& stream)
-{
- ::merge(&src[0], src.size(), dst, StreamAccessor::getStream(stream));
-}
-
-void cv::gpu::split(const GpuMat& src, GpuMat* dst, Stream& stream)
-{
- ::split(src, dst, StreamAccessor::getStream(stream));
-}
-
-void cv::gpu::split(const GpuMat& src, std::vector<GpuMat>& dst, Stream& stream)
-{
- dst.resize(src.channels());
- if(src.channels() > 0)
- ::split(src, &dst[0], StreamAccessor::getStream(stream));
-}
-
-#endif /* !defined (HAVE_CUDA) */
+++ /dev/null
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-// By downloading, copying, installing or using the software you agree to this license.
-// If you do not agree to this license, do not download, install,
-// copy or use the software.
-//
-//
-// License Agreement
-// For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-// * Redistribution's of source code must retain the above copyright notice,
-// this list of conditions and the following disclaimer.
-//
-// * Redistribution's in binary form must reproduce the above copyright notice,
-// this list of conditions and the following disclaimer in the documentation
-// and/or other materials provided with the distribution.
-//
-// * The name of the copyright holders may not be used to endorse or promote products
-// derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#include "thread_wrappers.h"
-
-#if defined(HAVE_CUDA) && defined(HAVE_NVCUVID)
-
-#ifdef WIN32
- #define NOMINMAX
- #include <windows.h>
-#else
- #include <pthread.h>
- #include <unistd.h>
-#endif
-
-#ifdef WIN32
- class cv::gpu::detail::CriticalSection::Impl
- {
- public:
- Impl()
- {
- InitializeCriticalSection(&criticalSection_);
- }
-
- ~Impl()
- {
- DeleteCriticalSection(&criticalSection_);
- }
-
- void enter()
- {
- EnterCriticalSection(&criticalSection_);
- }
-
- void leave()
- {
- LeaveCriticalSection(&criticalSection_);
- }
-
- private:
- CRITICAL_SECTION criticalSection_;
- };
-#else
- class cv::gpu::detail::CriticalSection::Impl
- {
- public:
- Impl()
- {
- pthread_mutexattr_t mutex_attribute;
- pthread_mutexattr_init(&mutex_attribute);
- pthread_mutexattr_settype(&mutex_attribute, PTHREAD_MUTEX_RECURSIVE);
- pthread_mutex_init(&mutex_, 0);
- pthread_mutexattr_destroy(&mutex_attribute);
- }
-
- ~Impl()
- {
- pthread_mutex_destroy(&mutex_);
- }
-
- void enter()
- {
- pthread_mutex_lock(&mutex_);
- }
-
- void leave()
- {
- pthread_mutex_unlock(&mutex_);
- }
-
- private:
- pthread_mutex_t mutex_;
- };
-#endif
-
-cv::gpu::detail::CriticalSection::CriticalSection() :
- impl_(new Impl)
-{
-}
-
-cv::gpu::detail::CriticalSection::~CriticalSection()
-{
-}
-
-void cv::gpu::detail::CriticalSection::enter()
-{
- impl_->enter();
-}
-
-void cv::gpu::detail::CriticalSection::leave()
-{
- impl_->leave();
-}
-
-#ifdef WIN32
- namespace
- {
- struct UserData
- {
- void (*func)(void* userData);
- void* param;
- };
-
- DWORD WINAPI WinThreadFunction(LPVOID lpParam)
- {
- UserData* userData = static_cast<UserData*>(lpParam);
-
- userData->func(userData->param);
-
- return 0;
- }
- }
-
- class cv::gpu::detail::Thread::Impl
- {
- public:
- Impl(void (*func)(void* userData), void* userData)
- {
- userData_.func = func;
- userData_.param = userData;
-
- thread_ = CreateThread(
- NULL, // default security attributes
- 0, // use default stack size
- WinThreadFunction, // thread function name
- &userData_, // argument to thread function
- 0, // use default creation flags
- &threadId_); // returns the thread identifier
- }
-
- ~Impl()
- {
- CloseHandle(thread_);
- }
-
- void wait()
- {
- WaitForSingleObject(thread_, INFINITE);
- }
-
- private:
- UserData userData_;
- HANDLE thread_;
- DWORD threadId_;
- };
-#else
- namespace
- {
- struct UserData
- {
- void (*func)(void* userData);
- void* param;
- };
-
- void* PThreadFunction(void* lpParam)
- {
- UserData* userData = static_cast<UserData*>(lpParam);
-
- userData->func(userData->param);
-
- return 0;
- }
- }
-
- class cv::gpu::detail::Thread::Impl
- {
- public:
- Impl(void (*func)(void* userData), void* userData)
- {
- userData_.func = func;
- userData_.param = userData;
-
- pthread_create(&thread_, NULL, PThreadFunction, &userData_);
- }
-
- ~Impl()
- {
- pthread_detach(thread_);
- }
-
- void wait()
- {
- pthread_join(thread_, NULL);
- }
-
- private:
- pthread_t thread_;
- UserData userData_;
- };
-#endif
-
-cv::gpu::detail::Thread::Thread(void (*func)(void* userData), void* userData) :
- impl_(new Impl(func, userData))
-{
-}
-
-cv::gpu::detail::Thread::~Thread()
-{
-}
-
-void cv::gpu::detail::Thread::wait()
-{
- impl_->wait();
-}
-
-void cv::gpu::detail::Thread::sleep(int ms)
-{
-#ifdef WIN32
- ::Sleep(ms);
-#else
- ::usleep(ms * 1000);
-#endif
-}
-
-#endif // HAVE_CUDA
+++ /dev/null
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-// By downloading, copying, installing or using the software you agree to this license.
-// If you do not agree to this license, do not download, install,
-// copy or use the software.
-//
-//
-// License Agreement
-// For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-// * Redistribution's of source code must retain the above copyright notice,
-// this list of conditions and the following disclaimer.
-//
-// * Redistribution's in binary form must reproduce the above copyright notice,
-// this list of conditions and the following disclaimer in the documentation
-// and/or other materials provided with the distribution.
-//
-// * The name of the copyright holders may not be used to endorse or promote products
-// derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#ifndef __VIDEO_DECODER_H__
-#define __VIDEO_DECODER_H__
-
-#include "precomp.hpp"
-
-#if defined(HAVE_CUDA) && defined(HAVE_NVCUVID)
-
-namespace cv { namespace gpu
-{
- namespace detail
- {
- class VideoDecoder
- {
- public:
- VideoDecoder(const VideoReader_GPU::FormatInfo& videoFormat, CUvideoctxlock lock) : lock_(lock), decoder_(0)
- {
- create(videoFormat);
- }
-
- ~VideoDecoder()
- {
- release();
- }
-
- void create(const VideoReader_GPU::FormatInfo& videoFormat);
- void release();
-
- // Get the code-type currently used.
- cudaVideoCodec codec() const { return createInfo_.CodecType; }
- unsigned long maxDecodeSurfaces() const { return createInfo_.ulNumDecodeSurfaces; }
-
- unsigned long frameWidth() const { return createInfo_.ulWidth; }
- unsigned long frameHeight() const { return createInfo_.ulHeight; }
-
- unsigned long targetWidth() const { return createInfo_.ulTargetWidth; }
- unsigned long targetHeight() const { return createInfo_.ulTargetHeight; }
-
- cudaVideoChromaFormat chromaFormat() const { return createInfo_.ChromaFormat; }
-
- bool decodePicture(CUVIDPICPARAMS* picParams)
- {
- return cuvidDecodePicture(decoder_, picParams) == CUDA_SUCCESS;
- }
-
- cv::gpu::GpuMat mapFrame(int picIdx, CUVIDPROCPARAMS& videoProcParams)
- {
- CUdeviceptr ptr;
- unsigned int pitch;
-
- cuSafeCall( cuvidMapVideoFrame(decoder_, picIdx, &ptr, &pitch, &videoProcParams) );
-
- return GpuMat(targetHeight() * 3 / 2, targetWidth(), CV_8UC1, (void*) ptr, pitch);
- }
-
- void unmapFrame(cv::gpu::GpuMat& frame)
- {
- cuSafeCall( cuvidUnmapVideoFrame(decoder_, (CUdeviceptr) frame.data) );
- frame.release();
- }
-
- private:
- VideoDecoder(const VideoDecoder&);
- VideoDecoder& operator =(const VideoDecoder&);
-
- CUvideoctxlock lock_;
- CUVIDDECODECREATEINFO createInfo_;
- CUvideodecoder decoder_;
- };
- }
-}}
-
-#endif // HAVE_CUDA
-
-#endif // __VIDEO_DECODER_H__
+++ /dev/null
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-// By downloading, copying, installing or using the software you agree to this license.
-// If you do not agree to this license, do not download, install,
-// copy or use the software.
-//
-//
-// License Agreement
-// For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-// * Redistribution's of source code must retain the above copyright notice,
-// this list of conditions and the following disclaimer.
-//
-// * Redistribution's in binary form must reproduce the above copyright notice,
-// this list of conditions and the following disclaimer in the documentation
-// and/or other materials provided with the distribution.
-//
-// * The name of the copyright holders may not be used to endorse or promote products
-// derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#ifndef __VIDEO_PARSER_H__
-#define __VIDEO_PARSER_H__
-
-#include "precomp.hpp"
-
-#include "frame_queue.h"
-#include "video_decoder.h"
-
-#if defined(HAVE_CUDA) && defined(HAVE_NVCUVID)
-
-namespace cv { namespace gpu
-{
- namespace detail
- {
- class VideoParser
- {
- public:
- VideoParser(VideoDecoder* videoDecoder, FrameQueue* frameQueue);
-
- ~VideoParser()
- {
- cuvidDestroyVideoParser(parser_);
- }
-
- bool parseVideoData(const unsigned char* data, size_t size, bool endOfStream);
-
- bool hasError() const { return hasError_; }
-
- private:
- VideoParser(const VideoParser&);
- VideoParser& operator =(const VideoParser&);
-
- VideoDecoder* videoDecoder_;
- FrameQueue* frameQueue_;
- CUvideoparser parser_;
- int unparsedPackets_;
- volatile bool hasError_;
-
- // Called when the decoder encounters a video format change (or initial sequence header)
- // This particular implementation of the callback returns 0 in case the video format changes
- // to something different than the original format. Returning 0 causes a stop of the app.
- static int CUDAAPI HandleVideoSequence(void* pUserData, CUVIDEOFORMAT* pFormat);
-
- // Called by the video parser to decode a single picture
- // Since the parser will deliver data as fast as it can, we need to make sure that the picture
- // index we're attempting to use for decode is no longer used for display
- static int CUDAAPI HandlePictureDecode(void* pUserData, CUVIDPICPARAMS* pPicParams);
-
- // Called by the video parser to display a video frame (in the case of field pictures, there may be
- // 2 decode calls per 1 display call, since two fields make up one frame)
- static int CUDAAPI HandlePictureDisplay(void* pUserData, CUVIDPARSERDISPINFO* pPicParams);
- };
- }
-}}
-
-#endif // HAVE_CUDA
-
-#endif // __VIDEO_PARSER_H__
using namespace cvtest;
-//////////////////////////////////////////////////////////////////////////
-// StereoBM
-
-struct StereoBM : testing::TestWithParam<cv::gpu::DeviceInfo>
-{
- cv::gpu::DeviceInfo devInfo;
-
- virtual void SetUp()
- {
- devInfo = GetParam();
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(StereoBM, Regression)
-{
- cv::Mat left_image = readImage("stereobm/aloe-L.png", cv::IMREAD_GRAYSCALE);
- cv::Mat right_image = readImage("stereobm/aloe-R.png", cv::IMREAD_GRAYSCALE);
- cv::Mat disp_gold = readImage("stereobm/aloe-disp.png", cv::IMREAD_GRAYSCALE);
-
- ASSERT_FALSE(left_image.empty());
- ASSERT_FALSE(right_image.empty());
- ASSERT_FALSE(disp_gold.empty());
-
- cv::gpu::StereoBM_GPU bm(0, 128, 19);
- cv::gpu::GpuMat disp;
-
- bm(loadMat(left_image), loadMat(right_image), disp);
-
- EXPECT_MAT_NEAR(disp_gold, disp, 0.0);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_Calib3D, StereoBM, ALL_DEVICES);
-
-//////////////////////////////////////////////////////////////////////////
-// StereoBeliefPropagation
-
-struct StereoBeliefPropagation : testing::TestWithParam<cv::gpu::DeviceInfo>
-{
- cv::gpu::DeviceInfo devInfo;
-
- virtual void SetUp()
- {
- devInfo = GetParam();
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(StereoBeliefPropagation, Regression)
-{
- cv::Mat left_image = readImage("stereobp/aloe-L.png");
- cv::Mat right_image = readImage("stereobp/aloe-R.png");
- cv::Mat disp_gold = readImage("stereobp/aloe-disp.png", cv::IMREAD_GRAYSCALE);
-
- ASSERT_FALSE(left_image.empty());
- ASSERT_FALSE(right_image.empty());
- ASSERT_FALSE(disp_gold.empty());
-
- cv::gpu::StereoBeliefPropagation bp(64, 8, 2, 25, 0.1f, 15, 1, CV_16S);
- cv::gpu::GpuMat disp;
-
- bp(loadMat(left_image), loadMat(right_image), disp);
-
- cv::Mat h_disp(disp);
- h_disp.convertTo(h_disp, disp_gold.depth());
-
- EXPECT_MAT_NEAR(disp_gold, h_disp, 0.0);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_Calib3D, StereoBeliefPropagation, ALL_DEVICES);
-
-//////////////////////////////////////////////////////////////////////////
-// StereoConstantSpaceBP
-
-struct StereoConstantSpaceBP : testing::TestWithParam<cv::gpu::DeviceInfo>
-{
- cv::gpu::DeviceInfo devInfo;
-
- virtual void SetUp()
- {
- devInfo = GetParam();
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(StereoConstantSpaceBP, Regression)
-{
- cv::Mat left_image = readImage("csstereobp/aloe-L.png");
- cv::Mat right_image = readImage("csstereobp/aloe-R.png");
-
- cv::Mat disp_gold;
-
- if (supportFeature(devInfo, cv::gpu::FEATURE_SET_COMPUTE_20))
- disp_gold = readImage("csstereobp/aloe-disp.png", cv::IMREAD_GRAYSCALE);
- else
- disp_gold = readImage("csstereobp/aloe-disp_CC1X.png", cv::IMREAD_GRAYSCALE);
-
- ASSERT_FALSE(left_image.empty());
- ASSERT_FALSE(right_image.empty());
- ASSERT_FALSE(disp_gold.empty());
-
- cv::gpu::StereoConstantSpaceBP csbp(128, 16, 4, 4);
- cv::gpu::GpuMat disp;
-
- csbp(loadMat(left_image), loadMat(right_image), disp);
-
- cv::Mat h_disp(disp);
- h_disp.convertTo(h_disp, disp_gold.depth());
-
- EXPECT_MAT_NEAR(disp_gold, h_disp, 1.0);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_Calib3D, StereoConstantSpaceBP, ALL_DEVICES);
-
///////////////////////////////////////////////////////////////////////////////////////////////////////
// transformPoints
INSTANTIATE_TEST_CASE_P(GPU_Calib3D, SolvePnPRansac, ALL_DEVICES);
-////////////////////////////////////////////////////////////////////////////////
-// reprojectImageTo3D
-
-PARAM_TEST_CASE(ReprojectImageTo3D, cv::gpu::DeviceInfo, cv::Size, MatDepth, UseRoi)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- int depth;
- bool useRoi;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- depth = GET_PARAM(2);
- useRoi = GET_PARAM(3);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(ReprojectImageTo3D, Accuracy)
-{
- cv::Mat disp = randomMat(size, depth, 5.0, 30.0);
- cv::Mat Q = randomMat(cv::Size(4, 4), CV_32FC1, 0.1, 1.0);
-
- cv::gpu::GpuMat dst;
- cv::gpu::reprojectImageTo3D(loadMat(disp, useRoi), dst, Q, 3);
-
- cv::Mat dst_gold;
- cv::reprojectImageTo3D(disp, dst_gold, Q, false);
-
- EXPECT_MAT_NEAR(dst_gold, dst, 1e-5);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_Calib3D, ReprojectImageTo3D, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- testing::Values(MatDepth(CV_8U), MatDepth(CV_16S)),
- WHOLE_SUBMAT));
-
#endif // HAVE_CUDA
+++ /dev/null
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-// By downloading, copying, installing or using the software you agree to this license.
-// If you do not agree to this license, do not download, install,
-// copy or use the software.
-//
-//
-// License Agreement
-// For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-// * Redistribution's of source code must retain the above copyright notice,
-// this list of conditions and the following disclaimer.
-//
-// * Redistribution's in binary form must reproduce the above copyright notice,
-// this list of conditions and the following disclaimer in the documentation
-// and/or other materials provided with the distribution.
-//
-// * The name of the copyright holders may not be used to endorse or promote products
-// derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#include "test_precomp.hpp"
-
-#ifdef HAVE_CUDA
-
-using namespace cvtest;
-
-///////////////////////////////////////////////////////////////////////////////////////////////////////
-// Integral
-
-PARAM_TEST_CASE(Integral, cv::gpu::DeviceInfo, cv::Size, UseRoi)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- bool useRoi;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- useRoi = GET_PARAM(2);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(Integral, Accuracy)
-{
- cv::Mat src = randomMat(size, CV_8UC1);
-
- cv::gpu::GpuMat dst = createMat(cv::Size(src.cols + 1, src.rows + 1), CV_32SC1, useRoi);
- cv::gpu::integral(loadMat(src, useRoi), dst);
-
- cv::Mat dst_gold;
- cv::integral(src, dst_gold, CV_32S);
-
- EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, Integral, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- WHOLE_SUBMAT));
-
-///////////////////////////////////////////////////////////////////////////////////////////////////////
-// HistEven
-
-struct HistEven : testing::TestWithParam<cv::gpu::DeviceInfo>
-{
- cv::gpu::DeviceInfo devInfo;
-
- virtual void SetUp()
- {
- devInfo = GetParam();
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(HistEven, Accuracy)
-{
- cv::Mat img = readImage("stereobm/aloe-L.png");
- ASSERT_FALSE(img.empty());
-
- cv::Mat hsv;
- cv::cvtColor(img, hsv, cv::COLOR_BGR2HSV);
-
- int hbins = 30;
- float hranges[] = {0.0f, 180.0f};
-
- std::vector<cv::gpu::GpuMat> srcs;
- cv::gpu::split(loadMat(hsv), srcs);
-
- cv::gpu::GpuMat hist;
- cv::gpu::histEven(srcs[0], hist, hbins, (int)hranges[0], (int)hranges[1]);
-
- cv::MatND histnd;
- int histSize[] = {hbins};
- const float* ranges[] = {hranges};
- int channels[] = {0};
- cv::calcHist(&hsv, 1, channels, cv::Mat(), histnd, 1, histSize, ranges);
-
- cv::Mat hist_gold = histnd;
- hist_gold = hist_gold.t();
- hist_gold.convertTo(hist_gold, CV_32S);
-
- EXPECT_MAT_NEAR(hist_gold, hist, 0.0);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, HistEven, ALL_DEVICES);
-
-///////////////////////////////////////////////////////////////////////////////////////////////////////
-// CalcHist
-
-namespace
-{
- void calcHistGold(const cv::Mat& src, cv::Mat& hist)
- {
- hist.create(1, 256, CV_32SC1);
- hist.setTo(cv::Scalar::all(0));
-
- int* hist_row = hist.ptr<int>();
- for (int y = 0; y < src.rows; ++y)
- {
- const uchar* src_row = src.ptr(y);
-
- for (int x = 0; x < src.cols; ++x)
- ++hist_row[src_row[x]];
- }
- }
-}
-
-PARAM_TEST_CASE(CalcHist, cv::gpu::DeviceInfo, cv::Size)
-{
- cv::gpu::DeviceInfo devInfo;
-
- cv::Size size;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(CalcHist, Accuracy)
-{
- cv::Mat src = randomMat(size, CV_8UC1);
-
- cv::gpu::GpuMat hist;
- cv::gpu::calcHist(loadMat(src), hist);
-
- cv::Mat hist_gold;
- calcHistGold(src, hist_gold);
-
- EXPECT_MAT_NEAR(hist_gold, hist, 0.0);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, CalcHist, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES));
-
-///////////////////////////////////////////////////////////////////////////////////////////////////////
-// EqualizeHist
-
-PARAM_TEST_CASE(EqualizeHist, cv::gpu::DeviceInfo, cv::Size)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(EqualizeHist, Accuracy)
-{
- cv::Mat src = randomMat(size, CV_8UC1);
-
- cv::gpu::GpuMat dst;
- cv::gpu::equalizeHist(loadMat(src), dst);
-
- cv::Mat dst_gold;
- cv::equalizeHist(src, dst_gold);
-
- EXPECT_MAT_NEAR(dst_gold, dst, 3.0);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, EqualizeHist, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES));
-
-///////////////////////////////////////////////////////////////////////////////////////////////////////
-// CLAHE
-
-namespace
-{
- IMPLEMENT_PARAM_CLASS(ClipLimit, double)
-}
-
-PARAM_TEST_CASE(CLAHE, cv::gpu::DeviceInfo, cv::Size, ClipLimit)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- double clipLimit;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- clipLimit = GET_PARAM(2);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(CLAHE, Accuracy)
-{
- cv::Mat src = randomMat(size, CV_8UC1);
-
- cv::Ptr<cv::gpu::CLAHE> clahe = cv::gpu::createCLAHE(clipLimit);
- cv::gpu::GpuMat dst;
- clahe->apply(loadMat(src), dst);
-
- cv::Ptr<cv::CLAHE> clahe_gold = cv::createCLAHE(clipLimit);
- cv::Mat dst_gold;
- clahe_gold->apply(src, dst_gold);
-
- ASSERT_MAT_NEAR(dst_gold, dst, 1.0);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, CLAHE, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- testing::Values(0.0, 40.0)));
-
-////////////////////////////////////////////////////////////////////////
-// ColumnSum
-
-PARAM_TEST_CASE(ColumnSum, cv::gpu::DeviceInfo, cv::Size)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(ColumnSum, Accuracy)
-{
- cv::Mat src = randomMat(size, CV_32FC1);
-
- cv::gpu::GpuMat d_dst;
- cv::gpu::columnSum(loadMat(src), d_dst);
-
- cv::Mat dst(d_dst);
-
- for (int j = 0; j < src.cols; ++j)
- {
- float gold = src.at<float>(0, j);
- float res = dst.at<float>(0, j);
- ASSERT_NEAR(res, gold, 1e-5);
- }
-
- for (int i = 1; i < src.rows; ++i)
- {
- for (int j = 0; j < src.cols; ++j)
- {
- float gold = src.at<float>(i, j) += src.at<float>(i - 1, j);
- float res = dst.at<float>(i, j);
- ASSERT_NEAR(res, gold, 1e-5);
- }
- }
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, ColumnSum, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES));
-
-////////////////////////////////////////////////////////
-// Canny
-
-namespace
-{
- IMPLEMENT_PARAM_CLASS(AppertureSize, int);
- IMPLEMENT_PARAM_CLASS(L2gradient, bool);
-}
-
-PARAM_TEST_CASE(Canny, cv::gpu::DeviceInfo, AppertureSize, L2gradient, UseRoi)
-{
- cv::gpu::DeviceInfo devInfo;
- int apperture_size;
- bool useL2gradient;
- bool useRoi;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- apperture_size = GET_PARAM(1);
- useL2gradient = GET_PARAM(2);
- useRoi = GET_PARAM(3);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(Canny, Accuracy)
-{
- cv::Mat img = readImage("stereobm/aloe-L.png", cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(img.empty());
-
- double low_thresh = 50.0;
- double high_thresh = 100.0;
-
- if (!supportFeature(devInfo, cv::gpu::SHARED_ATOMICS))
- {
- try
- {
- cv::gpu::GpuMat edges;
- cv::gpu::Canny(loadMat(img), edges, low_thresh, high_thresh, apperture_size, useL2gradient);
- }
- catch (const cv::Exception& e)
- {
- ASSERT_EQ(cv::Error::StsNotImplemented, e.code);
- }
- }
- else
- {
- cv::gpu::GpuMat edges;
- cv::gpu::Canny(loadMat(img, useRoi), edges, low_thresh, high_thresh, apperture_size, useL2gradient);
-
- cv::Mat edges_gold;
- cv::Canny(img, edges_gold, low_thresh, high_thresh, apperture_size, useL2gradient);
-
- EXPECT_MAT_SIMILAR(edges_gold, edges, 2e-2);
- }
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, Canny, testing::Combine(
- ALL_DEVICES,
- testing::Values(AppertureSize(3), AppertureSize(5)),
- testing::Values(L2gradient(false), L2gradient(true)),
- WHOLE_SUBMAT));
-
-////////////////////////////////////////////////////////////////////////////////
-// MeanShift
-
-struct MeanShift : testing::TestWithParam<cv::gpu::DeviceInfo>
-{
- cv::gpu::DeviceInfo devInfo;
-
- cv::Mat img;
-
- int spatialRad;
- int colorRad;
-
- virtual void SetUp()
- {
- devInfo = GetParam();
-
- cv::gpu::setDevice(devInfo.deviceID());
-
- img = readImageType("meanshift/cones.png", CV_8UC4);
- ASSERT_FALSE(img.empty());
-
- spatialRad = 30;
- colorRad = 30;
- }
-};
-
-GPU_TEST_P(MeanShift, Filtering)
-{
- cv::Mat img_template;
- if (supportFeature(devInfo, cv::gpu::FEATURE_SET_COMPUTE_20))
- img_template = readImage("meanshift/con_result.png");
- else
- img_template = readImage("meanshift/con_result_CC1X.png");
- ASSERT_FALSE(img_template.empty());
-
- cv::gpu::GpuMat d_dst;
- cv::gpu::meanShiftFiltering(loadMat(img), d_dst, spatialRad, colorRad);
-
- ASSERT_EQ(CV_8UC4, d_dst.type());
-
- cv::Mat dst(d_dst);
-
- cv::Mat result;
- cv::cvtColor(dst, result, cv::COLOR_BGRA2BGR);
-
- EXPECT_MAT_NEAR(img_template, result, 0.0);
-}
-
-GPU_TEST_P(MeanShift, Proc)
-{
- cv::FileStorage fs;
- if (supportFeature(devInfo, cv::gpu::FEATURE_SET_COMPUTE_20))
- fs.open(std::string(cvtest::TS::ptr()->get_data_path()) + "meanshift/spmap.yaml", cv::FileStorage::READ);
- else
- fs.open(std::string(cvtest::TS::ptr()->get_data_path()) + "meanshift/spmap_CC1X.yaml", cv::FileStorage::READ);
- ASSERT_TRUE(fs.isOpened());
-
- cv::Mat spmap_template;
- fs["spmap"] >> spmap_template;
- ASSERT_FALSE(spmap_template.empty());
-
- cv::gpu::GpuMat rmap_filtered;
- cv::gpu::meanShiftFiltering(loadMat(img), rmap_filtered, spatialRad, colorRad);
-
- cv::gpu::GpuMat rmap;
- cv::gpu::GpuMat spmap;
- cv::gpu::meanShiftProc(loadMat(img), rmap, spmap, spatialRad, colorRad);
-
- ASSERT_EQ(CV_8UC4, rmap.type());
-
- EXPECT_MAT_NEAR(rmap_filtered, rmap, 0.0);
- EXPECT_MAT_NEAR(spmap_template, spmap, 0.0);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, MeanShift, ALL_DEVICES);
-
-////////////////////////////////////////////////////////////////////////////////
-// MeanShiftSegmentation
-
-namespace
-{
- IMPLEMENT_PARAM_CLASS(MinSize, int);
-}
-
-PARAM_TEST_CASE(MeanShiftSegmentation, cv::gpu::DeviceInfo, MinSize)
-{
- cv::gpu::DeviceInfo devInfo;
- int minsize;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- minsize = GET_PARAM(1);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(MeanShiftSegmentation, Regression)
-{
- cv::Mat img = readImageType("meanshift/cones.png", CV_8UC4);
- ASSERT_FALSE(img.empty());
-
- std::ostringstream path;
- path << "meanshift/cones_segmented_sp10_sr10_minsize" << minsize;
- if (supportFeature(devInfo, cv::gpu::FEATURE_SET_COMPUTE_20))
- path << ".png";
- else
- path << "_CC1X.png";
- cv::Mat dst_gold = readImage(path.str());
- ASSERT_FALSE(dst_gold.empty());
-
- cv::Mat dst;
- cv::gpu::meanShiftSegmentation(loadMat(img), dst, 10, 10, minsize);
-
- cv::Mat dst_rgb;
- cv::cvtColor(dst, dst_rgb, cv::COLOR_BGRA2BGR);
-
- EXPECT_MAT_SIMILAR(dst_gold, dst_rgb, 1e-3);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, MeanShiftSegmentation, testing::Combine(
- ALL_DEVICES,
- testing::Values(MinSize(0), MinSize(4), MinSize(20), MinSize(84), MinSize(340), MinSize(1364))));
-
-////////////////////////////////////////////////////////////////////////////
-// Blend
-
-namespace
-{
- template <typename T>
- void blendLinearGold(const cv::Mat& img1, const cv::Mat& img2, const cv::Mat& weights1, const cv::Mat& weights2, cv::Mat& result_gold)
- {
- result_gold.create(img1.size(), img1.type());
-
- int cn = img1.channels();
-
- for (int y = 0; y < img1.rows; ++y)
- {
- const float* weights1_row = weights1.ptr<float>(y);
- const float* weights2_row = weights2.ptr<float>(y);
- const T* img1_row = img1.ptr<T>(y);
- const T* img2_row = img2.ptr<T>(y);
- T* result_gold_row = result_gold.ptr<T>(y);
-
- for (int x = 0; x < img1.cols * cn; ++x)
- {
- float w1 = weights1_row[x / cn];
- float w2 = weights2_row[x / cn];
- result_gold_row[x] = static_cast<T>((img1_row[x] * w1 + img2_row[x] * w2) / (w1 + w2 + 1e-5f));
- }
- }
- }
-}
-
-PARAM_TEST_CASE(Blend, cv::gpu::DeviceInfo, cv::Size, MatType, UseRoi)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- int type;
- bool useRoi;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- type = GET_PARAM(2);
- useRoi = GET_PARAM(3);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(Blend, Accuracy)
-{
- int depth = CV_MAT_DEPTH(type);
-
- cv::Mat img1 = randomMat(size, type, 0.0, depth == CV_8U ? 255.0 : 1.0);
- cv::Mat img2 = randomMat(size, type, 0.0, depth == CV_8U ? 255.0 : 1.0);
- cv::Mat weights1 = randomMat(size, CV_32F, 0, 1);
- cv::Mat weights2 = randomMat(size, CV_32F, 0, 1);
-
- cv::gpu::GpuMat result;
- cv::gpu::blendLinear(loadMat(img1, useRoi), loadMat(img2, useRoi), loadMat(weights1, useRoi), loadMat(weights2, useRoi), result);
-
- cv::Mat result_gold;
- if (depth == CV_8U)
- blendLinearGold<uchar>(img1, img2, weights1, weights2, result_gold);
- else
- blendLinearGold<float>(img1, img2, weights1, weights2, result_gold);
-
- EXPECT_MAT_NEAR(result_gold, result, CV_MAT_DEPTH(type) == CV_8U ? 1.0 : 1e-5);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, Blend, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- testing::Values(MatType(CV_8UC1), MatType(CV_8UC3), MatType(CV_8UC4), MatType(CV_32FC1), MatType(CV_32FC3), MatType(CV_32FC4)),
- WHOLE_SUBMAT));
-
-////////////////////////////////////////////////////////
-// Convolve
-
-namespace
-{
- void convolveDFT(const cv::Mat& A, const cv::Mat& B, cv::Mat& C, bool ccorr = false)
- {
- // reallocate the output array if needed
- C.create(std::abs(A.rows - B.rows) + 1, std::abs(A.cols - B.cols) + 1, A.type());
- cv::Size dftSize;
-
- // compute the size of DFT transform
- dftSize.width = cv::getOptimalDFTSize(A.cols + B.cols - 1);
- dftSize.height = cv::getOptimalDFTSize(A.rows + B.rows - 1);
-
- // allocate temporary buffers and initialize them with 0s
- cv::Mat tempA(dftSize, A.type(), cv::Scalar::all(0));
- cv::Mat tempB(dftSize, B.type(), cv::Scalar::all(0));
-
- // copy A and B to the top-left corners of tempA and tempB, respectively
- cv::Mat roiA(tempA, cv::Rect(0, 0, A.cols, A.rows));
- A.copyTo(roiA);
- cv::Mat roiB(tempB, cv::Rect(0, 0, B.cols, B.rows));
- B.copyTo(roiB);
-
- // now transform the padded A & B in-place;
- // use "nonzeroRows" hint for faster processing
- cv::dft(tempA, tempA, 0, A.rows);
- cv::dft(tempB, tempB, 0, B.rows);
-
- // multiply the spectrums;
- // the function handles packed spectrum representations well
- cv::mulSpectrums(tempA, tempB, tempA, 0, ccorr);
-
- // transform the product back from the frequency domain.
- // Even though all the result rows will be non-zero,
- // you need only the first C.rows of them, and thus you
- // pass nonzeroRows == C.rows
- cv::dft(tempA, tempA, cv::DFT_INVERSE + cv::DFT_SCALE, C.rows);
-
- // now copy the result back to C.
- tempA(cv::Rect(0, 0, C.cols, C.rows)).copyTo(C);
- }
-
- IMPLEMENT_PARAM_CLASS(KSize, int);
- IMPLEMENT_PARAM_CLASS(Ccorr, bool);
-}
-
-PARAM_TEST_CASE(Convolve, cv::gpu::DeviceInfo, cv::Size, KSize, Ccorr)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- int ksize;
- bool ccorr;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- ksize = GET_PARAM(2);
- ccorr = GET_PARAM(3);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(Convolve, Accuracy)
-{
- cv::Mat src = randomMat(size, CV_32FC1, 0.0, 100.0);
- cv::Mat kernel = randomMat(cv::Size(ksize, ksize), CV_32FC1, 0.0, 1.0);
-
- cv::gpu::GpuMat dst;
- cv::gpu::convolve(loadMat(src), loadMat(kernel), dst, ccorr);
-
- cv::Mat dst_gold;
- convolveDFT(src, kernel, dst_gold, ccorr);
-
- EXPECT_MAT_NEAR(dst, dst_gold, 1e-1);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, Convolve, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- testing::Values(KSize(3), KSize(7), KSize(11), KSize(17), KSize(19), KSize(23), KSize(45)),
- testing::Values(Ccorr(false), Ccorr(true))));
-
-////////////////////////////////////////////////////////////////////////////////
-// MatchTemplate8U
-
-CV_ENUM(TemplateMethod, TM_SQDIFF, TM_SQDIFF_NORMED, TM_CCORR, TM_CCORR_NORMED, TM_CCOEFF, TM_CCOEFF_NORMED)
-
-namespace
-{
- IMPLEMENT_PARAM_CLASS(TemplateSize, cv::Size);
-}
-
-PARAM_TEST_CASE(MatchTemplate8U, cv::gpu::DeviceInfo, cv::Size, TemplateSize, Channels, TemplateMethod)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- cv::Size templ_size;
- int cn;
- int method;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- templ_size = GET_PARAM(2);
- cn = GET_PARAM(3);
- method = GET_PARAM(4);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(MatchTemplate8U, Accuracy)
-{
- cv::Mat image = randomMat(size, CV_MAKETYPE(CV_8U, cn));
- cv::Mat templ = randomMat(templ_size, CV_MAKETYPE(CV_8U, cn));
-
- cv::gpu::GpuMat dst;
- cv::gpu::matchTemplate(loadMat(image), loadMat(templ), dst, method);
-
- cv::Mat dst_gold;
- cv::matchTemplate(image, templ, dst_gold, method);
-
- EXPECT_MAT_NEAR(dst_gold, dst, templ_size.area() * 1e-1);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, MatchTemplate8U, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- testing::Values(TemplateSize(cv::Size(5, 5)), TemplateSize(cv::Size(16, 16)), TemplateSize(cv::Size(30, 30))),
- testing::Values(Channels(1), Channels(3), Channels(4)),
- TemplateMethod::all()));
-
-////////////////////////////////////////////////////////////////////////////////
-// MatchTemplate32F
-
-PARAM_TEST_CASE(MatchTemplate32F, cv::gpu::DeviceInfo, cv::Size, TemplateSize, Channels, TemplateMethod)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- cv::Size templ_size;
- int cn;
- int method;
-
- int n, m, h, w;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- templ_size = GET_PARAM(2);
- cn = GET_PARAM(3);
- method = GET_PARAM(4);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(MatchTemplate32F, Regression)
-{
- cv::Mat image = randomMat(size, CV_MAKETYPE(CV_32F, cn));
- cv::Mat templ = randomMat(templ_size, CV_MAKETYPE(CV_32F, cn));
-
- cv::gpu::GpuMat dst;
- cv::gpu::matchTemplate(loadMat(image), loadMat(templ), dst, method);
-
- cv::Mat dst_gold;
- cv::matchTemplate(image, templ, dst_gold, method);
-
- EXPECT_MAT_NEAR(dst_gold, dst, templ_size.area() * 1e-1);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, MatchTemplate32F, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- testing::Values(TemplateSize(cv::Size(5, 5)), TemplateSize(cv::Size(16, 16)), TemplateSize(cv::Size(30, 30))),
- testing::Values(Channels(1), Channels(3), Channels(4)),
- testing::Values(TemplateMethod(cv::TM_SQDIFF), TemplateMethod(cv::TM_CCORR))));
-
-////////////////////////////////////////////////////////////////////////////////
-// MatchTemplateBlackSource
-
-PARAM_TEST_CASE(MatchTemplateBlackSource, cv::gpu::DeviceInfo, TemplateMethod)
-{
- cv::gpu::DeviceInfo devInfo;
- int method;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- method = GET_PARAM(1);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(MatchTemplateBlackSource, Accuracy)
-{
- cv::Mat image = readImage("matchtemplate/black.png");
- ASSERT_FALSE(image.empty());
-
- cv::Mat pattern = readImage("matchtemplate/cat.png");
- ASSERT_FALSE(pattern.empty());
-
- cv::gpu::GpuMat d_dst;
- cv::gpu::matchTemplate(loadMat(image), loadMat(pattern), d_dst, method);
-
- cv::Mat dst(d_dst);
-
- double maxValue;
- cv::Point maxLoc;
- cv::minMaxLoc(dst, NULL, &maxValue, NULL, &maxLoc);
-
- cv::Point maxLocGold = cv::Point(284, 12);
-
- ASSERT_EQ(maxLocGold, maxLoc);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, MatchTemplateBlackSource, testing::Combine(
- ALL_DEVICES,
- testing::Values(TemplateMethod(cv::TM_CCOEFF_NORMED), TemplateMethod(cv::TM_CCORR_NORMED))));
-
-////////////////////////////////////////////////////////////////////////////////
-// MatchTemplate_CCOEF_NORMED
-
-PARAM_TEST_CASE(MatchTemplate_CCOEF_NORMED, cv::gpu::DeviceInfo, std::pair<std::string, std::string>)
-{
- cv::gpu::DeviceInfo devInfo;
- std::string imageName;
- std::string patternName;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- imageName = GET_PARAM(1).first;
- patternName = GET_PARAM(1).second;
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(MatchTemplate_CCOEF_NORMED, Accuracy)
-{
- cv::Mat image = readImage(imageName);
- ASSERT_FALSE(image.empty());
-
- cv::Mat pattern = readImage(patternName);
- ASSERT_FALSE(pattern.empty());
-
- cv::gpu::GpuMat d_dst;
- cv::gpu::matchTemplate(loadMat(image), loadMat(pattern), d_dst, cv::TM_CCOEFF_NORMED);
-
- cv::Mat dst(d_dst);
-
- cv::Point minLoc, maxLoc;
- double minVal, maxVal;
- cv::minMaxLoc(dst, &minVal, &maxVal, &minLoc, &maxLoc);
-
- cv::Mat dstGold;
- cv::matchTemplate(image, pattern, dstGold, cv::TM_CCOEFF_NORMED);
-
- double minValGold, maxValGold;
- cv::Point minLocGold, maxLocGold;
- cv::minMaxLoc(dstGold, &minValGold, &maxValGold, &minLocGold, &maxLocGold);
-
- ASSERT_EQ(minLocGold, minLoc);
- ASSERT_EQ(maxLocGold, maxLoc);
- ASSERT_LE(maxVal, 1.0);
- ASSERT_GE(minVal, -1.0);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, MatchTemplate_CCOEF_NORMED, testing::Combine(
- ALL_DEVICES,
- testing::Values(std::make_pair(std::string("matchtemplate/source-0.png"), std::string("matchtemplate/target-0.png")))));
-
-////////////////////////////////////////////////////////////////////////////////
-// MatchTemplate_CanFindBigTemplate
-
-struct MatchTemplate_CanFindBigTemplate : testing::TestWithParam<cv::gpu::DeviceInfo>
-{
- cv::gpu::DeviceInfo devInfo;
-
- virtual void SetUp()
- {
- devInfo = GetParam();
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(MatchTemplate_CanFindBigTemplate, SQDIFF_NORMED)
-{
- cv::Mat scene = readImage("matchtemplate/scene.png");
- ASSERT_FALSE(scene.empty());
-
- cv::Mat templ = readImage("matchtemplate/template.png");
- ASSERT_FALSE(templ.empty());
-
- cv::gpu::GpuMat d_result;
- cv::gpu::matchTemplate(loadMat(scene), loadMat(templ), d_result, cv::TM_SQDIFF_NORMED);
-
- cv::Mat result(d_result);
-
- double minVal;
- cv::Point minLoc;
- cv::minMaxLoc(result, &minVal, 0, &minLoc, 0);
-
- ASSERT_GE(minVal, 0);
- ASSERT_LT(minVal, 1e-3);
- ASSERT_EQ(344, minLoc.x);
- ASSERT_EQ(0, minLoc.y);
-}
-
-GPU_TEST_P(MatchTemplate_CanFindBigTemplate, SQDIFF)
-{
- cv::Mat scene = readImage("matchtemplate/scene.png");
- ASSERT_FALSE(scene.empty());
-
- cv::Mat templ = readImage("matchtemplate/template.png");
- ASSERT_FALSE(templ.empty());
-
- cv::gpu::GpuMat d_result;
- cv::gpu::matchTemplate(loadMat(scene), loadMat(templ), d_result, cv::TM_SQDIFF);
-
- cv::Mat result(d_result);
-
- double minVal;
- cv::Point minLoc;
- cv::minMaxLoc(result, &minVal, 0, &minLoc, 0);
-
- ASSERT_GE(minVal, 0);
- ASSERT_EQ(344, minLoc.x);
- ASSERT_EQ(0, minLoc.y);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, MatchTemplate_CanFindBigTemplate, ALL_DEVICES);
-
-////////////////////////////////////////////////////////////////////////////
-// MulSpectrums
-
-CV_FLAGS(DftFlags, 0, DFT_INVERSE, DFT_SCALE, DFT_ROWS, DFT_COMPLEX_OUTPUT, DFT_REAL_OUTPUT)
-
-PARAM_TEST_CASE(MulSpectrums, cv::gpu::DeviceInfo, cv::Size, DftFlags)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- int flag;
-
- cv::Mat a, b;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- flag = GET_PARAM(2);
-
- cv::gpu::setDevice(devInfo.deviceID());
-
- a = randomMat(size, CV_32FC2);
- b = randomMat(size, CV_32FC2);
- }
-};
-
-GPU_TEST_P(MulSpectrums, Simple)
-{
- cv::gpu::GpuMat c;
- cv::gpu::mulSpectrums(loadMat(a), loadMat(b), c, flag, false);
-
- cv::Mat c_gold;
- cv::mulSpectrums(a, b, c_gold, flag, false);
-
- EXPECT_MAT_NEAR(c_gold, c, 1e-2);
-}
-
-GPU_TEST_P(MulSpectrums, Scaled)
-{
- float scale = 1.f / size.area();
-
- cv::gpu::GpuMat c;
- cv::gpu::mulAndScaleSpectrums(loadMat(a), loadMat(b), c, flag, scale, false);
-
- cv::Mat c_gold;
- cv::mulSpectrums(a, b, c_gold, flag, false);
- c_gold.convertTo(c_gold, c_gold.type(), scale);
-
- EXPECT_MAT_NEAR(c_gold, c, 1e-2);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, MulSpectrums, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- testing::Values(DftFlags(0), DftFlags(cv::DFT_ROWS))));
-
-////////////////////////////////////////////////////////////////////////////
-// Dft
-
-struct Dft : testing::TestWithParam<cv::gpu::DeviceInfo>
-{
- cv::gpu::DeviceInfo devInfo;
-
- virtual void SetUp()
- {
- devInfo = GetParam();
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-namespace
-{
- void testC2C(const std::string& hint, int cols, int rows, int flags, bool inplace)
- {
- SCOPED_TRACE(hint);
-
- cv::Mat a = randomMat(cv::Size(cols, rows), CV_32FC2, 0.0, 10.0);
-
- cv::Mat b_gold;
- cv::dft(a, b_gold, flags);
-
- cv::gpu::GpuMat d_b;
- cv::gpu::GpuMat d_b_data;
- if (inplace)
- {
- d_b_data.create(1, a.size().area(), CV_32FC2);
- d_b = cv::gpu::GpuMat(a.rows, a.cols, CV_32FC2, d_b_data.ptr(), a.cols * d_b_data.elemSize());
- }
- cv::gpu::dft(loadMat(a), d_b, cv::Size(cols, rows), flags);
-
- EXPECT_TRUE(!inplace || d_b.ptr() == d_b_data.ptr());
- ASSERT_EQ(CV_32F, d_b.depth());
- ASSERT_EQ(2, d_b.channels());
- EXPECT_MAT_NEAR(b_gold, cv::Mat(d_b), rows * cols * 1e-4);
- }
-}
-
-GPU_TEST_P(Dft, C2C)
-{
- int cols = randomInt(2, 100);
- int rows = randomInt(2, 100);
-
- for (int i = 0; i < 2; ++i)
- {
- bool inplace = i != 0;
-
- testC2C("no flags", cols, rows, 0, inplace);
- testC2C("no flags 0 1", cols, rows + 1, 0, inplace);
- testC2C("no flags 1 0", cols, rows + 1, 0, inplace);
- testC2C("no flags 1 1", cols + 1, rows, 0, inplace);
- testC2C("DFT_INVERSE", cols, rows, cv::DFT_INVERSE, inplace);
- testC2C("DFT_ROWS", cols, rows, cv::DFT_ROWS, inplace);
- testC2C("single col", 1, rows, 0, inplace);
- testC2C("single row", cols, 1, 0, inplace);
- testC2C("single col inversed", 1, rows, cv::DFT_INVERSE, inplace);
- testC2C("single row inversed", cols, 1, cv::DFT_INVERSE, inplace);
- testC2C("single row DFT_ROWS", cols, 1, cv::DFT_ROWS, inplace);
- testC2C("size 1 2", 1, 2, 0, inplace);
- testC2C("size 2 1", 2, 1, 0, inplace);
- }
-}
-
-namespace
-{
- void testR2CThenC2R(const std::string& hint, int cols, int rows, bool inplace)
- {
- SCOPED_TRACE(hint);
-
- cv::Mat a = randomMat(cv::Size(cols, rows), CV_32FC1, 0.0, 10.0);
-
- cv::gpu::GpuMat d_b, d_c;
- cv::gpu::GpuMat d_b_data, d_c_data;
- if (inplace)
- {
- if (a.cols == 1)
- {
- d_b_data.create(1, (a.rows / 2 + 1) * a.cols, CV_32FC2);
- d_b = cv::gpu::GpuMat(a.rows / 2 + 1, a.cols, CV_32FC2, d_b_data.ptr(), a.cols * d_b_data.elemSize());
- }
- else
- {
- d_b_data.create(1, a.rows * (a.cols / 2 + 1), CV_32FC2);
- d_b = cv::gpu::GpuMat(a.rows, a.cols / 2 + 1, CV_32FC2, d_b_data.ptr(), (a.cols / 2 + 1) * d_b_data.elemSize());
- }
- d_c_data.create(1, a.size().area(), CV_32F);
- d_c = cv::gpu::GpuMat(a.rows, a.cols, CV_32F, d_c_data.ptr(), a.cols * d_c_data.elemSize());
- }
-
- cv::gpu::dft(loadMat(a), d_b, cv::Size(cols, rows), 0);
- cv::gpu::dft(d_b, d_c, cv::Size(cols, rows), cv::DFT_REAL_OUTPUT | cv::DFT_SCALE);
-
- EXPECT_TRUE(!inplace || d_b.ptr() == d_b_data.ptr());
- EXPECT_TRUE(!inplace || d_c.ptr() == d_c_data.ptr());
- ASSERT_EQ(CV_32F, d_c.depth());
- ASSERT_EQ(1, d_c.channels());
-
- cv::Mat c(d_c);
- EXPECT_MAT_NEAR(a, c, rows * cols * 1e-5);
- }
-}
-
-GPU_TEST_P(Dft, R2CThenC2R)
-{
- int cols = randomInt(2, 100);
- int rows = randomInt(2, 100);
-
- testR2CThenC2R("sanity", cols, rows, false);
- testR2CThenC2R("sanity 0 1", cols, rows + 1, false);
- testR2CThenC2R("sanity 1 0", cols + 1, rows, false);
- testR2CThenC2R("sanity 1 1", cols + 1, rows + 1, false);
- testR2CThenC2R("single col", 1, rows, false);
- testR2CThenC2R("single col 1", 1, rows + 1, false);
- testR2CThenC2R("single row", cols, 1, false);
- testR2CThenC2R("single row 1", cols + 1, 1, false);
-
- testR2CThenC2R("sanity", cols, rows, true);
- testR2CThenC2R("sanity 0 1", cols, rows + 1, true);
- testR2CThenC2R("sanity 1 0", cols + 1, rows, true);
- testR2CThenC2R("sanity 1 1", cols + 1, rows + 1, true);
- testR2CThenC2R("single row", cols, 1, true);
- testR2CThenC2R("single row 1", cols + 1, 1, true);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, Dft, ALL_DEVICES);
-
-///////////////////////////////////////////////////////////////////////////////////////////////////////
-// CornerHarris
-
-namespace
-{
- IMPLEMENT_PARAM_CLASS(BlockSize, int);
- IMPLEMENT_PARAM_CLASS(ApertureSize, int);
-}
-
-PARAM_TEST_CASE(CornerHarris, cv::gpu::DeviceInfo, MatType, BorderType, BlockSize, ApertureSize)
-{
- cv::gpu::DeviceInfo devInfo;
- int type;
- int borderType;
- int blockSize;
- int apertureSize;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- type = GET_PARAM(1);
- borderType = GET_PARAM(2);
- blockSize = GET_PARAM(3);
- apertureSize = GET_PARAM(4);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(CornerHarris, Accuracy)
-{
- cv::Mat src = readImageType("stereobm/aloe-L.png", type);
- ASSERT_FALSE(src.empty());
-
- double k = randomDouble(0.1, 0.9);
-
- cv::gpu::GpuMat dst;
- cv::gpu::cornerHarris(loadMat(src), dst, blockSize, apertureSize, k, borderType);
-
- cv::Mat dst_gold;
- cv::cornerHarris(src, dst_gold, blockSize, apertureSize, k, borderType);
-
- EXPECT_MAT_NEAR(dst_gold, dst, 0.02);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, CornerHarris, testing::Combine(
- ALL_DEVICES,
- testing::Values(MatType(CV_8UC1), MatType(CV_32FC1)),
- testing::Values(BorderType(cv::BORDER_REFLECT101), BorderType(cv::BORDER_REPLICATE), BorderType(cv::BORDER_REFLECT)),
- testing::Values(BlockSize(3), BlockSize(5), BlockSize(7)),
- testing::Values(ApertureSize(0), ApertureSize(3), ApertureSize(5), ApertureSize(7))));
-
-///////////////////////////////////////////////////////////////////////////////////////////////////////
-// cornerMinEigen
-
-PARAM_TEST_CASE(CornerMinEigen, cv::gpu::DeviceInfo, MatType, BorderType, BlockSize, ApertureSize)
-{
- cv::gpu::DeviceInfo devInfo;
- int type;
- int borderType;
- int blockSize;
- int apertureSize;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- type = GET_PARAM(1);
- borderType = GET_PARAM(2);
- blockSize = GET_PARAM(3);
- apertureSize = GET_PARAM(4);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(CornerMinEigen, Accuracy)
-{
- cv::Mat src = readImageType("stereobm/aloe-L.png", type);
- ASSERT_FALSE(src.empty());
-
- cv::gpu::GpuMat dst;
- cv::gpu::cornerMinEigenVal(loadMat(src), dst, blockSize, apertureSize, borderType);
-
- cv::Mat dst_gold;
- cv::cornerMinEigenVal(src, dst_gold, blockSize, apertureSize, borderType);
-
- EXPECT_MAT_NEAR(dst_gold, dst, 0.02);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, CornerMinEigen, testing::Combine(
- ALL_DEVICES,
- testing::Values(MatType(CV_8UC1), MatType(CV_32FC1)),
- testing::Values(BorderType(cv::BORDER_REFLECT101), BorderType(cv::BORDER_REPLICATE), BorderType(cv::BORDER_REFLECT)),
- testing::Values(BlockSize(3), BlockSize(5), BlockSize(7)),
- testing::Values(ApertureSize(0), ApertureSize(3), ApertureSize(5), ApertureSize(7))));
-
-#endif // HAVE_CUDA
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+CV_GPU_TEST_MAIN("gpu")
#ifndef __OPENCV_TEST_PRECOMP_HPP__
#define __OPENCV_TEST_PRECOMP_HPP__
-#include <cmath>
-#include <ctime>
-#include <cstdio>
-#include <iostream>
#include <fstream>
-#include <functional>
-#include <sstream>
-#include <string>
-#include <limits>
-#include <algorithm>
-#include <iterator>
-#include <stdexcept>
-#include "opencv2/core.hpp"
-#include "opencv2/core/opengl.hpp"
-#include "opencv2/highgui.hpp"
-#include "opencv2/calib3d.hpp"
-#include "opencv2/imgproc.hpp"
-#include "opencv2/video.hpp"
#include "opencv2/ts.hpp"
#include "opencv2/ts/gpu_test.hpp"
-#include "opencv2/gpu.hpp"
-
-#include "interpolation.hpp"
-#include "main_test_nvidia.h"
-#include "opencv2/core/gpu_private.hpp"
+#include "opencv2/gpu.hpp"
+#include "opencv2/core.hpp"
+#include "opencv2/core/opengl.hpp"
+#include "opencv2/calib3d.hpp"
+#include "opencv2/objdetect.hpp"
#endif
--- /dev/null
+if(ANDROID OR IOS)
+ ocv_module_disable(gpuarithm)
+endif()
+
+set(the_description "GPU-accelerated Operations on Matrices")
+
+ocv_warnings_disable(CMAKE_CXX_FLAGS /wd4127 /wd4324 /wd4512 -Wundef -Wmissing-declarations)
+
+ocv_add_module(gpuarithm opencv_core OPTIONAL opencv_gpulegacy)
+
+ocv_module_include_directories()
+ocv_glob_module_sources()
+
+set(extra_libs "")
+
+if(HAVE_CUBLAS)
+ list(APPEND extra_libs ${CUDA_cublas_LIBRARY})
+endif()
+
+if(HAVE_CUFFT)
+ list(APPEND extra_libs ${CUDA_cufft_LIBRARY})
+endif()
+
+ocv_create_module(${extra_libs})
+
+ocv_add_precompiled_headers(${the_module})
+
+ocv_add_accuracy_tests(DEPENDS_ON opencv_imgproc)
+ocv_add_perf_tests(DEPENDS_ON opencv_imgproc)
--- /dev/null
+Arithm Operations on Matrices
+=============================
+
+.. highlight:: cpp
+
+
+
+gpu::gemm
+------------------
+Performs generalized matrix multiplication.
+
+.. ocv:function:: void gpu::gemm(const GpuMat& src1, const GpuMat& src2, double alpha, const GpuMat& src3, double beta, GpuMat& dst, int flags = 0, Stream& stream = Stream::Null())
+
+ :param src1: First multiplied input matrix that should have ``CV_32FC1`` , ``CV_64FC1`` , ``CV_32FC2`` , or ``CV_64FC2`` type.
+
+ :param src2: Second multiplied input matrix of the same type as ``src1`` .
+
+ :param alpha: Weight of the matrix product.
+
+ :param src3: Third optional delta matrix added to the matrix product. It should have the same type as ``src1`` and ``src2`` .
+
+ :param beta: Weight of ``src3`` .
+
+ :param dst: Destination matrix. It has the proper size and the same type as input matrices.
+
+ :param flags: Operation flags:
+
+ * **GEMM_1_T** transpose ``src1``
+ * **GEMM_2_T** transpose ``src2``
+ * **GEMM_3_T** transpose ``src3``
+
+ :param stream: Stream for the asynchronous version.
+
+The function performs generalized matrix multiplication similar to the ``gemm`` functions in BLAS level 3. For example, ``gemm(src1, src2, alpha, src3, beta, dst, GEMM_1_T + GEMM_3_T)`` corresponds to
+
+.. math::
+
+ \texttt{dst} = \texttt{alpha} \cdot \texttt{src1} ^T \cdot \texttt{src2} + \texttt{beta} \cdot \texttt{src3} ^T
+
+.. note:: Transposition operation doesn't support ``CV_64FC2`` input type.
+
+.. seealso:: :ocv:func:`gemm`
+
+
+
+gpu::mulSpectrums
+---------------------
+Performs a per-element multiplication of two Fourier spectrums.
+
+.. ocv:function:: void gpu::mulSpectrums( const GpuMat& a, const GpuMat& b, GpuMat& c, int flags, bool conjB=false, Stream& stream=Stream::Null() )
+
+ :param a: First spectrum.
+
+ :param b: Second spectrum with the same size and type as ``a`` .
+
+ :param c: Destination spectrum.
+
+ :param flags: Mock parameter used for CPU/GPU interfaces similarity.
+
+ :param conjB: Optional flag to specify if the second spectrum needs to be conjugated before the multiplication.
+
+ Only full (not packed) ``CV_32FC2`` complex spectrums in the interleaved format are supported for now.
+
+.. seealso:: :ocv:func:`mulSpectrums`
+
+
+
+gpu::mulAndScaleSpectrums
+-----------------------------
+Performs a per-element multiplication of two Fourier spectrums and scales the result.
+
+.. ocv:function:: void gpu::mulAndScaleSpectrums( const GpuMat& a, const GpuMat& b, GpuMat& c, int flags, float scale, bool conjB=false, Stream& stream=Stream::Null() )
+
+ :param a: First spectrum.
+
+ :param b: Second spectrum with the same size and type as ``a`` .
+
+ :param c: Destination spectrum.
+
+ :param flags: Mock parameter used for CPU/GPU interfaces similarity.
+
+ :param scale: Scale constant.
+
+ :param conjB: Optional flag to specify if the second spectrum needs to be conjugated before the multiplication.
+
+ Only full (not packed) ``CV_32FC2`` complex spectrums in the interleaved format are supported for now.
+
+.. seealso:: :ocv:func:`mulSpectrums`
+
+
+
+gpu::dft
+------------
+Performs a forward or inverse discrete Fourier transform (1D or 2D) of the floating point matrix.
+
+.. ocv:function:: void gpu::dft( const GpuMat& src, GpuMat& dst, Size dft_size, int flags=0, Stream& stream=Stream::Null() )
+
+ :param src: Source matrix (real or complex).
+
+ :param dst: Destination matrix (real or complex).
+
+ :param dft_size: Size of a discrete Fourier transform.
+
+ :param flags: Optional flags:
+
+ * **DFT_ROWS** transforms each individual row of the source matrix.
+
+ * **DFT_SCALE** scales the result: divide it by the number of elements in the transform (obtained from ``dft_size`` ).
+
+ * **DFT_INVERSE** inverts DFT. Use for complex-complex cases (real-complex and complex-real cases are always forward and inverse, respectively).
+
+ * **DFT_REAL_OUTPUT** specifies the output as real. The source matrix is the result of real-complex transform, so the destination matrix must be real.
+
+Use to handle real matrices ( ``CV32FC1`` ) and complex matrices in the interleaved format ( ``CV32FC2`` ).
+
+The source matrix should be continuous, otherwise reallocation and data copying is performed. The function chooses an operation mode depending on the flags, size, and channel count of the source matrix:
+
+ * If the source matrix is complex and the output is not specified as real, the destination matrix is complex and has the ``dft_size`` size and ``CV_32FC2`` type. The destination matrix contains a full result of the DFT (forward or inverse).
+
+ * If the source matrix is complex and the output is specified as real, the function assumes that its input is the result of the forward transform (see the next item). The destination matrix has the ``dft_size`` size and ``CV_32FC1`` type. It contains the result of the inverse DFT.
+
+ * If the source matrix is real (its type is ``CV_32FC1`` ), forward DFT is performed. The result of the DFT is packed into complex ( ``CV_32FC2`` ) matrix. So, the width of the destination matrix is ``dft_size.width / 2 + 1`` . But if the source is a single column, the height is reduced instead of the width.
+
+.. seealso:: :ocv:func:`dft`
+
+
+
+gpu::ConvolveBuf
+----------------
+.. ocv:struct:: gpu::ConvolveBuf
+
+Class providing a memory buffer for :ocv:func:`gpu::convolve` function, plus it allows to adjust some specific parameters. ::
+
+ struct CV_EXPORTS ConvolveBuf
+ {
+ Size result_size;
+ Size block_size;
+ Size user_block_size;
+ Size dft_size;
+ int spect_len;
+
+ GpuMat image_spect, templ_spect, result_spect;
+ GpuMat image_block, templ_block, result_data;
+
+ void create(Size image_size, Size templ_size);
+ static Size estimateBlockSize(Size result_size, Size templ_size);
+ };
+
+You can use field `user_block_size` to set specific block size for :ocv:func:`gpu::convolve` function. If you leave its default value `Size(0,0)` then automatic estimation of block size will be used (which is optimized for speed). By varying `user_block_size` you can reduce memory requirements at the cost of speed.
+
+
+
+gpu::ConvolveBuf::create
+------------------------
+.. ocv:function:: gpu::ConvolveBuf::create(Size image_size, Size templ_size)
+
+Constructs a buffer for :ocv:func:`gpu::convolve` function with respective arguments.
+
+
+
+gpu::convolve
+-----------------
+Computes a convolution (or cross-correlation) of two images.
+
+.. ocv:function:: void gpu::convolve(const GpuMat& image, const GpuMat& templ, GpuMat& result, bool ccorr=false)
+
+.. ocv:function:: void gpu::convolve( const GpuMat& image, const GpuMat& templ, GpuMat& result, bool ccorr, ConvolveBuf& buf, Stream& stream=Stream::Null() )
+
+ :param image: Source image. Only ``CV_32FC1`` images are supported for now.
+
+ :param templ: Template image. The size is not greater than the ``image`` size. The type is the same as ``image`` .
+
+ :param result: Result image. If ``image`` is *W x H* and ``templ`` is *w x h*, then ``result`` must be *W-w+1 x H-h+1*.
+
+ :param ccorr: Flags to evaluate cross-correlation instead of convolution.
+
+ :param buf: Optional buffer to avoid extra memory allocations and to adjust some specific parameters. See :ocv:struct:`gpu::ConvolveBuf`.
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`gpu::filter2D`
+
+
+
+gpu::integral
+-----------------
+Computes an integral image.
+
+.. ocv:function:: void gpu::integral(const GpuMat& src, GpuMat& sum, Stream& stream = Stream::Null())
+
+ :param src: Source image. Only ``CV_8UC1`` images are supported for now.
+
+ :param sum: Integral image containing 32-bit unsigned integer values packed into ``CV_32SC1`` .
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`integral`
+
+
+
+gpu::sqrIntegral
+--------------------
+Computes a squared integral image.
+
+.. ocv:function:: void gpu::sqrIntegral(const GpuMat& src, GpuMat& sqsum, Stream& stream = Stream::Null())
+
+ :param src: Source image. Only ``CV_8UC1`` images are supported for now.
+
+ :param sqsum: Squared integral image containing 64-bit unsigned integer values packed into ``CV_64FC1`` .
+
+ :param stream: Stream for the asynchronous version.
--- /dev/null
+Core Operations on Matrices
+===========================
+
+.. highlight:: cpp
+
+
+
+gpu::merge
+--------------
+Makes a multi-channel matrix out of several single-channel matrices.
+
+.. ocv:function:: void gpu::merge(const GpuMat* src, size_t n, GpuMat& dst, Stream& stream = Stream::Null())
+
+.. ocv:function:: void gpu::merge(const vector<GpuMat>& src, GpuMat& dst, Stream& stream = Stream::Null())
+
+ :param src: Array/vector of source matrices.
+
+ :param n: Number of source matrices.
+
+ :param dst: Destination matrix.
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`merge`
+
+
+
+gpu::split
+--------------
+Copies each plane of a multi-channel matrix into an array.
+
+.. ocv:function:: void gpu::split(const GpuMat& src, GpuMat* dst, Stream& stream = Stream::Null())
+
+.. ocv:function:: void gpu::split(const GpuMat& src, vector<GpuMat>& dst, Stream& stream = Stream::Null())
+
+ :param src: Source matrix.
+
+ :param dst: Destination array/vector of single-channel matrices.
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`split`
+
+
+
+gpu::copyMakeBorder
+-----------------------
+Forms a border around an image.
+
+.. ocv:function:: void gpu::copyMakeBorder(const GpuMat& src, GpuMat& dst, int top, int bottom, int left, int right, int borderType, const Scalar& value = Scalar(), Stream& stream = Stream::Null())
+
+ :param src: Source image. ``CV_8UC1`` , ``CV_8UC4`` , ``CV_32SC1`` , and ``CV_32FC1`` types are supported.
+
+ :param dst: Destination image with the same type as ``src``. The size is ``Size(src.cols+left+right, src.rows+top+bottom)`` .
+
+ :param top:
+
+ :param bottom:
+
+ :param left:
+
+ :param right: Number of pixels in each direction from the source image rectangle to extrapolate. For example: ``top=1, bottom=1, left=1, right=1`` mean that 1 pixel-wide border needs to be built.
+
+ :param borderType: Border type. See :ocv:func:`borderInterpolate` for details. ``BORDER_REFLECT101`` , ``BORDER_REPLICATE`` , ``BORDER_CONSTANT`` , ``BORDER_REFLECT`` and ``BORDER_WRAP`` are supported for now.
+
+ :param value: Border value.
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`copyMakeBorder`
+
+
+
+gpu::transpose
+------------------
+Transposes a matrix.
+
+.. ocv:function:: void gpu::transpose( const GpuMat& src1, GpuMat& dst, Stream& stream=Stream::Null() )
+
+ :param src1: Source matrix. 1-, 4-, 8-byte element sizes are supported for now (CV_8UC1, CV_8UC4, CV_16UC2, CV_32FC1, etc).
+
+ :param dst: Destination matrix.
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`transpose`
+
+
+
+gpu::flip
+-------------
+Flips a 2D matrix around vertical, horizontal, or both axes.
+
+.. ocv:function:: void gpu::flip( const GpuMat& a, GpuMat& b, int flipCode, Stream& stream=Stream::Null() )
+
+ :param a: Source matrix. Supports 1, 3 and 4 channels images with ``CV_8U``, ``CV_16U``, ``CV_32S`` or ``CV_32F`` depth.
+
+ :param b: Destination matrix.
+
+ :param flipCode: Flip mode for the source:
+
+ * ``0`` Flips around x-axis.
+
+ * ``>0`` Flips around y-axis.
+
+ * ``<0`` Flips around both axes.
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`flip`
+
+
+
+gpu::LUT
+------------
+Transforms the source matrix into the destination matrix using the given look-up table: ``dst(I) = lut(src(I))``
+
+.. ocv:function:: void gpu::LUT(const GpuMat& src, const Mat& lut, GpuMat& dst, Stream& stream = Stream::Null())
+
+ :param src: Source matrix. ``CV_8UC1`` and ``CV_8UC3`` matrices are supported for now.
+
+ :param lut: Look-up table of 256 elements. It is a continuous ``CV_8U`` matrix.
+
+ :param dst: Destination matrix with the same depth as ``lut`` and the same number of channels as ``src`` .
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`LUT`
Per-element Operations
-=======================
+======================
.. highlight:: cpp
.. seealso:: :ocv:func:`divide`
+
gpu::addWeighted
----------------
Computes the weighted sum of two arrays.
:param stream: Stream for the asynchronous version.
.. seealso:: :ocv:func:`max`
+
+
+
+gpu::threshold
+------------------
+Applies a fixed-level threshold to each array element.
+
+.. ocv:function:: double gpu::threshold(const GpuMat& src, GpuMat& dst, double thresh, double maxval, int type, Stream& stream = Stream::Null())
+
+ :param src: Source array (single-channel).
+
+ :param dst: Destination array with the same size and type as ``src`` .
+
+ :param thresh: Threshold value.
+
+ :param maxval: Maximum value to use with ``THRESH_BINARY`` and ``THRESH_BINARY_INV`` threshold types.
+
+ :param type: Threshold type. For details, see :ocv:func:`threshold` . The ``THRESH_OTSU`` threshold type is not supported.
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`threshold`
+
+
+
+gpu::magnitude
+------------------
+Computes magnitudes of complex matrix elements.
+
+.. ocv:function:: void gpu::magnitude( const GpuMat& xy, GpuMat& magnitude, Stream& stream=Stream::Null() )
+
+.. ocv:function:: void gpu::magnitude(const GpuMat& x, const GpuMat& y, GpuMat& magnitude, Stream& stream = Stream::Null())
+
+ :param xy: Source complex matrix in the interleaved format ( ``CV_32FC2`` ).
+
+ :param x: Source matrix containing real components ( ``CV_32FC1`` ).
+
+ :param y: Source matrix containing imaginary components ( ``CV_32FC1`` ).
+
+ :param magnitude: Destination matrix of float magnitudes ( ``CV_32FC1`` ).
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`magnitude`
+
+
+
+gpu::magnitudeSqr
+---------------------
+Computes squared magnitudes of complex matrix elements.
+
+.. ocv:function:: void gpu::magnitudeSqr( const GpuMat& xy, GpuMat& magnitude, Stream& stream=Stream::Null() )
+
+.. ocv:function:: void gpu::magnitudeSqr(const GpuMat& x, const GpuMat& y, GpuMat& magnitude, Stream& stream = Stream::Null())
+
+ :param xy: Source complex matrix in the interleaved format ( ``CV_32FC2`` ).
+
+ :param x: Source matrix containing real components ( ``CV_32FC1`` ).
+
+ :param y: Source matrix containing imaginary components ( ``CV_32FC1`` ).
+
+ :param magnitude: Destination matrix of float magnitude squares ( ``CV_32FC1`` ).
+
+ :param stream: Stream for the asynchronous version.
+
+
+
+gpu::phase
+--------------
+Computes polar angles of complex matrix elements.
+
+.. ocv:function:: void gpu::phase(const GpuMat& x, const GpuMat& y, GpuMat& angle, bool angleInDegrees=false, Stream& stream = Stream::Null())
+
+ :param x: Source matrix containing real components ( ``CV_32FC1`` ).
+
+ :param y: Source matrix containing imaginary components ( ``CV_32FC1`` ).
+
+ :param angle: Destination matrix of angles ( ``CV_32FC1`` ).
+
+ :param angleInDegrees: Flag for angles that must be evaluated in degrees.
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`phase`
+
+
+
+gpu::cartToPolar
+--------------------
+Converts Cartesian coordinates into polar.
+
+.. ocv:function:: void gpu::cartToPolar(const GpuMat& x, const GpuMat& y, GpuMat& magnitude, GpuMat& angle, bool angleInDegrees=false, Stream& stream = Stream::Null())
+
+ :param x: Source matrix containing real components ( ``CV_32FC1`` ).
+
+ :param y: Source matrix containing imaginary components ( ``CV_32FC1`` ).
+
+ :param magnitude: Destination matrix of float magnitudes ( ``CV_32FC1`` ).
+
+ :param angle: Destination matrix of angles ( ``CV_32FC1`` ).
+
+ :param angleInDegrees: Flag for angles that must be evaluated in degrees.
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`cartToPolar`
+
+
+
+gpu::polarToCart
+--------------------
+Converts polar coordinates into Cartesian.
+
+.. ocv:function:: void gpu::polarToCart(const GpuMat& magnitude, const GpuMat& angle, GpuMat& x, GpuMat& y, bool angleInDegrees=false, Stream& stream = Stream::Null())
+
+ :param magnitude: Source matrix containing magnitudes ( ``CV_32FC1`` ).
+
+ :param angle: Source matrix containing angles ( ``CV_32FC1`` ).
+
+ :param x: Destination matrix of real components ( ``CV_32FC1`` ).
+
+ :param y: Destination matrix of imaginary components ( ``CV_32FC1`` ).
+
+ :param angleInDegrees: Flag that indicates angles in degrees.
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`polarToCart`
--- /dev/null
+*************************************************
+gpuarithm. GPU-accelerated Operations on Matrices
+*************************************************
+
+.. toctree::
+ :maxdepth: 1
+
+ core
+ element_operations
+ reductions
+ arithm
-gpu::meanStdDev
--------------------
-Computes a mean value and a standard deviation of matrix elements.
-
-.. ocv:function:: void gpu::meanStdDev(const GpuMat& mtx, Scalar& mean, Scalar& stddev)
-.. ocv:function:: void gpu::meanStdDev(const GpuMat& mtx, Scalar& mean, Scalar& stddev, GpuMat& buf)
-
- :param mtx: Source matrix. ``CV_8UC1`` matrices are supported for now.
-
- :param mean: Mean value.
-
- :param stddev: Standard deviation value.
-
- :param buf: Optional buffer to avoid extra memory allocations. It is resized automatically.
-
-.. seealso:: :ocv:func:`meanStdDev`
-
-
-
gpu::norm
-------------
Returns the norm of a matrix (or difference of two matrices).
The function ``reduce`` reduces the matrix to a vector by treating the matrix rows/columns as a set of 1D vectors and performing the specified operation on the vectors until a single row/column is obtained. For example, the function can be used to compute horizontal and vertical projections of a raster image. In case of ``CV_REDUCE_SUM`` and ``CV_REDUCE_AVG`` , the output may have a larger element bit-depth to preserve accuracy. And multi-channel arrays are also supported in these two reduction modes.
.. seealso:: :ocv:func:`reduce`
+
+
+
+gpu::normalize
+--------------
+Normalizes the norm or value range of an array.
+
+.. ocv:function:: void gpu::normalize(const GpuMat& src, GpuMat& dst, double alpha = 1, double beta = 0, int norm_type = NORM_L2, int dtype = -1, const GpuMat& mask = GpuMat())
+
+.. ocv:function:: void gpu::normalize(const GpuMat& src, GpuMat& dst, double a, double b, int norm_type, int dtype, const GpuMat& mask, GpuMat& norm_buf, GpuMat& cvt_buf)
+
+ :param src: input array.
+
+ :param dst: output array of the same size as ``src`` .
+
+ :param alpha: norm value to normalize to or the lower range boundary in case of the range normalization.
+
+ :param beta: upper range boundary in case of the range normalization; it is not used for the norm normalization.
+
+ :param normType: normalization type (see the details below).
+
+ :param dtype: when negative, the output array has the same type as ``src``; otherwise, it has the same number of channels as ``src`` and the depth ``=CV_MAT_DEPTH(dtype)``.
+
+ :param mask: optional operation mask.
+
+ :param norm_buf: Optional buffer to avoid extra memory allocations. It is resized automatically.
+
+ :param cvt_buf: Optional buffer to avoid extra memory allocations. It is resized automatically.
+
+.. seealso:: :ocv:func:`normalize`
+
+
+
+gpu::meanStdDev
+-------------------
+Computes a mean value and a standard deviation of matrix elements.
+
+.. ocv:function:: void gpu::meanStdDev(const GpuMat& mtx, Scalar& mean, Scalar& stddev)
+.. ocv:function:: void gpu::meanStdDev(const GpuMat& mtx, Scalar& mean, Scalar& stddev, GpuMat& buf)
+
+ :param mtx: Source matrix. ``CV_8UC1`` matrices are supported for now.
+
+ :param mean: Mean value.
+
+ :param stddev: Standard deviation value.
+
+ :param buf: Optional buffer to avoid extra memory allocations. It is resized automatically.
+
+.. seealso:: :ocv:func:`meanStdDev`
+
+
+
+gpu::rectStdDev
+-------------------
+Computes a standard deviation of integral images.
+
+.. ocv:function:: void gpu::rectStdDev(const GpuMat& src, const GpuMat& sqr, GpuMat& dst, const Rect& rect, Stream& stream = Stream::Null())
+
+ :param src: Source image. Only the ``CV_32SC1`` type is supported.
+
+ :param sqr: Squared source image. Only the ``CV_32FC1`` type is supported.
+
+ :param dst: Destination image with the same type and size as ``src`` .
+
+ :param rect: Rectangular window.
+
+ :param stream: Stream for the asynchronous version.
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_GPUARITHM_HPP__
+#define __OPENCV_GPUARITHM_HPP__
+
+#ifndef __cplusplus
+# error gpuarithm.hpp header must be compiled as C++
+#endif
+
+#include "opencv2/core/gpumat.hpp"
+
+namespace cv { namespace gpu {
+
+//! adds one matrix to another (c = a + b)
+CV_EXPORTS void add(const GpuMat& a, const GpuMat& b, GpuMat& c, const GpuMat& mask = GpuMat(), int dtype = -1, Stream& stream = Stream::Null());
+//! adds scalar to a matrix (c = a + s)
+CV_EXPORTS void add(const GpuMat& a, const Scalar& sc, GpuMat& c, const GpuMat& mask = GpuMat(), int dtype = -1, Stream& stream = Stream::Null());
+
+//! subtracts one matrix from another (c = a - b)
+CV_EXPORTS void subtract(const GpuMat& a, const GpuMat& b, GpuMat& c, const GpuMat& mask = GpuMat(), int dtype = -1, Stream& stream = Stream::Null());
+//! subtracts scalar from a matrix (c = a - s)
+CV_EXPORTS void subtract(const GpuMat& a, const Scalar& sc, GpuMat& c, const GpuMat& mask = GpuMat(), int dtype = -1, Stream& stream = Stream::Null());
+
+//! computes element-wise weighted product of the two arrays (c = scale * a * b)
+CV_EXPORTS void multiply(const GpuMat& a, const GpuMat& b, GpuMat& c, double scale = 1, int dtype = -1, Stream& stream = Stream::Null());
+//! weighted multiplies matrix to a scalar (c = scale * a * s)
+CV_EXPORTS void multiply(const GpuMat& a, const Scalar& sc, GpuMat& c, double scale = 1, int dtype = -1, Stream& stream = Stream::Null());
+
+//! computes element-wise weighted quotient of the two arrays (c = a / b)
+CV_EXPORTS void divide(const GpuMat& a, const GpuMat& b, GpuMat& c, double scale = 1, int dtype = -1, Stream& stream = Stream::Null());
+//! computes element-wise weighted quotient of matrix and scalar (c = a / s)
+CV_EXPORTS void divide(const GpuMat& a, const Scalar& sc, GpuMat& c, double scale = 1, int dtype = -1, Stream& stream = Stream::Null());
+//! computes element-wise weighted reciprocal of an array (dst = scale/src2)
+CV_EXPORTS void divide(double scale, const GpuMat& b, GpuMat& c, int dtype = -1, Stream& stream = Stream::Null());
+
+//! computes the weighted sum of two arrays (dst = alpha*src1 + beta*src2 + gamma)
+CV_EXPORTS void addWeighted(const GpuMat& src1, double alpha, const GpuMat& src2, double beta, double gamma, GpuMat& dst,
+ int dtype = -1, Stream& stream = Stream::Null());
+
+//! adds scaled array to another one (dst = alpha*src1 + src2)
+static inline void scaleAdd(const GpuMat& src1, double alpha, const GpuMat& src2, GpuMat& dst, Stream& stream = Stream::Null())
+{
+ addWeighted(src1, alpha, src2, 1.0, 0.0, dst, -1, stream);
+}
+
+//! computes element-wise absolute difference of two arrays (c = abs(a - b))
+CV_EXPORTS void absdiff(const GpuMat& a, const GpuMat& b, GpuMat& c, Stream& stream = Stream::Null());
+//! computes element-wise absolute difference of array and scalar (c = abs(a - s))
+CV_EXPORTS void absdiff(const GpuMat& a, const Scalar& s, GpuMat& c, Stream& stream = Stream::Null());
+
+//! computes absolute value of each matrix element
+//! supports CV_16S and CV_32F depth
+CV_EXPORTS void abs(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null());
+
+//! computes square of each pixel in an image
+//! supports CV_8U, CV_16U, CV_16S and CV_32F depth
+CV_EXPORTS void sqr(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null());
+
+//! computes square root of each pixel in an image
+//! supports CV_8U, CV_16U, CV_16S and CV_32F depth
+CV_EXPORTS void sqrt(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null());
+
+//! computes exponent of each matrix element (b = e**a)
+//! supports CV_8U, CV_16U, CV_16S and CV_32F depth
+CV_EXPORTS void exp(const GpuMat& a, GpuMat& b, Stream& stream = Stream::Null());
+
+//! computes natural logarithm of absolute value of each matrix element: b = log(abs(a))
+//! supports CV_8U, CV_16U, CV_16S and CV_32F depth
+CV_EXPORTS void log(const GpuMat& a, GpuMat& b, Stream& stream = Stream::Null());
+
+//! computes power of each matrix element:
+// (dst(i,j) = pow( src(i,j) , power), if src.type() is integer
+// (dst(i,j) = pow(fabs(src(i,j)), power), otherwise
+//! supports all, except depth == CV_64F
+CV_EXPORTS void pow(const GpuMat& src, double power, GpuMat& dst, Stream& stream = Stream::Null());
+
+//! compares elements of two arrays (c = a <cmpop> b)
+CV_EXPORTS void compare(const GpuMat& a, const GpuMat& b, GpuMat& c, int cmpop, Stream& stream = Stream::Null());
+CV_EXPORTS void compare(const GpuMat& a, Scalar sc, GpuMat& c, int cmpop, Stream& stream = Stream::Null());
+
+//! performs per-elements bit-wise inversion
+CV_EXPORTS void bitwise_not(const GpuMat& src, GpuMat& dst, const GpuMat& mask=GpuMat(), Stream& stream = Stream::Null());
+
+//! calculates per-element bit-wise disjunction of two arrays
+CV_EXPORTS void bitwise_or(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, const GpuMat& mask=GpuMat(), Stream& stream = Stream::Null());
+//! calculates per-element bit-wise disjunction of array and scalar
+//! supports 1, 3 and 4 channels images with CV_8U, CV_16U or CV_32S depth
+CV_EXPORTS void bitwise_or(const GpuMat& src1, const Scalar& sc, GpuMat& dst, Stream& stream = Stream::Null());
+
+//! calculates per-element bit-wise conjunction of two arrays
+CV_EXPORTS void bitwise_and(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, const GpuMat& mask=GpuMat(), Stream& stream = Stream::Null());
+//! calculates per-element bit-wise conjunction of array and scalar
+//! supports 1, 3 and 4 channels images with CV_8U, CV_16U or CV_32S depth
+CV_EXPORTS void bitwise_and(const GpuMat& src1, const Scalar& sc, GpuMat& dst, Stream& stream = Stream::Null());
+
+//! calculates per-element bit-wise "exclusive or" operation
+CV_EXPORTS void bitwise_xor(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, const GpuMat& mask=GpuMat(), Stream& stream = Stream::Null());
+//! calculates per-element bit-wise "exclusive or" of array and scalar
+//! supports 1, 3 and 4 channels images with CV_8U, CV_16U or CV_32S depth
+CV_EXPORTS void bitwise_xor(const GpuMat& src1, const Scalar& sc, GpuMat& dst, Stream& stream = Stream::Null());
+
+//! pixel by pixel right shift of an image by a constant value
+//! supports 1, 3 and 4 channels images with integers elements
+CV_EXPORTS void rshift(const GpuMat& src, Scalar_<int> sc, GpuMat& dst, Stream& stream = Stream::Null());
+
+//! pixel by pixel left shift of an image by a constant value
+//! supports 1, 3 and 4 channels images with CV_8U, CV_16U or CV_32S depth
+CV_EXPORTS void lshift(const GpuMat& src, Scalar_<int> sc, GpuMat& dst, Stream& stream = Stream::Null());
+
+//! computes per-element minimum of two arrays (dst = min(src1, src2))
+CV_EXPORTS void min(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, Stream& stream = Stream::Null());
+
+//! computes per-element minimum of array and scalar (dst = min(src1, src2))
+CV_EXPORTS void min(const GpuMat& src1, double src2, GpuMat& dst, Stream& stream = Stream::Null());
+
+//! computes per-element maximum of two arrays (dst = max(src1, src2))
+CV_EXPORTS void max(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, Stream& stream = Stream::Null());
+
+//! computes per-element maximum of array and scalar (dst = max(src1, src2))
+CV_EXPORTS void max(const GpuMat& src1, double src2, GpuMat& dst, Stream& stream = Stream::Null());
+
+//! implements generalized matrix product algorithm GEMM from BLAS
+CV_EXPORTS void gemm(const GpuMat& src1, const GpuMat& src2, double alpha,
+ const GpuMat& src3, double beta, GpuMat& dst, int flags = 0, Stream& stream = Stream::Null());
+
+//! transposes the matrix
+//! supports matrix with element size = 1, 4 and 8 bytes (CV_8UC1, CV_8UC4, CV_16UC2, CV_32FC1, etc)
+CV_EXPORTS void transpose(const GpuMat& src1, GpuMat& dst, Stream& stream = Stream::Null());
+
+//! reverses the order of the rows, columns or both in a matrix
+//! supports 1, 3 and 4 channels images with CV_8U, CV_16U, CV_32S or CV_32F depth
+CV_EXPORTS void flip(const GpuMat& a, GpuMat& b, int flipCode, Stream& stream = Stream::Null());
+
+//! transforms 8-bit unsigned integers using lookup table: dst(i)=lut(src(i))
+//! destination array will have the depth type as lut and the same channels number as source
+//! supports CV_8UC1, CV_8UC3 types
+CV_EXPORTS void LUT(const GpuMat& src, const Mat& lut, GpuMat& dst, Stream& stream = Stream::Null());
+
+//! makes multi-channel array out of several single-channel arrays
+CV_EXPORTS void merge(const GpuMat* src, size_t n, GpuMat& dst, Stream& stream = Stream::Null());
+
+//! makes multi-channel array out of several single-channel arrays
+CV_EXPORTS void merge(const std::vector<GpuMat>& src, GpuMat& dst, Stream& stream = Stream::Null());
+
+//! copies each plane of a multi-channel array to a dedicated array
+CV_EXPORTS void split(const GpuMat& src, GpuMat* dst, Stream& stream = Stream::Null());
+
+//! copies each plane of a multi-channel array to a dedicated array
+CV_EXPORTS void split(const GpuMat& src, std::vector<GpuMat>& dst, Stream& stream = Stream::Null());
+
+//! computes magnitude of complex (x(i).re, x(i).im) vector
+//! supports only CV_32FC2 type
+CV_EXPORTS void magnitude(const GpuMat& xy, GpuMat& magnitude, Stream& stream = Stream::Null());
+
+//! computes squared magnitude of complex (x(i).re, x(i).im) vector
+//! supports only CV_32FC2 type
+CV_EXPORTS void magnitudeSqr(const GpuMat& xy, GpuMat& magnitude, Stream& stream = Stream::Null());
+
+//! computes magnitude of each (x(i), y(i)) vector
+//! supports only floating-point source
+CV_EXPORTS void magnitude(const GpuMat& x, const GpuMat& y, GpuMat& magnitude, Stream& stream = Stream::Null());
+
+//! computes squared magnitude of each (x(i), y(i)) vector
+//! supports only floating-point source
+CV_EXPORTS void magnitudeSqr(const GpuMat& x, const GpuMat& y, GpuMat& magnitude, Stream& stream = Stream::Null());
+
+//! computes angle (angle(i)) of each (x(i), y(i)) vector
+//! supports only floating-point source
+CV_EXPORTS void phase(const GpuMat& x, const GpuMat& y, GpuMat& angle, bool angleInDegrees = false, Stream& stream = Stream::Null());
+
+//! converts Cartesian coordinates to polar
+//! supports only floating-point source
+CV_EXPORTS void cartToPolar(const GpuMat& x, const GpuMat& y, GpuMat& magnitude, GpuMat& angle, bool angleInDegrees = false, Stream& stream = Stream::Null());
+
+//! converts polar coordinates to Cartesian
+//! supports only floating-point source
+CV_EXPORTS void polarToCart(const GpuMat& magnitude, const GpuMat& angle, GpuMat& x, GpuMat& y, bool angleInDegrees = false, Stream& stream = Stream::Null());
+
+//! scales and shifts array elements so that either the specified norm (alpha) or the minimum (alpha) and maximum (beta) array values get the specified values
+CV_EXPORTS void normalize(const GpuMat& src, GpuMat& dst, double alpha = 1, double beta = 0,
+ int norm_type = NORM_L2, int dtype = -1, const GpuMat& mask = GpuMat());
+CV_EXPORTS void normalize(const GpuMat& src, GpuMat& dst, double a, double b,
+ int norm_type, int dtype, const GpuMat& mask, GpuMat& norm_buf, GpuMat& cvt_buf);
+
+//! computes norm of array
+//! supports NORM_INF, NORM_L1, NORM_L2
+//! supports all matrices except 64F
+CV_EXPORTS double norm(const GpuMat& src1, int normType=NORM_L2);
+CV_EXPORTS double norm(const GpuMat& src1, int normType, GpuMat& buf);
+CV_EXPORTS double norm(const GpuMat& src1, int normType, const GpuMat& mask, GpuMat& buf);
+
+//! computes norm of the difference between two arrays
+//! supports NORM_INF, NORM_L1, NORM_L2
+//! supports only CV_8UC1 type
+CV_EXPORTS double norm(const GpuMat& src1, const GpuMat& src2, int normType=NORM_L2);
+
+//! computes sum of array elements
+//! supports only single channel images
+CV_EXPORTS Scalar sum(const GpuMat& src);
+CV_EXPORTS Scalar sum(const GpuMat& src, GpuMat& buf);
+CV_EXPORTS Scalar sum(const GpuMat& src, const GpuMat& mask, GpuMat& buf);
+
+//! computes sum of array elements absolute values
+//! supports only single channel images
+CV_EXPORTS Scalar absSum(const GpuMat& src);
+CV_EXPORTS Scalar absSum(const GpuMat& src, GpuMat& buf);
+CV_EXPORTS Scalar absSum(const GpuMat& src, const GpuMat& mask, GpuMat& buf);
+
+//! computes squared sum of array elements
+//! supports only single channel images
+CV_EXPORTS Scalar sqrSum(const GpuMat& src);
+CV_EXPORTS Scalar sqrSum(const GpuMat& src, GpuMat& buf);
+CV_EXPORTS Scalar sqrSum(const GpuMat& src, const GpuMat& mask, GpuMat& buf);
+
+//! finds global minimum and maximum array elements and returns their values
+CV_EXPORTS void minMax(const GpuMat& src, double* minVal, double* maxVal=0, const GpuMat& mask=GpuMat());
+CV_EXPORTS void minMax(const GpuMat& src, double* minVal, double* maxVal, const GpuMat& mask, GpuMat& buf);
+
+//! finds global minimum and maximum array elements and returns their values with locations
+CV_EXPORTS void minMaxLoc(const GpuMat& src, double* minVal, double* maxVal=0, Point* minLoc=0, Point* maxLoc=0,
+ const GpuMat& mask=GpuMat());
+CV_EXPORTS void minMaxLoc(const GpuMat& src, double* minVal, double* maxVal, Point* minLoc, Point* maxLoc,
+ const GpuMat& mask, GpuMat& valbuf, GpuMat& locbuf);
+
+//! counts non-zero array elements
+CV_EXPORTS int countNonZero(const GpuMat& src);
+CV_EXPORTS int countNonZero(const GpuMat& src, GpuMat& buf);
+
+//! reduces a matrix to a vector
+CV_EXPORTS void reduce(const GpuMat& mtx, GpuMat& vec, int dim, int reduceOp, int dtype = -1, Stream& stream = Stream::Null());
+
+//! computes mean value and standard deviation of all or selected array elements
+//! supports only CV_8UC1 type
+CV_EXPORTS void meanStdDev(const GpuMat& mtx, Scalar& mean, Scalar& stddev);
+//! buffered version
+CV_EXPORTS void meanStdDev(const GpuMat& mtx, Scalar& mean, Scalar& stddev, GpuMat& buf);
+
+//! computes the standard deviation of integral images
+//! supports only CV_32SC1 source type and CV_32FC1 sqr type
+//! output will have CV_32FC1 type
+CV_EXPORTS void rectStdDev(const GpuMat& src, const GpuMat& sqr, GpuMat& dst, const Rect& rect, Stream& stream = Stream::Null());
+
+//! copies 2D array to a larger destination array and pads borders with user-specifiable constant
+CV_EXPORTS void copyMakeBorder(const GpuMat& src, GpuMat& dst, int top, int bottom, int left, int right, int borderType,
+ const Scalar& value = Scalar(), Stream& stream = Stream::Null());
+
+//! applies fixed threshold to the image
+CV_EXPORTS double threshold(const GpuMat& src, GpuMat& dst, double thresh, double maxval, int type, Stream& stream = Stream::Null());
+
+//! computes the integral image
+//! sum will have CV_32S type, but will contain unsigned int values
+//! supports only CV_8UC1 source type
+CV_EXPORTS void integral(const GpuMat& src, GpuMat& sum, Stream& stream = Stream::Null());
+//! buffered version
+CV_EXPORTS void integralBuffered(const GpuMat& src, GpuMat& sum, GpuMat& buffer, Stream& stream = Stream::Null());
+
+//! computes squared integral image
+//! result matrix will have 64F type, but will contain 64U values
+//! supports source images of 8UC1 type only
+CV_EXPORTS void sqrIntegral(const GpuMat& src, GpuMat& sqsum, Stream& stream = Stream::Null());
+
+//! performs per-element multiplication of two full (not packed) Fourier spectrums
+//! supports 32FC2 matrixes only (interleaved format)
+CV_EXPORTS void mulSpectrums(const GpuMat& a, const GpuMat& b, GpuMat& c, int flags, bool conjB=false, Stream& stream = Stream::Null());
+
+//! performs per-element multiplication of two full (not packed) Fourier spectrums
+//! supports 32FC2 matrixes only (interleaved format)
+CV_EXPORTS void mulAndScaleSpectrums(const GpuMat& a, const GpuMat& b, GpuMat& c, int flags, float scale, bool conjB=false, Stream& stream = Stream::Null());
+
+//! Performs a forward or inverse discrete Fourier transform (1D or 2D) of floating point matrix.
+//! Param dft_size is the size of DFT transform.
+//!
+//! If the source matrix is not continous, then additional copy will be done,
+//! so to avoid copying ensure the source matrix is continous one. If you want to use
+//! preallocated output ensure it is continuous too, otherwise it will be reallocated.
+//!
+//! Being implemented via CUFFT real-to-complex transform result contains only non-redundant values
+//! in CUFFT's format. Result as full complex matrix for such kind of transform cannot be retrieved.
+//!
+//! For complex-to-real transform it is assumed that the source matrix is packed in CUFFT's format.
+CV_EXPORTS void dft(const GpuMat& src, GpuMat& dst, Size dft_size, int flags=0, Stream& stream = Stream::Null());
+
+struct CV_EXPORTS ConvolveBuf
+{
+ Size result_size;
+ Size block_size;
+ Size user_block_size;
+ Size dft_size;
+ int spect_len;
+
+ GpuMat image_spect, templ_spect, result_spect;
+ GpuMat image_block, templ_block, result_data;
+
+ void create(Size image_size, Size templ_size);
+ static Size estimateBlockSize(Size result_size, Size templ_size);
+};
+
+//! computes convolution (or cross-correlation) of two images using discrete Fourier transform
+//! supports source images of 32FC1 type only
+//! result matrix will have 32FC1 type
+CV_EXPORTS void convolve(const GpuMat& image, const GpuMat& templ, GpuMat& result, bool ccorr = false);
+CV_EXPORTS void convolve(const GpuMat& image, const GpuMat& templ, GpuMat& result, bool ccorr, ConvolveBuf& buf, Stream& stream = Stream::Null());
+
+}} // namespace cv { namespace gpu {
+
+#endif /* __OPENCV_GPUARITHM_HPP__ */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace std;
+using namespace testing;
+using namespace perf;
+
+//////////////////////////////////////////////////////////////////////
+// GEMM
+
+CV_FLAGS(GemmFlags, 0, cv::GEMM_1_T, cv::GEMM_2_T, cv::GEMM_3_T)
+#define ALL_GEMM_FLAGS Values(GemmFlags(0), GemmFlags(cv::GEMM_1_T), GemmFlags(cv::GEMM_2_T), GemmFlags(cv::GEMM_3_T), \
+ GemmFlags(cv::GEMM_1_T | cv::GEMM_2_T), GemmFlags(cv::GEMM_1_T | cv::GEMM_3_T), GemmFlags(cv::GEMM_1_T | cv::GEMM_2_T | cv::GEMM_3_T))
+
+DEF_PARAM_TEST(Sz_Type_Flags, cv::Size, MatType, GemmFlags);
+
+PERF_TEST_P(Sz_Type_Flags, GEMM,
+ Combine(Values(cv::Size(512, 512), cv::Size(1024, 1024)),
+ Values(CV_32FC1, CV_32FC2, CV_64FC1),
+ ALL_GEMM_FLAGS))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int type = GET_PARAM(1);
+ const int flags = GET_PARAM(2);
+
+ cv::Mat src1(size, type);
+ declare.in(src1, WARMUP_RNG);
+
+ cv::Mat src2(size, type);
+ declare.in(src2, WARMUP_RNG);
+
+ cv::Mat src3(size, type);
+ declare.in(src3, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ declare.time(5.0);
+
+ const cv::gpu::GpuMat d_src1(src1);
+ const cv::gpu::GpuMat d_src2(src2);
+ const cv::gpu::GpuMat d_src3(src3);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::gemm(d_src1, d_src2, 1.0, d_src3, 1.0, dst, flags);
+
+ GPU_SANITY_CHECK(dst, 1e-6);
+ }
+ else
+ {
+ declare.time(50.0);
+
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::gemm(src1, src2, 1.0, src3, 1.0, dst, flags);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// MulSpectrums
+
+CV_FLAGS(DftFlags, 0, cv::DFT_INVERSE, cv::DFT_SCALE, cv::DFT_ROWS, cv::DFT_COMPLEX_OUTPUT, cv::DFT_REAL_OUTPUT)
+
+DEF_PARAM_TEST(Sz_Flags, cv::Size, DftFlags);
+
+PERF_TEST_P(Sz_Flags, MulSpectrums,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(0, DftFlags(cv::DFT_ROWS))))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int flag = GET_PARAM(1);
+
+ cv::Mat a(size, CV_32FC2);
+ cv::Mat b(size, CV_32FC2);
+ declare.in(a, b, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_a(a);
+ const cv::gpu::GpuMat d_b(b);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::mulSpectrums(d_a, d_b, dst, flag);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::mulSpectrums(a, b, dst, flag);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// MulAndScaleSpectrums
+
+PERF_TEST_P(Sz, MulAndScaleSpectrums,
+ GPU_TYPICAL_MAT_SIZES)
+{
+ const cv::Size size = GetParam();
+
+ const float scale = 1.f / size.area();
+
+ cv::Mat src1(size, CV_32FC2);
+ cv::Mat src2(size, CV_32FC2);
+ declare.in(src1,src2, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src1(src1);
+ const cv::gpu::GpuMat d_src2(src2);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::mulAndScaleSpectrums(d_src1, d_src2, dst, cv::DFT_ROWS, scale, false);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// Dft
+
+PERF_TEST_P(Sz_Flags, Dft,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(0, DftFlags(cv::DFT_ROWS), DftFlags(cv::DFT_INVERSE))))
+{
+ declare.time(10.0);
+
+ const cv::Size size = GET_PARAM(0);
+ const int flag = GET_PARAM(1);
+
+ cv::Mat src(size, CV_32FC2);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::dft(d_src, dst, size, flag);
+
+ GPU_SANITY_CHECK(dst, 1e-6, ERROR_RELATIVE);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::dft(src, dst, flag);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// Convolve
+
+DEF_PARAM_TEST(Sz_KernelSz_Ccorr, cv::Size, int, bool);
+
+PERF_TEST_P(Sz_KernelSz_Ccorr, Convolve,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(17, 27, 32, 64),
+ Bool()))
+{
+ declare.time(10.0);
+
+ const cv::Size size = GET_PARAM(0);
+ const int templ_size = GET_PARAM(1);
+ const bool ccorr = GET_PARAM(2);
+
+ const cv::Mat image(size, CV_32FC1);
+ const cv::Mat templ(templ_size, templ_size, CV_32FC1);
+ declare.in(image, templ, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ cv::gpu::GpuMat d_image = cv::gpu::createContinuous(size, CV_32FC1);
+ d_image.upload(image);
+
+ cv::gpu::GpuMat d_templ = cv::gpu::createContinuous(templ_size, templ_size, CV_32FC1);
+ d_templ.upload(templ);
+
+ cv::gpu::GpuMat dst;
+ cv::gpu::ConvolveBuf d_buf;
+
+ TEST_CYCLE() cv::gpu::convolve(d_image, d_templ, dst, ccorr, d_buf);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ if (ccorr)
+ FAIL_NO_CPU();
+
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::filter2D(image, dst, image.depth(), templ);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// Integral
+
+PERF_TEST_P(Sz, Integral,
+ GPU_TYPICAL_MAT_SIZES)
+{
+ const cv::Size size = GetParam();
+
+ cv::Mat src(size, CV_8UC1);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+ cv::gpu::GpuMat d_buf;
+
+ TEST_CYCLE() cv::gpu::integralBuffered(d_src, dst, d_buf);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::integral(src, dst);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// IntegralSqr
+
+PERF_TEST_P(Sz, IntegralSqr,
+ GPU_TYPICAL_MAT_SIZES)
+{
+ const cv::Size size = GetParam();
+
+ cv::Mat src(size, CV_8UC1);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::sqrIntegral(d_src, dst);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace std;
+using namespace testing;
+using namespace perf;
+
+#define ARITHM_MAT_DEPTH Values(CV_8U, CV_16U, CV_32F, CV_64F)
+
+//////////////////////////////////////////////////////////////////////
+// Merge
+
+PERF_TEST_P(Sz_Depth_Cn, Merge,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ ARITHM_MAT_DEPTH,
+ Values(2, 3, 4)))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+
+ std::vector<cv::Mat> src(channels);
+ for (int i = 0; i < channels; ++i)
+ {
+ src[i].create(size, depth);
+ declare.in(src[i], WARMUP_RNG);
+ }
+
+ if (PERF_RUN_GPU())
+ {
+ std::vector<cv::gpu::GpuMat> d_src(channels);
+ for (int i = 0; i < channels; ++i)
+ d_src[i].upload(src[i]);
+
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::merge(d_src, dst);
+
+ GPU_SANITY_CHECK(dst, 1e-10);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::merge(src, dst);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// Split
+
+PERF_TEST_P(Sz_Depth_Cn, Split,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ ARITHM_MAT_DEPTH,
+ Values(2, 3, 4)))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+
+ cv::Mat src(size, CV_MAKE_TYPE(depth, channels));
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ std::vector<cv::gpu::GpuMat> dst;
+
+ TEST_CYCLE() cv::gpu::split(d_src, dst);
+
+ const cv::gpu::GpuMat& dst0 = dst[0];
+ const cv::gpu::GpuMat& dst1 = dst[1];
+
+ GPU_SANITY_CHECK(dst0, 1e-10);
+ GPU_SANITY_CHECK(dst1, 1e-10);
+ }
+ else
+ {
+ std::vector<cv::Mat> dst;
+
+ TEST_CYCLE() cv::split(src, dst);
+
+ const cv::Mat& dst0 = dst[0];
+ const cv::Mat& dst1 = dst[1];
+
+ CPU_SANITY_CHECK(dst0);
+ CPU_SANITY_CHECK(dst1);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// Transpose
+
+PERF_TEST_P(Sz_Type, Transpose,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8UC1, CV_8UC4, CV_16UC2, CV_16SC2, CV_32SC1, CV_32SC2, CV_64FC1)))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int type = GET_PARAM(1);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::transpose(d_src, dst);
+
+ GPU_SANITY_CHECK(dst, 1e-10);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::transpose(src, dst);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// Flip
+
+enum {FLIP_BOTH = 0, FLIP_X = 1, FLIP_Y = -1};
+CV_ENUM(FlipCode, FLIP_BOTH, FLIP_X, FLIP_Y)
+
+DEF_PARAM_TEST(Sz_Depth_Cn_Code, cv::Size, MatDepth, MatCn, FlipCode);
+
+PERF_TEST_P(Sz_Depth_Cn_Code, Flip,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_32F),
+ GPU_CHANNELS_1_3_4,
+ FlipCode::all()))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+ const int flipCode = GET_PARAM(3);
+
+ const int type = CV_MAKE_TYPE(depth, channels);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::flip(d_src, dst, flipCode);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::flip(src, dst, flipCode);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// LutOneChannel
+
+PERF_TEST_P(Sz_Type, LutOneChannel,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8UC1, CV_8UC3)))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int type = GET_PARAM(1);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ cv::Mat lut(1, 256, CV_8UC1);
+ declare.in(lut, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::LUT(d_src, lut, dst);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::LUT(src, lut, dst);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// LutMultiChannel
+
+PERF_TEST_P(Sz_Type, LutMultiChannel,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values<MatType>(CV_8UC3)))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int type = GET_PARAM(1);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ cv::Mat lut(1, 256, CV_MAKE_TYPE(CV_8U, src.channels()));
+ declare.in(lut, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::LUT(d_src, lut, dst);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::LUT(src, lut, dst);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// CopyMakeBorder
+
+DEF_PARAM_TEST(Sz_Depth_Cn_Border, cv::Size, MatDepth, MatCn, BorderMode);
+
+PERF_TEST_P(Sz_Depth_Cn_Border, CopyMakeBorder,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_32F),
+ GPU_CHANNELS_1_3_4,
+ ALL_BORDER_MODES))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+ const int borderMode = GET_PARAM(3);
+
+ const int type = CV_MAKE_TYPE(depth, channels);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::copyMakeBorder(d_src, dst, 5, 5, 5, 5, borderMode);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::copyMakeBorder(src, dst, 5, 5, 5, 5, borderMode);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
#define ARITHM_MAT_DEPTH Values(CV_8U, CV_16U, CV_32F, CV_64F)
//////////////////////////////////////////////////////////////////////
-// Merge
-
-PERF_TEST_P(Sz_Depth_Cn, Core_Merge,
- Combine(GPU_TYPICAL_MAT_SIZES,
- ARITHM_MAT_DEPTH,
- Values(2, 3, 4)))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
-
- std::vector<cv::Mat> src(channels);
- for (int i = 0; i < channels; ++i)
- {
- src[i].create(size, depth);
- declare.in(src[i], WARMUP_RNG);
- }
-
- if (PERF_RUN_GPU())
- {
- std::vector<cv::gpu::GpuMat> d_src(channels);
- for (int i = 0; i < channels; ++i)
- d_src[i].upload(src[i]);
-
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::merge(d_src, dst);
-
- GPU_SANITY_CHECK(dst, 1e-10);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::merge(src, dst);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// Split
-
-PERF_TEST_P(Sz_Depth_Cn, Core_Split,
- Combine(GPU_TYPICAL_MAT_SIZES,
- ARITHM_MAT_DEPTH,
- Values(2, 3, 4)))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
-
- cv::Mat src(size, CV_MAKE_TYPE(depth, channels));
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- std::vector<cv::gpu::GpuMat> dst;
-
- TEST_CYCLE() cv::gpu::split(d_src, dst);
-
- const cv::gpu::GpuMat& dst0 = dst[0];
- const cv::gpu::GpuMat& dst1 = dst[1];
-
- GPU_SANITY_CHECK(dst0, 1e-10);
- GPU_SANITY_CHECK(dst1, 1e-10);
- }
- else
- {
- std::vector<cv::Mat> dst;
-
- TEST_CYCLE() cv::split(src, dst);
-
- const cv::Mat& dst0 = dst[0];
- const cv::Mat& dst1 = dst[1];
-
- CPU_SANITY_CHECK(dst0);
- CPU_SANITY_CHECK(dst1);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
// AddMat
-PERF_TEST_P(Sz_Depth, Core_AddMat,
+PERF_TEST_P(Sz_Depth, AddMat,
Combine(GPU_TYPICAL_MAT_SIZES,
ARITHM_MAT_DEPTH))
{
//////////////////////////////////////////////////////////////////////
// AddScalar
-PERF_TEST_P(Sz_Depth, Core_AddScalar,
+PERF_TEST_P(Sz_Depth, AddScalar,
Combine(GPU_TYPICAL_MAT_SIZES,
ARITHM_MAT_DEPTH))
{
//////////////////////////////////////////////////////////////////////
// SubtractMat
-PERF_TEST_P(Sz_Depth, Core_SubtractMat,
+PERF_TEST_P(Sz_Depth, SubtractMat,
Combine(GPU_TYPICAL_MAT_SIZES,
ARITHM_MAT_DEPTH))
{
//////////////////////////////////////////////////////////////////////
// SubtractScalar
-PERF_TEST_P(Sz_Depth, Core_SubtractScalar,
+PERF_TEST_P(Sz_Depth, SubtractScalar,
Combine(GPU_TYPICAL_MAT_SIZES,
ARITHM_MAT_DEPTH))
{
//////////////////////////////////////////////////////////////////////
// MultiplyMat
-PERF_TEST_P(Sz_Depth, Core_MultiplyMat,
+PERF_TEST_P(Sz_Depth, MultiplyMat,
Combine(GPU_TYPICAL_MAT_SIZES,
ARITHM_MAT_DEPTH))
{
//////////////////////////////////////////////////////////////////////
// MultiplyScalar
-PERF_TEST_P(Sz_Depth, Core_MultiplyScalar,
+PERF_TEST_P(Sz_Depth, MultiplyScalar,
Combine(GPU_TYPICAL_MAT_SIZES,
ARITHM_MAT_DEPTH))
{
//////////////////////////////////////////////////////////////////////
// DivideMat
-PERF_TEST_P(Sz_Depth, Core_DivideMat,
+PERF_TEST_P(Sz_Depth, DivideMat,
Combine(GPU_TYPICAL_MAT_SIZES,
ARITHM_MAT_DEPTH))
{
//////////////////////////////////////////////////////////////////////
// DivideScalar
-PERF_TEST_P(Sz_Depth, Core_DivideScalar,
+PERF_TEST_P(Sz_Depth, DivideScalar,
Combine(GPU_TYPICAL_MAT_SIZES,
ARITHM_MAT_DEPTH))
{
//////////////////////////////////////////////////////////////////////
// DivideScalarInv
-PERF_TEST_P(Sz_Depth, Core_DivideScalarInv,
+PERF_TEST_P(Sz_Depth, DivideScalarInv,
Combine(GPU_TYPICAL_MAT_SIZES,
ARITHM_MAT_DEPTH))
{
//////////////////////////////////////////////////////////////////////
// AbsDiffMat
-PERF_TEST_P(Sz_Depth, Core_AbsDiffMat,
+PERF_TEST_P(Sz_Depth, AbsDiffMat,
Combine(GPU_TYPICAL_MAT_SIZES,
ARITHM_MAT_DEPTH))
{
//////////////////////////////////////////////////////////////////////
// AbsDiffScalar
-PERF_TEST_P(Sz_Depth, Core_AbsDiffScalar,
+PERF_TEST_P(Sz_Depth, AbsDiffScalar,
Combine(GPU_TYPICAL_MAT_SIZES,
ARITHM_MAT_DEPTH))
{
//////////////////////////////////////////////////////////////////////
// Abs
-PERF_TEST_P(Sz_Depth, Core_Abs,
+PERF_TEST_P(Sz_Depth, Abs,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_16S, CV_32F)))
{
//////////////////////////////////////////////////////////////////////
// Sqr
-PERF_TEST_P(Sz_Depth, Core_Sqr,
+PERF_TEST_P(Sz_Depth, Sqr,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8U, CV_16S, CV_32F)))
{
//////////////////////////////////////////////////////////////////////
// Sqrt
-PERF_TEST_P(Sz_Depth, Core_Sqrt,
+PERF_TEST_P(Sz_Depth, Sqrt,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8U, CV_16S, CV_32F)))
{
//////////////////////////////////////////////////////////////////////
// Log
-PERF_TEST_P(Sz_Depth, Core_Log,
+PERF_TEST_P(Sz_Depth, Log,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8U, CV_16S, CV_32F)))
{
//////////////////////////////////////////////////////////////////////
// Exp
-PERF_TEST_P(Sz_Depth, Core_Exp,
+PERF_TEST_P(Sz_Depth, Exp,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8U, CV_16S, CV_32F)))
{
DEF_PARAM_TEST(Sz_Depth_Power, cv::Size, MatDepth, double);
-PERF_TEST_P(Sz_Depth_Power, Core_Pow,
+PERF_TEST_P(Sz_Depth_Power, Pow,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8U, CV_16S, CV_32F),
Values(0.3, 2.0, 2.4)))
//////////////////////////////////////////////////////////////////////
// CompareMat
-CV_ENUM(CmpCode, CMP_EQ, CMP_GT, CMP_GE, CMP_LT, CMP_LE, CMP_NE)
+CV_ENUM(CmpCode, cv::CMP_EQ, cv::CMP_GT, cv::CMP_GE, cv::CMP_LT, cv::CMP_LE, cv::CMP_NE)
DEF_PARAM_TEST(Sz_Depth_Code, cv::Size, MatDepth, CmpCode);
-PERF_TEST_P(Sz_Depth_Code, Core_CompareMat,
+PERF_TEST_P(Sz_Depth_Code, CompareMat,
Combine(GPU_TYPICAL_MAT_SIZES,
ARITHM_MAT_DEPTH,
CmpCode::all()))
//////////////////////////////////////////////////////////////////////
// CompareScalar
-PERF_TEST_P(Sz_Depth_Code, Core_CompareScalar,
+PERF_TEST_P(Sz_Depth_Code, CompareScalar,
Combine(GPU_TYPICAL_MAT_SIZES,
ARITHM_MAT_DEPTH,
CmpCode::all()))
//////////////////////////////////////////////////////////////////////
// BitwiseNot
-PERF_TEST_P(Sz_Depth, Core_BitwiseNot,
+PERF_TEST_P(Sz_Depth, BitwiseNot,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8U, CV_16U, CV_32S)))
{
//////////////////////////////////////////////////////////////////////
// BitwiseAndMat
-PERF_TEST_P(Sz_Depth, Core_BitwiseAndMat,
+PERF_TEST_P(Sz_Depth, BitwiseAndMat,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8U, CV_16U, CV_32S)))
{
//////////////////////////////////////////////////////////////////////
// BitwiseAndScalar
-PERF_TEST_P(Sz_Depth_Cn, Core_BitwiseAndScalar,
+PERF_TEST_P(Sz_Depth_Cn, BitwiseAndScalar,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8U, CV_16U, CV_32S),
GPU_CHANNELS_1_3_4))
//////////////////////////////////////////////////////////////////////
// BitwiseOrMat
-PERF_TEST_P(Sz_Depth, Core_BitwiseOrMat,
+PERF_TEST_P(Sz_Depth, BitwiseOrMat,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8U, CV_16U, CV_32S)))
{
//////////////////////////////////////////////////////////////////////
// BitwiseOrScalar
-PERF_TEST_P(Sz_Depth_Cn, Core_BitwiseOrScalar,
+PERF_TEST_P(Sz_Depth_Cn, BitwiseOrScalar,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8U, CV_16U, CV_32S),
GPU_CHANNELS_1_3_4))
//////////////////////////////////////////////////////////////////////
// BitwiseXorMat
-PERF_TEST_P(Sz_Depth, Core_BitwiseXorMat,
+PERF_TEST_P(Sz_Depth, BitwiseXorMat,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8U, CV_16U, CV_32S)))
{
//////////////////////////////////////////////////////////////////////
// BitwiseXorScalar
-PERF_TEST_P(Sz_Depth_Cn, Core_BitwiseXorScalar,
+PERF_TEST_P(Sz_Depth_Cn, BitwiseXorScalar,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8U, CV_16U, CV_32S),
GPU_CHANNELS_1_3_4))
//////////////////////////////////////////////////////////////////////
// RShift
-PERF_TEST_P(Sz_Depth_Cn, Core_RShift,
+PERF_TEST_P(Sz_Depth_Cn, RShift,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8U, CV_16U, CV_32S),
GPU_CHANNELS_1_3_4))
//////////////////////////////////////////////////////////////////////
// LShift
-PERF_TEST_P(Sz_Depth_Cn, Core_LShift,
+PERF_TEST_P(Sz_Depth_Cn, LShift,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8U, CV_16U, CV_32S),
GPU_CHANNELS_1_3_4))
//////////////////////////////////////////////////////////////////////
// MinMat
-PERF_TEST_P(Sz_Depth, Core_MinMat,
+PERF_TEST_P(Sz_Depth, MinMat,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8U, CV_16U, CV_32F)))
{
//////////////////////////////////////////////////////////////////////
// MinScalar
-PERF_TEST_P(Sz_Depth, Core_MinScalar,
+PERF_TEST_P(Sz_Depth, MinScalar,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8U, CV_16U, CV_32F)))
{
//////////////////////////////////////////////////////////////////////
// MaxMat
-PERF_TEST_P(Sz_Depth, Core_MaxMat,
+PERF_TEST_P(Sz_Depth, MaxMat,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8U, CV_16U, CV_32F)))
{
//////////////////////////////////////////////////////////////////////
// MaxScalar
-PERF_TEST_P(Sz_Depth, Core_MaxScalar,
+PERF_TEST_P(Sz_Depth, MaxScalar,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8U, CV_16U, CV_32F)))
{
DEF_PARAM_TEST(Sz_3Depth, cv::Size, MatDepth, MatDepth, MatDepth);
-PERF_TEST_P(Sz_3Depth, Core_AddWeighted,
+PERF_TEST_P(Sz_3Depth, AddWeighted,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8U, CV_16U, CV_32F, CV_64F),
Values(CV_8U, CV_16U, CV_32F, CV_64F),
}
//////////////////////////////////////////////////////////////////////
-// GEMM
-
-CV_FLAGS(GemmFlags, 0, GEMM_1_T, GEMM_2_T, GEMM_3_T)
-#define ALL_GEMM_FLAGS Values(0, (int)cv::GEMM_1_T, (int)cv::GEMM_2_T, (int)cv::GEMM_3_T, \
- (int)cv::GEMM_1_T | cv::GEMM_2_T, (int)cv::GEMM_1_T | cv::GEMM_3_T, \
- (int)cv::GEMM_1_T | cv::GEMM_2_T | cv::GEMM_3_T)
-
-DEF_PARAM_TEST(Sz_Type_Flags, cv::Size, MatType, GemmFlags);
-
-PERF_TEST_P(Sz_Type_Flags, Core_GEMM,
- Combine(Values(cv::Size(512, 512), cv::Size(1024, 1024)),
- Values(CV_32FC1, CV_32FC2, CV_64FC1),
- ALL_GEMM_FLAGS))
-{
- const cv::Size size = GET_PARAM(0);
- const int type = GET_PARAM(1);
- const int flags = GET_PARAM(2);
-
- cv::Mat src1(size, type);
- declare.in(src1, WARMUP_RNG);
-
- cv::Mat src2(size, type);
- declare.in(src2, WARMUP_RNG);
-
- cv::Mat src3(size, type);
- declare.in(src3, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- declare.time(5.0);
-
- const cv::gpu::GpuMat d_src1(src1);
- const cv::gpu::GpuMat d_src2(src2);
- const cv::gpu::GpuMat d_src3(src3);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::gemm(d_src1, d_src2, 1.0, d_src3, 1.0, dst, flags);
-
- GPU_SANITY_CHECK(dst, 1e-6);
- }
- else
- {
- declare.time(50.0);
-
- cv::Mat dst;
-
- TEST_CYCLE() cv::gemm(src1, src2, 1.0, src3, 1.0, dst, flags);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// Transpose
-
-PERF_TEST_P(Sz_Type, Core_Transpose,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8UC1, CV_8UC4, CV_16UC2, CV_16SC2, CV_32SC1, CV_32SC2, CV_64FC1)))
-{
- const cv::Size size = GET_PARAM(0);
- const int type = GET_PARAM(1);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::transpose(d_src, dst);
-
- GPU_SANITY_CHECK(dst, 1e-10);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::transpose(src, dst);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// Flip
-
-enum {FLIP_BOTH = 0, FLIP_X = 1, FLIP_Y = -1};
-CV_ENUM(FlipCode, FLIP_BOTH, FLIP_X, FLIP_Y)
-
-DEF_PARAM_TEST(Sz_Depth_Cn_Code, cv::Size, MatDepth, MatCn, FlipCode);
-
-PERF_TEST_P(Sz_Depth_Cn_Code, Core_Flip,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32F),
- GPU_CHANNELS_1_3_4,
- FlipCode::all()))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
- const int flipCode = GET_PARAM(3);
-
- const int type = CV_MAKE_TYPE(depth, channels);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::flip(d_src, dst, flipCode);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::flip(src, dst, flipCode);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// LutOneChannel
-
-PERF_TEST_P(Sz_Type, Core_LutOneChannel,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8UC1, CV_8UC3)))
-{
- const cv::Size size = GET_PARAM(0);
- const int type = GET_PARAM(1);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- cv::Mat lut(1, 256, CV_8UC1);
- declare.in(lut, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::LUT(d_src, lut, dst);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::LUT(src, lut, dst);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// LutMultiChannel
-
-PERF_TEST_P(Sz_Type, Core_LutMultiChannel,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values<MatType>(CV_8UC3)))
-{
- const cv::Size size = GET_PARAM(0);
- const int type = GET_PARAM(1);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- cv::Mat lut(1, 256, CV_MAKE_TYPE(CV_8U, src.channels()));
- declare.in(lut, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::LUT(d_src, lut, dst);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::LUT(src, lut, dst);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
// MagnitudeComplex
-PERF_TEST_P(Sz, Core_MagnitudeComplex,
+PERF_TEST_P(Sz, MagnitudeComplex,
GPU_TYPICAL_MAT_SIZES)
{
const cv::Size size = GetParam();
//////////////////////////////////////////////////////////////////////
// MagnitudeSqrComplex
-PERF_TEST_P(Sz, Core_MagnitudeSqrComplex,
+PERF_TEST_P(Sz, MagnitudeSqrComplex,
GPU_TYPICAL_MAT_SIZES)
{
const cv::Size size = GetParam();
//////////////////////////////////////////////////////////////////////
// Magnitude
-PERF_TEST_P(Sz, Core_Magnitude,
+PERF_TEST_P(Sz, Magnitude,
GPU_TYPICAL_MAT_SIZES)
{
const cv::Size size = GetParam();
//////////////////////////////////////////////////////////////////////
// MagnitudeSqr
-PERF_TEST_P(Sz, Core_MagnitudeSqr,
+PERF_TEST_P(Sz, MagnitudeSqr,
GPU_TYPICAL_MAT_SIZES)
{
const cv::Size size = GetParam();
DEF_PARAM_TEST(Sz_AngleInDegrees, cv::Size, bool);
-PERF_TEST_P(Sz_AngleInDegrees, Core_Phase,
+PERF_TEST_P(Sz_AngleInDegrees, Phase,
Combine(GPU_TYPICAL_MAT_SIZES,
Bool()))
{
//////////////////////////////////////////////////////////////////////
// CartToPolar
-PERF_TEST_P(Sz_AngleInDegrees, Core_CartToPolar,
+PERF_TEST_P(Sz_AngleInDegrees, CartToPolar,
Combine(GPU_TYPICAL_MAT_SIZES,
Bool()))
{
//////////////////////////////////////////////////////////////////////
// PolarToCart
-PERF_TEST_P(Sz_AngleInDegrees, Core_PolarToCart,
+PERF_TEST_P(Sz_AngleInDegrees, PolarToCart,
Combine(GPU_TYPICAL_MAT_SIZES,
Bool()))
{
}
//////////////////////////////////////////////////////////////////////
-// MeanStdDev
+// Threshold
-PERF_TEST_P(Sz, Core_MeanStdDev,
- GPU_TYPICAL_MAT_SIZES)
-{
- const cv::Size size = GetParam();
-
- cv::Mat src(size, CV_8UC1);
- declare.in(src, WARMUP_RNG);
-
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat d_buf;
- cv::Scalar gpu_mean;
- cv::Scalar gpu_stddev;
-
- TEST_CYCLE() cv::gpu::meanStdDev(d_src, gpu_mean, gpu_stddev, d_buf);
-
- SANITY_CHECK(gpu_mean);
- SANITY_CHECK(gpu_stddev);
- }
- else
- {
- cv::Scalar cpu_mean;
- cv::Scalar cpu_stddev;
+CV_ENUM(ThreshOp, cv::THRESH_BINARY, cv::THRESH_BINARY_INV, cv::THRESH_TRUNC, cv::THRESH_TOZERO, cv::THRESH_TOZERO_INV)
- TEST_CYCLE() cv::meanStdDev(src, cpu_mean, cpu_stddev);
+DEF_PARAM_TEST(Sz_Depth_Op, cv::Size, MatDepth, ThreshOp);
- SANITY_CHECK(cpu_mean);
- SANITY_CHECK(cpu_stddev);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// Norm
-
-DEF_PARAM_TEST(Sz_Depth_Norm, cv::Size, MatDepth, NormType);
-
-PERF_TEST_P(Sz_Depth_Norm, Core_Norm,
+PERF_TEST_P(Sz_Depth_Op, Threshold,
Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32S, CV_32F),
- Values(NormType(cv::NORM_INF), NormType(cv::NORM_L1), NormType(cv::NORM_L2))))
+ Values(CV_8U, CV_16U, CV_32F, CV_64F),
+ ThreshOp::all()))
{
const cv::Size size = GET_PARAM(0);
const int depth = GET_PARAM(1);
- const int normType = GET_PARAM(2);
+ const int threshOp = GET_PARAM(2);
cv::Mat src(size, depth);
- if (depth == CV_8U)
- cv::randu(src, 0, 254);
- else
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat d_buf;
- double gpu_dst;
-
- TEST_CYCLE() gpu_dst = cv::gpu::norm(d_src, normType, d_buf);
-
- SANITY_CHECK(gpu_dst, 1e-6, ERROR_RELATIVE);
- }
- else
- {
- double cpu_dst;
-
- TEST_CYCLE() cpu_dst = cv::norm(src, normType);
-
- SANITY_CHECK(cpu_dst, 1e-6, ERROR_RELATIVE);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// NormDiff
-
-DEF_PARAM_TEST(Sz_Norm, cv::Size, NormType);
-
-PERF_TEST_P(Sz_Norm, Core_NormDiff,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(NormType(cv::NORM_INF), NormType(cv::NORM_L1), NormType(cv::NORM_L2))))
-{
- const cv::Size size = GET_PARAM(0);
- const int normType = GET_PARAM(1);
-
- cv::Mat src1(size, CV_8UC1);
- declare.in(src1, WARMUP_RNG);
-
- cv::Mat src2(size, CV_8UC1);
- declare.in(src2, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src1(src1);
- const cv::gpu::GpuMat d_src2(src2);
- double gpu_dst;
-
- TEST_CYCLE() gpu_dst = cv::gpu::norm(d_src1, d_src2, normType);
-
- SANITY_CHECK(gpu_dst);
-
- }
- else
- {
- double cpu_dst;
-
- TEST_CYCLE() cpu_dst = cv::norm(src1, src2, normType);
-
- SANITY_CHECK(cpu_dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// Sum
-
-PERF_TEST_P(Sz_Depth_Cn, Core_Sum,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32F),
- GPU_CHANNELS_1_3_4))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
-
- const int type = CV_MAKE_TYPE(depth, channels);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat d_buf;
- cv::Scalar gpu_dst;
-
- TEST_CYCLE() gpu_dst = cv::gpu::sum(d_src, d_buf);
-
- SANITY_CHECK(gpu_dst, 1e-5, ERROR_RELATIVE);
- }
- else
- {
- cv::Scalar cpu_dst;
-
- TEST_CYCLE() cpu_dst = cv::sum(src);
-
- SANITY_CHECK(cpu_dst, 1e-6, ERROR_RELATIVE);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// SumAbs
-
-PERF_TEST_P(Sz_Depth_Cn, Core_SumAbs,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32F),
- GPU_CHANNELS_1_3_4))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
-
- const int type = CV_MAKE_TYPE(depth, channels);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat d_buf;
- cv::Scalar gpu_dst;
-
- TEST_CYCLE() gpu_dst = cv::gpu::absSum(d_src, d_buf);
-
- SANITY_CHECK(gpu_dst, 1e-6, ERROR_RELATIVE);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// SumSqr
-
-PERF_TEST_P(Sz_Depth_Cn, Core_SumSqr,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values<MatDepth>(CV_8U, CV_16U, CV_32F),
- GPU_CHANNELS_1_3_4))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
-
- const int type = CV_MAKE_TYPE(depth, channels);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat d_buf;
- cv::Scalar gpu_dst;
-
- TEST_CYCLE() gpu_dst = cv::gpu::sqrSum(d_src, d_buf);
-
- SANITY_CHECK(gpu_dst, 1e-6, ERROR_RELATIVE);
- }
- else
- {
- FAIL_NO_CPU();
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// MinMax
-
-PERF_TEST_P(Sz_Depth, Core_MinMax,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32F, CV_64F)))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
-
- cv::Mat src(size, depth);
- if (depth == CV_8U)
- cv::randu(src, 0, 254);
- else
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat d_buf;
- double gpu_minVal, gpu_maxVal;
-
- TEST_CYCLE() cv::gpu::minMax(d_src, &gpu_minVal, &gpu_maxVal, cv::gpu::GpuMat(), d_buf);
-
- SANITY_CHECK(gpu_minVal, 1e-10);
- SANITY_CHECK(gpu_maxVal, 1e-10);
- }
- else
- {
- double cpu_minVal, cpu_maxVal;
-
- TEST_CYCLE() cv::minMaxLoc(src, &cpu_minVal, &cpu_maxVal);
-
- SANITY_CHECK(cpu_minVal);
- SANITY_CHECK(cpu_maxVal);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// MinMaxLoc
-
-PERF_TEST_P(Sz_Depth, Core_MinMaxLoc,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32F, CV_64F)))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
-
- cv::Mat src(size, depth);
- if (depth == CV_8U)
- cv::randu(src, 0, 254);
- else
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat d_valbuf, d_locbuf;
- double gpu_minVal, gpu_maxVal;
- cv::Point gpu_minLoc, gpu_maxLoc;
-
- TEST_CYCLE() cv::gpu::minMaxLoc(d_src, &gpu_minVal, &gpu_maxVal, &gpu_minLoc, &gpu_maxLoc, cv::gpu::GpuMat(), d_valbuf, d_locbuf);
-
- SANITY_CHECK(gpu_minVal, 1e-10);
- SANITY_CHECK(gpu_maxVal, 1e-10);
- }
- else
- {
- double cpu_minVal, cpu_maxVal;
- cv::Point cpu_minLoc, cpu_maxLoc;
-
- TEST_CYCLE() cv::minMaxLoc(src, &cpu_minVal, &cpu_maxVal, &cpu_minLoc, &cpu_maxLoc);
-
- SANITY_CHECK(cpu_minVal);
- SANITY_CHECK(cpu_maxVal);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// CountNonZero
-
-PERF_TEST_P(Sz_Depth, Core_CountNonZero,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32F, CV_64F)))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
-
- cv::Mat src(size, depth);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat d_buf;
- int gpu_dst = 0;
-
- TEST_CYCLE() gpu_dst = cv::gpu::countNonZero(d_src, d_buf);
-
- SANITY_CHECK(gpu_dst);
- }
- else
- {
- int cpu_dst = 0;
-
- TEST_CYCLE() cpu_dst = cv::countNonZero(src);
-
- SANITY_CHECK(cpu_dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
-// Reduce
-
-enum {Rows = 0, Cols = 1};
-CV_ENUM(ReduceCode, REDUCE_SUM, REDUCE_AVG, REDUCE_MAX, REDUCE_MIN)
-CV_ENUM(ReduceDim, Rows, Cols)
-
-DEF_PARAM_TEST(Sz_Depth_Cn_Code_Dim, cv::Size, MatDepth, MatCn, ReduceCode, ReduceDim);
-
-PERF_TEST_P(Sz_Depth_Cn_Code_Dim, Core_Reduce,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_16S, CV_32F),
- Values(1, 2, 3, 4),
- ReduceCode::all(),
- ReduceDim::all()))
-{
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
- const int reduceOp = GET_PARAM(3);
- const int dim = GET_PARAM(4);
-
- const int type = CV_MAKE_TYPE(depth, channels);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::reduce(d_src, dst, dim, reduceOp);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::reduce(src, dst, dim, reduceOp);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-//////////////////////////////////////////////////////////////////////
-// Normalize
-
-DEF_PARAM_TEST(Sz_Depth_NormType, cv::Size, MatDepth, NormType);
-
-PERF_TEST_P(Sz_Depth_NormType, Core_Normalize,
- Combine(GPU_TYPICAL_MAT_SIZES,
- Values(CV_8U, CV_16U, CV_32F, CV_64F),
- Values(NormType(cv::NORM_INF),
- NormType(cv::NORM_L1),
- NormType(cv::NORM_L2),
- NormType(cv::NORM_MINMAX))))
-{
- const cv::Size size = GET_PARAM(0);
- const int type = GET_PARAM(1);
- const int norm_type = GET_PARAM(2);
-
- const double alpha = 1;
- const double beta = 0;
-
- cv::Mat src(size, type);
declare.in(src, WARMUP_RNG);
if (PERF_RUN_GPU())
{
const cv::gpu::GpuMat d_src(src);
cv::gpu::GpuMat dst;
- cv::gpu::GpuMat d_norm_buf, d_cvt_buf;
- TEST_CYCLE() cv::gpu::normalize(d_src, dst, alpha, beta, norm_type, type, cv::gpu::GpuMat(), d_norm_buf, d_cvt_buf);
+ TEST_CYCLE() cv::gpu::threshold(d_src, dst, 100.0, 255.0, threshOp);
- GPU_SANITY_CHECK(dst, 1e-6);
+ GPU_SANITY_CHECK(dst, 1e-10);
}
else
{
cv::Mat dst;
- TEST_CYCLE() cv::normalize(src, dst, alpha, beta, norm_type, type);
+ TEST_CYCLE() cv::threshold(src, dst, 100.0, 255.0, threshOp);
CPU_SANITY_CHECK(dst);
}
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace perf;
+
+CV_PERF_TEST_MAIN(gpuarithm, printCudaInfo())
//
//M*/
-#include "opencv2/core/gpumat.hpp"
+#include "perf_precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifdef __GNUC__
+# pragma GCC diagnostic ignored "-Wmissing-declarations"
+# if defined __clang__ || defined __APPLE__
+# pragma GCC diagnostic ignored "-Wmissing-prototypes"
+# pragma GCC diagnostic ignored "-Wextra"
+# endif
+#endif
+
+#ifndef __OPENCV_PERF_PRECOMP_HPP__
+#define __OPENCV_PERF_PRECOMP_HPP__
+
+#include "opencv2/ts.hpp"
+#include "opencv2/ts/gpu_perf.hpp"
+
+#include "opencv2/gpuarithm.hpp"
+#include "opencv2/core.hpp"
+#include "opencv2/imgproc.hpp"
+
+#ifdef GTEST_CREATE_SHARED_LIBRARY
+#error no modules except ts should have GTEST_CREATE_SHARED_LIBRARY defined
+#endif
+
+#endif
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace std;
+using namespace testing;
+using namespace perf;
+
+//////////////////////////////////////////////////////////////////////
+// Norm
+
+DEF_PARAM_TEST(Sz_Depth_Norm, cv::Size, MatDepth, NormType);
+
+PERF_TEST_P(Sz_Depth_Norm, Norm,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_32S, CV_32F),
+ Values(NormType(cv::NORM_INF), NormType(cv::NORM_L1), NormType(cv::NORM_L2))))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int normType = GET_PARAM(2);
+
+ cv::Mat src(size, depth);
+ if (depth == CV_8U)
+ cv::randu(src, 0, 254);
+ else
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat d_buf;
+ double gpu_dst;
+
+ TEST_CYCLE() gpu_dst = cv::gpu::norm(d_src, normType, d_buf);
+
+ SANITY_CHECK(gpu_dst, 1e-6, ERROR_RELATIVE);
+ }
+ else
+ {
+ double cpu_dst;
+
+ TEST_CYCLE() cpu_dst = cv::norm(src, normType);
+
+ SANITY_CHECK(cpu_dst, 1e-6, ERROR_RELATIVE);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// NormDiff
+
+DEF_PARAM_TEST(Sz_Norm, cv::Size, NormType);
+
+PERF_TEST_P(Sz_Norm, NormDiff,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(NormType(cv::NORM_INF), NormType(cv::NORM_L1), NormType(cv::NORM_L2))))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int normType = GET_PARAM(1);
+
+ cv::Mat src1(size, CV_8UC1);
+ declare.in(src1, WARMUP_RNG);
+
+ cv::Mat src2(size, CV_8UC1);
+ declare.in(src2, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src1(src1);
+ const cv::gpu::GpuMat d_src2(src2);
+ double gpu_dst;
+
+ TEST_CYCLE() gpu_dst = cv::gpu::norm(d_src1, d_src2, normType);
+
+ SANITY_CHECK(gpu_dst);
+
+ }
+ else
+ {
+ double cpu_dst;
+
+ TEST_CYCLE() cpu_dst = cv::norm(src1, src2, normType);
+
+ SANITY_CHECK(cpu_dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// Sum
+
+PERF_TEST_P(Sz_Depth_Cn, Sum,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_32F),
+ GPU_CHANNELS_1_3_4))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+
+ const int type = CV_MAKE_TYPE(depth, channels);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat d_buf;
+ cv::Scalar gpu_dst;
+
+ TEST_CYCLE() gpu_dst = cv::gpu::sum(d_src, d_buf);
+
+ SANITY_CHECK(gpu_dst, 1e-5, ERROR_RELATIVE);
+ }
+ else
+ {
+ cv::Scalar cpu_dst;
+
+ TEST_CYCLE() cpu_dst = cv::sum(src);
+
+ SANITY_CHECK(cpu_dst, 1e-6, ERROR_RELATIVE);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// SumAbs
+
+PERF_TEST_P(Sz_Depth_Cn, SumAbs,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_32F),
+ GPU_CHANNELS_1_3_4))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+
+ const int type = CV_MAKE_TYPE(depth, channels);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat d_buf;
+ cv::Scalar gpu_dst;
+
+ TEST_CYCLE() gpu_dst = cv::gpu::absSum(d_src, d_buf);
+
+ SANITY_CHECK(gpu_dst, 1e-6, ERROR_RELATIVE);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// SumSqr
+
+PERF_TEST_P(Sz_Depth_Cn, SumSqr,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values<MatDepth>(CV_8U, CV_16U, CV_32F),
+ GPU_CHANNELS_1_3_4))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+
+ const int type = CV_MAKE_TYPE(depth, channels);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat d_buf;
+ cv::Scalar gpu_dst;
+
+ TEST_CYCLE() gpu_dst = cv::gpu::sqrSum(d_src, d_buf);
+
+ SANITY_CHECK(gpu_dst, 1e-6, ERROR_RELATIVE);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// MinMax
+
+PERF_TEST_P(Sz_Depth, MinMax,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_32F, CV_64F)))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+
+ cv::Mat src(size, depth);
+ if (depth == CV_8U)
+ cv::randu(src, 0, 254);
+ else
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat d_buf;
+ double gpu_minVal, gpu_maxVal;
+
+ TEST_CYCLE() cv::gpu::minMax(d_src, &gpu_minVal, &gpu_maxVal, cv::gpu::GpuMat(), d_buf);
+
+ SANITY_CHECK(gpu_minVal, 1e-10);
+ SANITY_CHECK(gpu_maxVal, 1e-10);
+ }
+ else
+ {
+ double cpu_minVal, cpu_maxVal;
+
+ TEST_CYCLE() cv::minMaxLoc(src, &cpu_minVal, &cpu_maxVal);
+
+ SANITY_CHECK(cpu_minVal);
+ SANITY_CHECK(cpu_maxVal);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// MinMaxLoc
+
+PERF_TEST_P(Sz_Depth, MinMaxLoc,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_32F, CV_64F)))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+
+ cv::Mat src(size, depth);
+ if (depth == CV_8U)
+ cv::randu(src, 0, 254);
+ else
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat d_valbuf, d_locbuf;
+ double gpu_minVal, gpu_maxVal;
+ cv::Point gpu_minLoc, gpu_maxLoc;
+
+ TEST_CYCLE() cv::gpu::minMaxLoc(d_src, &gpu_minVal, &gpu_maxVal, &gpu_minLoc, &gpu_maxLoc, cv::gpu::GpuMat(), d_valbuf, d_locbuf);
+
+ SANITY_CHECK(gpu_minVal, 1e-10);
+ SANITY_CHECK(gpu_maxVal, 1e-10);
+ }
+ else
+ {
+ double cpu_minVal, cpu_maxVal;
+ cv::Point cpu_minLoc, cpu_maxLoc;
+
+ TEST_CYCLE() cv::minMaxLoc(src, &cpu_minVal, &cpu_maxVal, &cpu_minLoc, &cpu_maxLoc);
+
+ SANITY_CHECK(cpu_minVal);
+ SANITY_CHECK(cpu_maxVal);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// CountNonZero
+
+PERF_TEST_P(Sz_Depth, CountNonZero,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_32F, CV_64F)))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+
+ cv::Mat src(size, depth);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat d_buf;
+ int gpu_dst = 0;
+
+ TEST_CYCLE() gpu_dst = cv::gpu::countNonZero(d_src, d_buf);
+
+ SANITY_CHECK(gpu_dst);
+ }
+ else
+ {
+ int cpu_dst = 0;
+
+ TEST_CYCLE() cpu_dst = cv::countNonZero(src);
+
+ SANITY_CHECK(cpu_dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// Reduce
+
+CV_ENUM(ReduceCode, REDUCE_SUM, REDUCE_AVG, REDUCE_MAX, REDUCE_MIN)
+
+enum {Rows = 0, Cols = 1};
+CV_ENUM(ReduceDim, Rows, Cols)
+
+DEF_PARAM_TEST(Sz_Depth_Cn_Code_Dim, cv::Size, MatDepth, MatCn, ReduceCode, ReduceDim);
+
+PERF_TEST_P(Sz_Depth_Cn_Code_Dim, Reduce,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_16S, CV_32F),
+ Values(1, 2, 3, 4),
+ ReduceCode::all(),
+ ReduceDim::all()))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+ const int reduceOp = GET_PARAM(3);
+ const int dim = GET_PARAM(4);
+
+ const int type = CV_MAKE_TYPE(depth, channels);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::reduce(d_src, dst, dim, reduceOp);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::reduce(src, dst, dim, reduceOp);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// Normalize
+
+DEF_PARAM_TEST(Sz_Depth_NormType, cv::Size, MatDepth, NormType);
+
+PERF_TEST_P(Sz_Depth_NormType, Normalize,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_32F, CV_64F),
+ Values(NormType(cv::NORM_INF),
+ NormType(cv::NORM_L1),
+ NormType(cv::NORM_L2),
+ NormType(cv::NORM_MINMAX))))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int type = GET_PARAM(1);
+ const int norm_type = GET_PARAM(2);
+
+ const double alpha = 1;
+ const double beta = 0;
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+ cv::gpu::GpuMat d_norm_buf, d_cvt_buf;
+
+ TEST_CYCLE() cv::gpu::normalize(d_src, dst, alpha, beta, norm_type, type, cv::gpu::GpuMat(), d_norm_buf, d_cvt_buf);
+
+ GPU_SANITY_CHECK(dst, 1e-6);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::normalize(src, dst, alpha, beta, norm_type, type);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// MeanStdDev
+
+PERF_TEST_P(Sz, MeanStdDev,
+ GPU_TYPICAL_MAT_SIZES)
+{
+ const cv::Size size = GetParam();
+
+ cv::Mat src(size, CV_8UC1);
+ declare.in(src, WARMUP_RNG);
+
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat d_buf;
+ cv::Scalar gpu_mean;
+ cv::Scalar gpu_stddev;
+
+ TEST_CYCLE() cv::gpu::meanStdDev(d_src, gpu_mean, gpu_stddev, d_buf);
+
+ SANITY_CHECK(gpu_mean);
+ SANITY_CHECK(gpu_stddev);
+ }
+ else
+ {
+ cv::Scalar cpu_mean;
+ cv::Scalar cpu_stddev;
+
+ TEST_CYCLE() cv::meanStdDev(src, cpu_mean, cpu_stddev);
+
+ SANITY_CHECK(cpu_mean);
+ SANITY_CHECK(cpu_stddev);
+ }
+}
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "precomp.hpp"
+
+using namespace cv;
+using namespace cv::gpu;
+
+#if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
+
+void cv::gpu::gemm(const GpuMat&, const GpuMat&, double, const GpuMat&, double, GpuMat&, int, Stream&) { throw_no_cuda(); }
+
+void cv::gpu::integral(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+void cv::gpu::integralBuffered(const GpuMat&, GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+
+void cv::gpu::sqrIntegral(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+
+void cv::gpu::mulSpectrums(const GpuMat&, const GpuMat&, GpuMat&, int, bool, Stream&) { throw_no_cuda(); }
+void cv::gpu::mulAndScaleSpectrums(const GpuMat&, const GpuMat&, GpuMat&, int, float, bool, Stream&) { throw_no_cuda(); }
+
+void cv::gpu::dft(const GpuMat&, GpuMat&, Size, int, Stream&) { throw_no_cuda(); }
+
+void cv::gpu::ConvolveBuf::create(Size, Size) { throw_no_cuda(); }
+void cv::gpu::convolve(const GpuMat&, const GpuMat&, GpuMat&, bool) { throw_no_cuda(); }
+void cv::gpu::convolve(const GpuMat&, const GpuMat&, GpuMat&, bool, ConvolveBuf&, Stream&) { throw_no_cuda(); }
+
+#else /* !defined (HAVE_CUDA) */
+
+namespace
+{
+ #define error_entry(entry) { entry, #entry }
+
+ struct ErrorEntry
+ {
+ int code;
+ const char* str;
+ };
+
+ struct ErrorEntryComparer
+ {
+ int code;
+ ErrorEntryComparer(int code_) : code(code_) {}
+ bool operator()(const ErrorEntry& e) const { return e.code == code; }
+ };
+
+ String getErrorString(int code, const ErrorEntry* errors, size_t n)
+ {
+ size_t idx = std::find_if(errors, errors + n, ErrorEntryComparer(code)) - errors;
+
+ const char* msg = (idx != n) ? errors[idx].str : "Unknown error code";
+ String str = cv::format("%s [Code = %d]", msg, code);
+
+ return str;
+ }
+}
+
+#ifdef HAVE_CUBLAS
+ namespace
+ {
+ const ErrorEntry cublas_errors[] =
+ {
+ error_entry( CUBLAS_STATUS_SUCCESS ),
+ error_entry( CUBLAS_STATUS_NOT_INITIALIZED ),
+ error_entry( CUBLAS_STATUS_ALLOC_FAILED ),
+ error_entry( CUBLAS_STATUS_INVALID_VALUE ),
+ error_entry( CUBLAS_STATUS_ARCH_MISMATCH ),
+ error_entry( CUBLAS_STATUS_MAPPING_ERROR ),
+ error_entry( CUBLAS_STATUS_EXECUTION_FAILED ),
+ error_entry( CUBLAS_STATUS_INTERNAL_ERROR )
+ };
+
+ const size_t cublas_error_num = sizeof(cublas_errors) / sizeof(cublas_errors[0]);
+
+ static inline void ___cublasSafeCall(cublasStatus_t err, const char* file, const int line, const char* func)
+ {
+ if (CUBLAS_STATUS_SUCCESS != err)
+ {
+ String msg = getErrorString(err, cublas_errors, cublas_error_num);
+ cv::error(cv::Error::GpuApiCallError, msg, func, file, line);
+ }
+ }
+ }
+
+ #if defined(__GNUC__)
+ #define cublasSafeCall(expr) ___cublasSafeCall(expr, __FILE__, __LINE__, __func__)
+ #else /* defined(__CUDACC__) || defined(__MSVC__) */
+ #define cublasSafeCall(expr) ___cublasSafeCall(expr, __FILE__, __LINE__, "")
+ #endif
+#endif // HAVE_CUBLAS
+
+#ifdef HAVE_CUFFT
+ namespace
+ {
+ //////////////////////////////////////////////////////////////////////////
+ // CUFFT errors
+
+ const ErrorEntry cufft_errors[] =
+ {
+ error_entry( CUFFT_INVALID_PLAN ),
+ error_entry( CUFFT_ALLOC_FAILED ),
+ error_entry( CUFFT_INVALID_TYPE ),
+ error_entry( CUFFT_INVALID_VALUE ),
+ error_entry( CUFFT_INTERNAL_ERROR ),
+ error_entry( CUFFT_EXEC_FAILED ),
+ error_entry( CUFFT_SETUP_FAILED ),
+ error_entry( CUFFT_INVALID_SIZE ),
+ error_entry( CUFFT_UNALIGNED_DATA )
+ };
+
+ const int cufft_error_num = sizeof(cufft_errors) / sizeof(cufft_errors[0]);
+
+ void ___cufftSafeCall(int err, const char* file, const int line, const char* func)
+ {
+ if (CUFFT_SUCCESS != err)
+ {
+ String msg = getErrorString(err, cufft_errors, cufft_error_num);
+ cv::error(cv::Error::GpuApiCallError, msg, func, file, line);
+ }
+ }
+ }
+
+ #if defined(__GNUC__)
+ #define cufftSafeCall(expr) ___cufftSafeCall(expr, __FILE__, __LINE__, __func__)
+ #else /* defined(__CUDACC__) || defined(__MSVC__) */
+ #define cufftSafeCall(expr) ___cufftSafeCall(expr, __FILE__, __LINE__, "")
+ #endif
+
+#endif
+
+////////////////////////////////////////////////////////////////////////
+// gemm
+
+void cv::gpu::gemm(const GpuMat& src1, const GpuMat& src2, double alpha, const GpuMat& src3, double beta, GpuMat& dst, int flags, Stream& stream)
+{
+#ifndef HAVE_CUBLAS
+ (void)src1;
+ (void)src2;
+ (void)alpha;
+ (void)src3;
+ (void)beta;
+ (void)dst;
+ (void)flags;
+ (void)stream;
+ CV_Error(cv::Error::StsNotImplemented, "The library was build without CUBLAS");
+#else
+ // CUBLAS works with column-major matrices
+
+ CV_Assert(src1.type() == CV_32FC1 || src1.type() == CV_32FC2 || src1.type() == CV_64FC1 || src1.type() == CV_64FC2);
+ CV_Assert(src2.type() == src1.type() && (src3.empty() || src3.type() == src1.type()));
+
+ if (src1.depth() == CV_64F)
+ {
+ if (!deviceSupports(NATIVE_DOUBLE))
+ CV_Error(cv::Error::StsUnsupportedFormat, "The device doesn't support double");
+ }
+
+ bool tr1 = (flags & GEMM_1_T) != 0;
+ bool tr2 = (flags & GEMM_2_T) != 0;
+ bool tr3 = (flags & GEMM_3_T) != 0;
+
+ if (src1.type() == CV_64FC2)
+ {
+ if (tr1 || tr2 || tr3)
+ CV_Error(cv::Error::StsNotImplemented, "transpose operation doesn't implemented for CV_64FC2 type");
+ }
+
+ Size src1Size = tr1 ? Size(src1.rows, src1.cols) : src1.size();
+ Size src2Size = tr2 ? Size(src2.rows, src2.cols) : src2.size();
+ Size src3Size = tr3 ? Size(src3.rows, src3.cols) : src3.size();
+ Size dstSize(src2Size.width, src1Size.height);
+
+ CV_Assert(src1Size.width == src2Size.height);
+ CV_Assert(src3.empty() || src3Size == dstSize);
+
+ dst.create(dstSize, src1.type());
+
+ if (beta != 0)
+ {
+ if (src3.empty())
+ {
+ if (stream)
+ stream.enqueueMemSet(dst, Scalar::all(0));
+ else
+ dst.setTo(Scalar::all(0));
+ }
+ else
+ {
+ if (tr3)
+ {
+ gpu::transpose(src3, dst, stream);
+ }
+ else
+ {
+ if (stream)
+ stream.enqueueCopy(src3, dst);
+ else
+ src3.copyTo(dst);
+ }
+ }
+ }
+
+ cublasHandle_t handle;
+ cublasSafeCall( cublasCreate_v2(&handle) );
+
+ cublasSafeCall( cublasSetStream_v2(handle, StreamAccessor::getStream(stream)) );
+
+ cublasSafeCall( cublasSetPointerMode_v2(handle, CUBLAS_POINTER_MODE_HOST) );
+
+ const float alphaf = static_cast<float>(alpha);
+ const float betaf = static_cast<float>(beta);
+
+ const cuComplex alphacf = make_cuComplex(alphaf, 0);
+ const cuComplex betacf = make_cuComplex(betaf, 0);
+
+ const cuDoubleComplex alphac = make_cuDoubleComplex(alpha, 0);
+ const cuDoubleComplex betac = make_cuDoubleComplex(beta, 0);
+
+ cublasOperation_t transa = tr2 ? CUBLAS_OP_T : CUBLAS_OP_N;
+ cublasOperation_t transb = tr1 ? CUBLAS_OP_T : CUBLAS_OP_N;
+
+ switch (src1.type())
+ {
+ case CV_32FC1:
+ cublasSafeCall( cublasSgemm_v2(handle, transa, transb, tr2 ? src2.rows : src2.cols, tr1 ? src1.cols : src1.rows, tr2 ? src2.cols : src2.rows,
+ &alphaf,
+ src2.ptr<float>(), static_cast<int>(src2.step / sizeof(float)),
+ src1.ptr<float>(), static_cast<int>(src1.step / sizeof(float)),
+ &betaf,
+ dst.ptr<float>(), static_cast<int>(dst.step / sizeof(float))) );
+ break;
+
+ case CV_64FC1:
+ cublasSafeCall( cublasDgemm_v2(handle, transa, transb, tr2 ? src2.rows : src2.cols, tr1 ? src1.cols : src1.rows, tr2 ? src2.cols : src2.rows,
+ &alpha,
+ src2.ptr<double>(), static_cast<int>(src2.step / sizeof(double)),
+ src1.ptr<double>(), static_cast<int>(src1.step / sizeof(double)),
+ &beta,
+ dst.ptr<double>(), static_cast<int>(dst.step / sizeof(double))) );
+ break;
+
+ case CV_32FC2:
+ cublasSafeCall( cublasCgemm_v2(handle, transa, transb, tr2 ? src2.rows : src2.cols, tr1 ? src1.cols : src1.rows, tr2 ? src2.cols : src2.rows,
+ &alphacf,
+ src2.ptr<cuComplex>(), static_cast<int>(src2.step / sizeof(cuComplex)),
+ src1.ptr<cuComplex>(), static_cast<int>(src1.step / sizeof(cuComplex)),
+ &betacf,
+ dst.ptr<cuComplex>(), static_cast<int>(dst.step / sizeof(cuComplex))) );
+ break;
+
+ case CV_64FC2:
+ cublasSafeCall( cublasZgemm_v2(handle, transa, transb, tr2 ? src2.rows : src2.cols, tr1 ? src1.cols : src1.rows, tr2 ? src2.cols : src2.rows,
+ &alphac,
+ src2.ptr<cuDoubleComplex>(), static_cast<int>(src2.step / sizeof(cuDoubleComplex)),
+ src1.ptr<cuDoubleComplex>(), static_cast<int>(src1.step / sizeof(cuDoubleComplex)),
+ &betac,
+ dst.ptr<cuDoubleComplex>(), static_cast<int>(dst.step / sizeof(cuDoubleComplex))) );
+ break;
+ }
+
+ cublasSafeCall( cublasDestroy_v2(handle) );
+#endif
+}
+
+////////////////////////////////////////////////////////////////////////
+// integral
+
+void cv::gpu::integral(const GpuMat& src, GpuMat& sum, Stream& s)
+{
+ GpuMat buffer;
+ gpu::integralBuffered(src, sum, buffer, s);
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ namespace imgproc
+ {
+ void shfl_integral_gpu(const PtrStepSzb& img, PtrStepSz<unsigned int> integral, cudaStream_t stream);
+ }
+}}}
+
+void cv::gpu::integralBuffered(const GpuMat& src, GpuMat& sum, GpuMat& buffer, Stream& s)
+{
+ CV_Assert(src.type() == CV_8UC1);
+
+ cudaStream_t stream = StreamAccessor::getStream(s);
+
+ cv::Size whole;
+ cv::Point offset;
+
+ src.locateROI(whole, offset);
+
+ if (deviceSupports(WARP_SHUFFLE_FUNCTIONS) && src.cols <= 2048
+ && offset.x % 16 == 0 && ((src.cols + 63) / 64) * 64 <= (static_cast<int>(src.step) - offset.x))
+ {
+ ensureSizeIsEnough(((src.rows + 7) / 8) * 8, ((src.cols + 63) / 64) * 64, CV_32SC1, buffer);
+
+ cv::gpu::cudev::imgproc::shfl_integral_gpu(src, buffer, stream);
+
+ sum.create(src.rows + 1, src.cols + 1, CV_32SC1);
+ if (s)
+ s.enqueueMemSet(sum, Scalar::all(0));
+ else
+ sum.setTo(Scalar::all(0));
+
+ GpuMat inner = sum(Rect(1, 1, src.cols, src.rows));
+ GpuMat res = buffer(Rect(0, 0, src.cols, src.rows));
+
+ if (s)
+ s.enqueueCopy(res, inner);
+ else
+ res.copyTo(inner);
+ }
+ else
+ {
+#ifndef HAVE_OPENCV_GPULEGACY
+ throw_no_cuda();
+#else
+ sum.create(src.rows + 1, src.cols + 1, CV_32SC1);
+
+ NcvSize32u roiSize;
+ roiSize.width = src.cols;
+ roiSize.height = src.rows;
+
+ cudaDeviceProp prop;
+ cudaSafeCall( cudaGetDeviceProperties(&prop, cv::gpu::getDevice()) );
+
+ Ncv32u bufSize;
+ ncvSafeCall( nppiStIntegralGetSize_8u32u(roiSize, &bufSize, prop) );
+ ensureSizeIsEnough(1, bufSize, CV_8UC1, buffer);
+
+ NppStStreamHandler h(stream);
+
+ ncvSafeCall( nppiStIntegral_8u32u_C1R(const_cast<Ncv8u*>(src.ptr<Ncv8u>()), static_cast<int>(src.step),
+ sum.ptr<Ncv32u>(), static_cast<int>(sum.step), roiSize, buffer.ptr<Ncv8u>(), bufSize, prop) );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+#endif
+ }
+}
+
+//////////////////////////////////////////////////////////////////////////////
+// sqrIntegral
+
+void cv::gpu::sqrIntegral(const GpuMat& src, GpuMat& sqsum, Stream& s)
+{
+#ifndef HAVE_OPENCV_GPULEGACY
+ (void) src;
+ (void) sqsum;
+ (void) s;
+ throw_no_cuda();
+#else
+ CV_Assert(src.type() == CV_8U);
+
+ NcvSize32u roiSize;
+ roiSize.width = src.cols;
+ roiSize.height = src.rows;
+
+ cudaDeviceProp prop;
+ cudaSafeCall( cudaGetDeviceProperties(&prop, cv::gpu::getDevice()) );
+
+ Ncv32u bufSize;
+ ncvSafeCall(nppiStSqrIntegralGetSize_8u64u(roiSize, &bufSize, prop));
+ GpuMat buf(1, bufSize, CV_8U);
+
+ cudaStream_t stream = StreamAccessor::getStream(s);
+
+ NppStStreamHandler h(stream);
+
+ sqsum.create(src.rows + 1, src.cols + 1, CV_64F);
+ ncvSafeCall(nppiStSqrIntegral_8u64u_C1R(const_cast<Ncv8u*>(src.ptr<Ncv8u>(0)), static_cast<int>(src.step),
+ sqsum.ptr<Ncv64u>(0), static_cast<int>(sqsum.step), roiSize, buf.ptr<Ncv8u>(0), bufSize, prop));
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+#endif
+}
+
+//////////////////////////////////////////////////////////////////////////////
+// mulSpectrums
+
+#ifdef HAVE_CUFFT
+
+namespace cv { namespace gpu { namespace cudev
+{
+ void mulSpectrums(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, PtrStepSz<cufftComplex> c, cudaStream_t stream);
+
+ void mulSpectrums_CONJ(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, PtrStepSz<cufftComplex> c, cudaStream_t stream);
+}}}
+
+#endif
+
+void cv::gpu::mulSpectrums(const GpuMat& a, const GpuMat& b, GpuMat& c, int flags, bool conjB, Stream& stream)
+{
+#ifndef HAVE_CUFFT
+ (void) a;
+ (void) b;
+ (void) c;
+ (void) flags;
+ (void) conjB;
+ (void) stream;
+ throw_no_cuda();
+#else
+ (void) flags;
+
+ typedef void (*Caller)(const PtrStep<cufftComplex>, const PtrStep<cufftComplex>, PtrStepSz<cufftComplex>, cudaStream_t stream);
+
+ static Caller callers[] = { cudev::mulSpectrums, cudev::mulSpectrums_CONJ };
+
+ CV_Assert(a.type() == b.type() && a.type() == CV_32FC2);
+ CV_Assert(a.size() == b.size());
+
+ c.create(a.size(), CV_32FC2);
+
+ Caller caller = callers[(int)conjB];
+ caller(a, b, c, StreamAccessor::getStream(stream));
+#endif
+}
+
+//////////////////////////////////////////////////////////////////////////////
+// mulAndScaleSpectrums
+
+#ifdef HAVE_CUFFT
+
+namespace cv { namespace gpu { namespace cudev
+{
+ void mulAndScaleSpectrums(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, float scale, PtrStepSz<cufftComplex> c, cudaStream_t stream);
+
+ void mulAndScaleSpectrums_CONJ(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, float scale, PtrStepSz<cufftComplex> c, cudaStream_t stream);
+}}}
+
+#endif
+
+void cv::gpu::mulAndScaleSpectrums(const GpuMat& a, const GpuMat& b, GpuMat& c, int flags, float scale, bool conjB, Stream& stream)
+{
+#ifndef HAVE_CUFFT
+ (void) a;
+ (void) b;
+ (void) c;
+ (void) flags;
+ (void) scale;
+ (void) conjB;
+ (void) stream;
+ throw_no_cuda();
+#else
+ (void)flags;
+
+ typedef void (*Caller)(const PtrStep<cufftComplex>, const PtrStep<cufftComplex>, float scale, PtrStepSz<cufftComplex>, cudaStream_t stream);
+ static Caller callers[] = { cudev::mulAndScaleSpectrums, cudev::mulAndScaleSpectrums_CONJ };
+
+ CV_Assert(a.type() == b.type() && a.type() == CV_32FC2);
+ CV_Assert(a.size() == b.size());
+
+ c.create(a.size(), CV_32FC2);
+
+ Caller caller = callers[(int)conjB];
+ caller(a, b, scale, c, StreamAccessor::getStream(stream));
+#endif
+}
+
+//////////////////////////////////////////////////////////////////////////////
+// dft
+
+void cv::gpu::dft(const GpuMat& src, GpuMat& dst, Size dft_size, int flags, Stream& stream)
+{
+#ifndef HAVE_CUFFT
+ (void) src;
+ (void) dst;
+ (void) dft_size;
+ (void) flags;
+ (void) stream;
+ throw_no_cuda();
+#else
+
+ CV_Assert(src.type() == CV_32F || src.type() == CV_32FC2);
+
+ // We don't support unpacked output (in the case of real input)
+ CV_Assert(!(flags & DFT_COMPLEX_OUTPUT));
+
+ bool is_1d_input = (dft_size.height == 1) || (dft_size.width == 1);
+ int is_row_dft = flags & DFT_ROWS;
+ int is_scaled_dft = flags & DFT_SCALE;
+ int is_inverse = flags & DFT_INVERSE;
+ bool is_complex_input = src.channels() == 2;
+ bool is_complex_output = !(flags & DFT_REAL_OUTPUT);
+
+ // We don't support real-to-real transform
+ CV_Assert(is_complex_input || is_complex_output);
+
+ GpuMat src_data;
+
+ // Make sure here we work with the continuous input,
+ // as CUFFT can't handle gaps
+ src_data = src;
+ createContinuous(src.rows, src.cols, src.type(), src_data);
+ if (src_data.data != src.data)
+ src.copyTo(src_data);
+
+ Size dft_size_opt = dft_size;
+ if (is_1d_input && !is_row_dft)
+ {
+ // If the source matrix is single column handle it as single row
+ dft_size_opt.width = std::max(dft_size.width, dft_size.height);
+ dft_size_opt.height = std::min(dft_size.width, dft_size.height);
+ }
+
+ cufftType dft_type = CUFFT_R2C;
+ if (is_complex_input)
+ dft_type = is_complex_output ? CUFFT_C2C : CUFFT_C2R;
+
+ CV_Assert(dft_size_opt.width > 1);
+
+ cufftHandle plan;
+ if (is_1d_input || is_row_dft)
+ cufftPlan1d(&plan, dft_size_opt.width, dft_type, dft_size_opt.height);
+ else
+ cufftPlan2d(&plan, dft_size_opt.height, dft_size_opt.width, dft_type);
+
+ cufftSafeCall( cufftSetStream(plan, StreamAccessor::getStream(stream)) );
+
+ if (is_complex_input)
+ {
+ if (is_complex_output)
+ {
+ createContinuous(dft_size, CV_32FC2, dst);
+ cufftSafeCall(cufftExecC2C(
+ plan, src_data.ptr<cufftComplex>(), dst.ptr<cufftComplex>(),
+ is_inverse ? CUFFT_INVERSE : CUFFT_FORWARD));
+ }
+ else
+ {
+ createContinuous(dft_size, CV_32F, dst);
+ cufftSafeCall(cufftExecC2R(
+ plan, src_data.ptr<cufftComplex>(), dst.ptr<cufftReal>()));
+ }
+ }
+ else
+ {
+ // We could swap dft_size for efficiency. Here we must reflect it
+ if (dft_size == dft_size_opt)
+ createContinuous(Size(dft_size.width / 2 + 1, dft_size.height), CV_32FC2, dst);
+ else
+ createContinuous(Size(dft_size.width, dft_size.height / 2 + 1), CV_32FC2, dst);
+
+ cufftSafeCall(cufftExecR2C(
+ plan, src_data.ptr<cufftReal>(), dst.ptr<cufftComplex>()));
+ }
+
+ cufftSafeCall(cufftDestroy(plan));
+
+ if (is_scaled_dft)
+ multiply(dst, Scalar::all(1. / dft_size.area()), dst, 1, -1, stream);
+
+#endif
+}
+
+//////////////////////////////////////////////////////////////////////////////
+// convolve
+
+void cv::gpu::ConvolveBuf::create(Size image_size, Size templ_size)
+{
+ result_size = Size(image_size.width - templ_size.width + 1,
+ image_size.height - templ_size.height + 1);
+
+ block_size = user_block_size;
+ if (user_block_size.width == 0 || user_block_size.height == 0)
+ block_size = estimateBlockSize(result_size, templ_size);
+
+ dft_size.width = 1 << int(ceil(std::log(block_size.width + templ_size.width - 1.) / std::log(2.)));
+ dft_size.height = 1 << int(ceil(std::log(block_size.height + templ_size.height - 1.) / std::log(2.)));
+
+ // CUFFT has hard-coded kernels for power-of-2 sizes (up to 8192),
+ // see CUDA Toolkit 4.1 CUFFT Library Programming Guide
+ if (dft_size.width > 8192)
+ dft_size.width = getOptimalDFTSize(block_size.width + templ_size.width - 1);
+ if (dft_size.height > 8192)
+ dft_size.height = getOptimalDFTSize(block_size.height + templ_size.height - 1);
+
+ // To avoid wasting time doing small DFTs
+ dft_size.width = std::max(dft_size.width, 512);
+ dft_size.height = std::max(dft_size.height, 512);
+
+ createContinuous(dft_size, CV_32F, image_block);
+ createContinuous(dft_size, CV_32F, templ_block);
+ createContinuous(dft_size, CV_32F, result_data);
+
+ spect_len = dft_size.height * (dft_size.width / 2 + 1);
+ createContinuous(1, spect_len, CV_32FC2, image_spect);
+ createContinuous(1, spect_len, CV_32FC2, templ_spect);
+ createContinuous(1, spect_len, CV_32FC2, result_spect);
+
+ // Use maximum result matrix block size for the estimated DFT block size
+ block_size.width = std::min(dft_size.width - templ_size.width + 1, result_size.width);
+ block_size.height = std::min(dft_size.height - templ_size.height + 1, result_size.height);
+}
+
+
+Size cv::gpu::ConvolveBuf::estimateBlockSize(Size result_size, Size /*templ_size*/)
+{
+ int width = (result_size.width + 2) / 3;
+ int height = (result_size.height + 2) / 3;
+ width = std::min(width, result_size.width);
+ height = std::min(height, result_size.height);
+ return Size(width, height);
+}
+
+
+void cv::gpu::convolve(const GpuMat& image, const GpuMat& templ, GpuMat& result, bool ccorr)
+{
+ ConvolveBuf buf;
+ gpu::convolve(image, templ, result, ccorr, buf);
+}
+
+void cv::gpu::convolve(const GpuMat& image, const GpuMat& templ, GpuMat& result, bool ccorr, ConvolveBuf& buf, Stream& stream)
+{
+#ifndef HAVE_CUFFT
+ (void) image;
+ (void) templ;
+ (void) result;
+ (void) ccorr;
+ (void) buf;
+ (void) stream;
+ throw_no_cuda();
+#else
+ using namespace cv::gpu::cudev::imgproc;
+
+ CV_Assert(image.type() == CV_32F);
+ CV_Assert(templ.type() == CV_32F);
+
+ buf.create(image.size(), templ.size());
+ result.create(buf.result_size, CV_32F);
+
+ Size& block_size = buf.block_size;
+ Size& dft_size = buf.dft_size;
+
+ GpuMat& image_block = buf.image_block;
+ GpuMat& templ_block = buf.templ_block;
+ GpuMat& result_data = buf.result_data;
+
+ GpuMat& image_spect = buf.image_spect;
+ GpuMat& templ_spect = buf.templ_spect;
+ GpuMat& result_spect = buf.result_spect;
+
+ cufftHandle planR2C, planC2R;
+ cufftSafeCall(cufftPlan2d(&planC2R, dft_size.height, dft_size.width, CUFFT_C2R));
+ cufftSafeCall(cufftPlan2d(&planR2C, dft_size.height, dft_size.width, CUFFT_R2C));
+
+ cufftSafeCall( cufftSetStream(planR2C, StreamAccessor::getStream(stream)) );
+ cufftSafeCall( cufftSetStream(planC2R, StreamAccessor::getStream(stream)) );
+
+ GpuMat templ_roi(templ.size(), CV_32F, templ.data, templ.step);
+ gpu::copyMakeBorder(templ_roi, templ_block, 0, templ_block.rows - templ_roi.rows, 0,
+ templ_block.cols - templ_roi.cols, 0, Scalar(), stream);
+
+ cufftSafeCall(cufftExecR2C(planR2C, templ_block.ptr<cufftReal>(),
+ templ_spect.ptr<cufftComplex>()));
+
+ // Process all blocks of the result matrix
+ for (int y = 0; y < result.rows; y += block_size.height)
+ {
+ for (int x = 0; x < result.cols; x += block_size.width)
+ {
+ Size image_roi_size(std::min(x + dft_size.width, image.cols) - x,
+ std::min(y + dft_size.height, image.rows) - y);
+ GpuMat image_roi(image_roi_size, CV_32F, (void*)(image.ptr<float>(y) + x),
+ image.step);
+ gpu::copyMakeBorder(image_roi, image_block, 0, image_block.rows - image_roi.rows,
+ 0, image_block.cols - image_roi.cols, 0, Scalar(), stream);
+
+ cufftSafeCall(cufftExecR2C(planR2C, image_block.ptr<cufftReal>(),
+ image_spect.ptr<cufftComplex>()));
+ gpu::mulAndScaleSpectrums(image_spect, templ_spect, result_spect, 0,
+ 1.f / dft_size.area(), ccorr, stream);
+ cufftSafeCall(cufftExecC2R(planC2R, result_spect.ptr<cufftComplex>(),
+ result_data.ptr<cufftReal>()));
+
+ Size result_roi_size(std::min(x + block_size.width, result.cols) - x,
+ std::min(y + block_size.height, result.rows) - y);
+ GpuMat result_roi(result_roi_size, result.type(),
+ (void*)(result.ptr<float>(y) + x), result.step);
+ GpuMat result_block(result_roi_size, result_data.type(),
+ result_data.ptr(), result_data.step);
+
+ if (stream)
+ stream.enqueueCopy(result_block, result_roi);
+ else
+ result_block.copyTo(result_roi);
+ }
+ }
+
+ cufftSafeCall(cufftDestroy(planR2C));
+ cufftSafeCall(cufftDestroy(planC2R));
+#endif
+}
+
+#endif /* !defined (HAVE_CUDA) */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "precomp.hpp"
+
+using namespace cv;
+using namespace cv::gpu;
+
+#if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
+
+void cv::gpu::merge(const GpuMat* /*src*/, size_t /*count*/, GpuMat& /*dst*/, Stream& /*stream*/) { throw_no_cuda(); }
+void cv::gpu::merge(const std::vector<GpuMat>& /*src*/, GpuMat& /*dst*/, Stream& /*stream*/) { throw_no_cuda(); }
+
+void cv::gpu::split(const GpuMat& /*src*/, GpuMat* /*dst*/, Stream& /*stream*/) { throw_no_cuda(); }
+void cv::gpu::split(const GpuMat& /*src*/, std::vector<GpuMat>& /*dst*/, Stream& /*stream*/) { throw_no_cuda(); }
+
+void cv::gpu::transpose(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+
+void cv::gpu::flip(const GpuMat&, GpuMat&, int, Stream&) { throw_no_cuda(); }
+
+void cv::gpu::LUT(const GpuMat&, const Mat&, GpuMat&, Stream&) { throw_no_cuda(); }
+
+void cv::gpu::copyMakeBorder(const GpuMat&, GpuMat&, int, int, int, int, int, const Scalar&, Stream&) { throw_no_cuda(); }
+
+#else /* !defined (HAVE_CUDA) */
+
+////////////////////////////////////////////////////////////////////////
+// merge/split
+
+namespace cv { namespace gpu { namespace cudev
+{
+ namespace split_merge
+ {
+ void merge_caller(const PtrStepSzb* src, PtrStepSzb& dst, int total_channels, size_t elem_size, const cudaStream_t& stream);
+ void split_caller(const PtrStepSzb& src, PtrStepSzb* dst, int num_channels, size_t elem_size1, const cudaStream_t& stream);
+ }
+}}}
+
+namespace
+{
+ void merge(const GpuMat* src, size_t n, GpuMat& dst, const cudaStream_t& stream)
+ {
+ using namespace ::cv::gpu::cudev::split_merge;
+
+ CV_Assert(src);
+ CV_Assert(n > 0);
+
+ int depth = src[0].depth();
+ Size size = src[0].size();
+
+ if (depth == CV_64F)
+ {
+ if (!deviceSupports(NATIVE_DOUBLE))
+ CV_Error(cv::Error::StsUnsupportedFormat, "The device doesn't support double");
+ }
+
+ bool single_channel_only = true;
+ int total_channels = 0;
+
+ for (size_t i = 0; i < n; ++i)
+ {
+ CV_Assert(src[i].size() == size);
+ CV_Assert(src[i].depth() == depth);
+ single_channel_only = single_channel_only && src[i].channels() == 1;
+ total_channels += src[i].channels();
+ }
+
+ CV_Assert(single_channel_only);
+ CV_Assert(total_channels <= 4);
+
+ if (total_channels == 1)
+ src[0].copyTo(dst);
+ else
+ {
+ dst.create(size, CV_MAKETYPE(depth, total_channels));
+
+ PtrStepSzb src_as_devmem[4];
+ for(size_t i = 0; i < n; ++i)
+ src_as_devmem[i] = src[i];
+
+ PtrStepSzb dst_as_devmem(dst);
+ merge_caller(src_as_devmem, dst_as_devmem, total_channels, CV_ELEM_SIZE(depth), stream);
+ }
+ }
+
+ void split(const GpuMat& src, GpuMat* dst, const cudaStream_t& stream)
+ {
+ using namespace ::cv::gpu::cudev::split_merge;
+
+ CV_Assert(dst);
+
+ int depth = src.depth();
+ int num_channels = src.channels();
+
+ if (depth == CV_64F)
+ {
+ if (!deviceSupports(NATIVE_DOUBLE))
+ CV_Error(cv::Error::StsUnsupportedFormat, "The device doesn't support double");
+ }
+
+ if (num_channels == 1)
+ {
+ src.copyTo(dst[0]);
+ return;
+ }
+
+ for (int i = 0; i < num_channels; ++i)
+ dst[i].create(src.size(), depth);
+
+ CV_Assert(num_channels <= 4);
+
+ PtrStepSzb dst_as_devmem[4];
+ for (int i = 0; i < num_channels; ++i)
+ dst_as_devmem[i] = dst[i];
+
+ PtrStepSzb src_as_devmem(src);
+ split_caller(src_as_devmem, dst_as_devmem, num_channels, src.elemSize1(), stream);
+ }
+}
+
+void cv::gpu::merge(const GpuMat* src, size_t n, GpuMat& dst, Stream& stream)
+{
+ ::merge(src, n, dst, StreamAccessor::getStream(stream));
+}
+
+
+void cv::gpu::merge(const std::vector<GpuMat>& src, GpuMat& dst, Stream& stream)
+{
+ ::merge(&src[0], src.size(), dst, StreamAccessor::getStream(stream));
+}
+
+void cv::gpu::split(const GpuMat& src, GpuMat* dst, Stream& stream)
+{
+ ::split(src, dst, StreamAccessor::getStream(stream));
+}
+
+void cv::gpu::split(const GpuMat& src, std::vector<GpuMat>& dst, Stream& stream)
+{
+ dst.resize(src.channels());
+ if(src.channels() > 0)
+ ::split(src, &dst[0], StreamAccessor::getStream(stream));
+}
+
+////////////////////////////////////////////////////////////////////////
+// transpose
+
+namespace arithm
+{
+ template <typename T> void transpose(PtrStepSz<T> src, PtrStepSz<T> dst, cudaStream_t stream);
+}
+
+void cv::gpu::transpose(const GpuMat& src, GpuMat& dst, Stream& s)
+{
+ CV_Assert( src.elemSize() == 1 || src.elemSize() == 4 || src.elemSize() == 8 );
+
+ dst.create( src.cols, src.rows, src.type() );
+
+ cudaStream_t stream = StreamAccessor::getStream(s);
+
+ if (src.elemSize() == 1)
+ {
+ NppStreamHandler h(stream);
+
+ NppiSize sz;
+ sz.width = src.cols;
+ sz.height = src.rows;
+
+ nppSafeCall( nppiTranspose_8u_C1R(src.ptr<Npp8u>(), static_cast<int>(src.step),
+ dst.ptr<Npp8u>(), static_cast<int>(dst.step), sz) );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+ else if (src.elemSize() == 4)
+ {
+ arithm::transpose<int>(src, dst, stream);
+ }
+ else // if (src.elemSize() == 8)
+ {
+ if (!deviceSupports(NATIVE_DOUBLE))
+ CV_Error(cv::Error::StsUnsupportedFormat, "The device doesn't support double");
+
+ arithm::transpose<double>(src, dst, stream);
+ }
+}
+
+////////////////////////////////////////////////////////////////////////
+// flip
+
+namespace
+{
+ template<int DEPTH> struct NppTypeTraits;
+ template<> struct NppTypeTraits<CV_8U> { typedef Npp8u npp_t; };
+ template<> struct NppTypeTraits<CV_8S> { typedef Npp8s npp_t; };
+ template<> struct NppTypeTraits<CV_16U> { typedef Npp16u npp_t; };
+ template<> struct NppTypeTraits<CV_16S> { typedef Npp16s npp_t; };
+ template<> struct NppTypeTraits<CV_32S> { typedef Npp32s npp_t; };
+ template<> struct NppTypeTraits<CV_32F> { typedef Npp32f npp_t; };
+ template<> struct NppTypeTraits<CV_64F> { typedef Npp64f npp_t; };
+
+ template <int DEPTH> struct NppMirrorFunc
+ {
+ typedef typename NppTypeTraits<DEPTH>::npp_t npp_t;
+
+ typedef NppStatus (*func_t)(const npp_t* pSrc, int nSrcStep, npp_t* pDst, int nDstStep, NppiSize oROI, NppiAxis flip);
+ };
+
+ template <int DEPTH, typename NppMirrorFunc<DEPTH>::func_t func> struct NppMirror
+ {
+ typedef typename NppMirrorFunc<DEPTH>::npp_t npp_t;
+
+ static void call(const GpuMat& src, GpuMat& dst, int flipCode, cudaStream_t stream)
+ {
+ NppStreamHandler h(stream);
+
+ NppiSize sz;
+ sz.width = src.cols;
+ sz.height = src.rows;
+
+ nppSafeCall( func(src.ptr<npp_t>(), static_cast<int>(src.step),
+ dst.ptr<npp_t>(), static_cast<int>(dst.step), sz,
+ (flipCode == 0 ? NPP_HORIZONTAL_AXIS : (flipCode > 0 ? NPP_VERTICAL_AXIS : NPP_BOTH_AXIS))) );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+ };
+}
+
+void cv::gpu::flip(const GpuMat& src, GpuMat& dst, int flipCode, Stream& stream)
+{
+ typedef void (*func_t)(const GpuMat& src, GpuMat& dst, int flipCode, cudaStream_t stream);
+ static const func_t funcs[6][4] =
+ {
+ {NppMirror<CV_8U, nppiMirror_8u_C1R>::call, 0, NppMirror<CV_8U, nppiMirror_8u_C3R>::call, NppMirror<CV_8U, nppiMirror_8u_C4R>::call},
+ {0,0,0,0},
+ {NppMirror<CV_16U, nppiMirror_16u_C1R>::call, 0, NppMirror<CV_16U, nppiMirror_16u_C3R>::call, NppMirror<CV_16U, nppiMirror_16u_C4R>::call},
+ {0,0,0,0},
+ {NppMirror<CV_32S, nppiMirror_32s_C1R>::call, 0, NppMirror<CV_32S, nppiMirror_32s_C3R>::call, NppMirror<CV_32S, nppiMirror_32s_C4R>::call},
+ {NppMirror<CV_32F, nppiMirror_32f_C1R>::call, 0, NppMirror<CV_32F, nppiMirror_32f_C3R>::call, NppMirror<CV_32F, nppiMirror_32f_C4R>::call}
+ };
+
+ CV_Assert(src.depth() == CV_8U || src.depth() == CV_16U || src.depth() == CV_32S || src.depth() == CV_32F);
+ CV_Assert(src.channels() == 1 || src.channels() == 3 || src.channels() == 4);
+
+ dst.create(src.size(), src.type());
+
+ funcs[src.depth()][src.channels() - 1](src, dst, flipCode, StreamAccessor::getStream(stream));
+}
+
+////////////////////////////////////////////////////////////////////////
+// LUT
+
+void cv::gpu::LUT(const GpuMat& src, const Mat& lut, GpuMat& dst, Stream& s)
+{
+ const int cn = src.channels();
+
+ CV_Assert( src.type() == CV_8UC1 || src.type() == CV_8UC3 );
+ CV_Assert( lut.depth() == CV_8U );
+ CV_Assert( lut.channels() == 1 || lut.channels() == cn );
+ CV_Assert( lut.rows * lut.cols == 256 && lut.isContinuous() );
+
+ dst.create(src.size(), CV_MAKE_TYPE(lut.depth(), cn));
+
+ NppiSize sz;
+ sz.height = src.rows;
+ sz.width = src.cols;
+
+ Mat nppLut;
+ lut.convertTo(nppLut, CV_32S);
+
+ int nValues3[] = {256, 256, 256};
+
+ Npp32s pLevels[256];
+ for (int i = 0; i < 256; ++i)
+ pLevels[i] = i;
+
+ const Npp32s* pLevels3[3];
+
+#if (CUDA_VERSION <= 4020)
+ pLevels3[0] = pLevels3[1] = pLevels3[2] = pLevels;
+#else
+ GpuMat d_pLevels;
+ d_pLevels.upload(Mat(1, 256, CV_32S, pLevels));
+ pLevels3[0] = pLevels3[1] = pLevels3[2] = d_pLevels.ptr<Npp32s>();
+#endif
+
+ cudaStream_t stream = StreamAccessor::getStream(s);
+ NppStreamHandler h(stream);
+
+ if (src.type() == CV_8UC1)
+ {
+#if (CUDA_VERSION <= 4020)
+ nppSafeCall( nppiLUT_Linear_8u_C1R(src.ptr<Npp8u>(), static_cast<int>(src.step),
+ dst.ptr<Npp8u>(), static_cast<int>(dst.step), sz, nppLut.ptr<Npp32s>(), pLevels, 256) );
+#else
+ GpuMat d_nppLut(Mat(1, 256, CV_32S, nppLut.data));
+ nppSafeCall( nppiLUT_Linear_8u_C1R(src.ptr<Npp8u>(), static_cast<int>(src.step),
+ dst.ptr<Npp8u>(), static_cast<int>(dst.step), sz, d_nppLut.ptr<Npp32s>(), d_pLevels.ptr<Npp32s>(), 256) );
+#endif
+ }
+ else
+ {
+ const Npp32s* pValues3[3];
+
+ Mat nppLut3[3];
+ if (nppLut.channels() == 1)
+ {
+#if (CUDA_VERSION <= 4020)
+ pValues3[0] = pValues3[1] = pValues3[2] = nppLut.ptr<Npp32s>();
+#else
+ GpuMat d_nppLut(Mat(1, 256, CV_32S, nppLut.data));
+ pValues3[0] = pValues3[1] = pValues3[2] = d_nppLut.ptr<Npp32s>();
+#endif
+ }
+ else
+ {
+ cv::split(nppLut, nppLut3);
+
+#if (CUDA_VERSION <= 4020)
+ pValues3[0] = nppLut3[0].ptr<Npp32s>();
+ pValues3[1] = nppLut3[1].ptr<Npp32s>();
+ pValues3[2] = nppLut3[2].ptr<Npp32s>();
+#else
+ GpuMat d_nppLut0(Mat(1, 256, CV_32S, nppLut3[0].data));
+ GpuMat d_nppLut1(Mat(1, 256, CV_32S, nppLut3[1].data));
+ GpuMat d_nppLut2(Mat(1, 256, CV_32S, nppLut3[2].data));
+
+ pValues3[0] = d_nppLut0.ptr<Npp32s>();
+ pValues3[1] = d_nppLut1.ptr<Npp32s>();
+ pValues3[2] = d_nppLut2.ptr<Npp32s>();
+#endif
+ }
+
+ nppSafeCall( nppiLUT_Linear_8u_C3R(src.ptr<Npp8u>(), static_cast<int>(src.step),
+ dst.ptr<Npp8u>(), static_cast<int>(dst.step), sz, pValues3, pLevels3, nValues3) );
+ }
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+}
+
+////////////////////////////////////////////////////////////////////////
+// copyMakeBorder
+
+namespace cv { namespace gpu { namespace cudev
+{
+ namespace imgproc
+ {
+ template <typename T, int cn> void copyMakeBorder_gpu(const PtrStepSzb& src, const PtrStepSzb& dst, int top, int left, int borderMode, const T* borderValue, cudaStream_t stream);
+ }
+}}}
+
+namespace
+{
+ template <typename T, int cn> void copyMakeBorder_caller(const PtrStepSzb& src, const PtrStepSzb& dst, int top, int left, int borderType, const Scalar& value, cudaStream_t stream)
+ {
+ using namespace ::cv::gpu::cudev::imgproc;
+
+ Scalar_<T> val(saturate_cast<T>(value[0]), saturate_cast<T>(value[1]), saturate_cast<T>(value[2]), saturate_cast<T>(value[3]));
+
+ copyMakeBorder_gpu<T, cn>(src, dst, top, left, borderType, val.val, stream);
+ }
+}
+
+#if defined __GNUC__ && __GNUC__ > 2 && __GNUC_MINOR__ > 4
+typedef Npp32s __attribute__((__may_alias__)) Npp32s_a;
+#else
+typedef Npp32s Npp32s_a;
+#endif
+
+void cv::gpu::copyMakeBorder(const GpuMat& src, GpuMat& dst, int top, int bottom, int left, int right, int borderType, const Scalar& value, Stream& s)
+{
+ CV_Assert(src.depth() <= CV_32F && src.channels() <= 4);
+ CV_Assert(borderType == BORDER_REFLECT_101 || borderType == BORDER_REPLICATE || borderType == BORDER_CONSTANT || borderType == BORDER_REFLECT || borderType == BORDER_WRAP);
+
+ dst.create(src.rows + top + bottom, src.cols + left + right, src.type());
+
+ cudaStream_t stream = StreamAccessor::getStream(s);
+
+ if (borderType == BORDER_CONSTANT && (src.type() == CV_8UC1 || src.type() == CV_8UC4 || src.type() == CV_32SC1 || src.type() == CV_32FC1))
+ {
+ NppiSize srcsz;
+ srcsz.width = src.cols;
+ srcsz.height = src.rows;
+
+ NppiSize dstsz;
+ dstsz.width = dst.cols;
+ dstsz.height = dst.rows;
+
+ NppStreamHandler h(stream);
+
+ switch (src.type())
+ {
+ case CV_8UC1:
+ {
+ Npp8u nVal = saturate_cast<Npp8u>(value[0]);
+ nppSafeCall( nppiCopyConstBorder_8u_C1R(src.ptr<Npp8u>(), static_cast<int>(src.step), srcsz,
+ dst.ptr<Npp8u>(), static_cast<int>(dst.step), dstsz, top, left, nVal) );
+ break;
+ }
+ case CV_8UC4:
+ {
+ Npp8u nVal[] = {saturate_cast<Npp8u>(value[0]), saturate_cast<Npp8u>(value[1]), saturate_cast<Npp8u>(value[2]), saturate_cast<Npp8u>(value[3])};
+ nppSafeCall( nppiCopyConstBorder_8u_C4R(src.ptr<Npp8u>(), static_cast<int>(src.step), srcsz,
+ dst.ptr<Npp8u>(), static_cast<int>(dst.step), dstsz, top, left, nVal) );
+ break;
+ }
+ case CV_32SC1:
+ {
+ Npp32s nVal = saturate_cast<Npp32s>(value[0]);
+ nppSafeCall( nppiCopyConstBorder_32s_C1R(src.ptr<Npp32s>(), static_cast<int>(src.step), srcsz,
+ dst.ptr<Npp32s>(), static_cast<int>(dst.step), dstsz, top, left, nVal) );
+ break;
+ }
+ case CV_32FC1:
+ {
+ Npp32f val = saturate_cast<Npp32f>(value[0]);
+ Npp32s nVal = *(reinterpret_cast<Npp32s_a*>(&val));
+ nppSafeCall( nppiCopyConstBorder_32s_C1R(src.ptr<Npp32s>(), static_cast<int>(src.step), srcsz,
+ dst.ptr<Npp32s>(), static_cast<int>(dst.step), dstsz, top, left, nVal) );
+ break;
+ }
+ }
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+ else
+ {
+ typedef void (*caller_t)(const PtrStepSzb& src, const PtrStepSzb& dst, int top, int left, int borderType, const Scalar& value, cudaStream_t stream);
+ static const caller_t callers[6][4] =
+ {
+ { copyMakeBorder_caller<uchar, 1> , copyMakeBorder_caller<uchar, 2> , copyMakeBorder_caller<uchar, 3> , copyMakeBorder_caller<uchar, 4>},
+ {0/*copyMakeBorder_caller<schar, 1>*/, 0/*copyMakeBorder_caller<schar, 2>*/ , 0/*copyMakeBorder_caller<schar, 3>*/, 0/*copyMakeBorder_caller<schar, 4>*/},
+ { copyMakeBorder_caller<ushort, 1> , 0/*copyMakeBorder_caller<ushort, 2>*/, copyMakeBorder_caller<ushort, 3> , copyMakeBorder_caller<ushort, 4>},
+ { copyMakeBorder_caller<short, 1> , 0/*copyMakeBorder_caller<short, 2>*/ , copyMakeBorder_caller<short, 3> , copyMakeBorder_caller<short, 4>},
+ {0/*copyMakeBorder_caller<int, 1>*/, 0/*copyMakeBorder_caller<int, 2>*/ , 0/*copyMakeBorder_caller<int, 3>*/, 0/*copyMakeBorder_caller<int , 4>*/},
+ { copyMakeBorder_caller<float, 1> , 0/*copyMakeBorder_caller<float, 2>*/ , copyMakeBorder_caller<float, 3> , copyMakeBorder_caller<float ,4>}
+ };
+
+ caller_t func = callers[src.depth()][src.channels() - 1];
+ CV_Assert(func != 0);
+
+ func(src, dst, top, left, borderType, value, stream);
+ }
+}
+
+#endif /* !defined (HAVE_CUDA) */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/simd_functions.hpp"
+
+#include "arithm_func_traits.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace arithm
+{
+ struct VAbsDiff4 : binary_function<uint, uint, uint>
+ {
+ __device__ __forceinline__ uint operator ()(uint a, uint b) const
+ {
+ return vabsdiff4(a, b);
+ }
+
+ __device__ __forceinline__ VAbsDiff4() {}
+ __device__ __forceinline__ VAbsDiff4(const VAbsDiff4& other) {}
+ };
+
+ struct VAbsDiff2 : binary_function<uint, uint, uint>
+ {
+ __device__ __forceinline__ uint operator ()(uint a, uint b) const
+ {
+ return vabsdiff2(a, b);
+ }
+
+ __device__ __forceinline__ VAbsDiff2() {}
+ __device__ __forceinline__ VAbsDiff2(const VAbsDiff2& other) {}
+ };
+
+ __device__ __forceinline__ int _abs(int a)
+ {
+ return ::abs(a);
+ }
+ __device__ __forceinline__ float _abs(float a)
+ {
+ return ::fabsf(a);
+ }
+ __device__ __forceinline__ double _abs(double a)
+ {
+ return ::fabs(a);
+ }
+
+ template <typename T> struct AbsDiffMat : binary_function<T, T, T>
+ {
+ __device__ __forceinline__ T operator ()(T a, T b) const
+ {
+ return saturate_cast<T>(_abs(a - b));
+ }
+
+ __device__ __forceinline__ AbsDiffMat() {}
+ __device__ __forceinline__ AbsDiffMat(const AbsDiffMat& other) {}
+ };
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <> struct TransformFunctorTraits< arithm::VAbsDiff4 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
+ {
+ };
+
+ template <> struct TransformFunctorTraits< arithm::VAbsDiff2 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
+ {
+ };
+
+ template <typename T> struct TransformFunctorTraits< arithm::AbsDiffMat<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ void absDiffMat_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
+ {
+ cudev::transform(src1, src2, dst, VAbsDiff4(), WithOutMask(), stream);
+ }
+
+ void absDiffMat_v2(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
+ {
+ cudev::transform(src1, src2, dst, VAbsDiff2(), WithOutMask(), stream);
+ }
+
+ template <typename T>
+ void absDiffMat(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream)
+ {
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<T>) dst, AbsDiffMat<T>(), WithOutMask(), stream);
+ }
+
+ template void absDiffMat<uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void absDiffMat<schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void absDiffMat<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void absDiffMat<short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void absDiffMat<int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void absDiffMat<float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void absDiffMat<double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/simd_functions.hpp"
+
+#include "arithm_func_traits.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace arithm
+{
+ template <typename T, typename S> struct AbsDiffScalar : unary_function<T, T>
+ {
+ S val;
+
+ explicit AbsDiffScalar(S val_) : val(val_) {}
+
+ __device__ __forceinline__ T operator ()(T a) const
+ {
+ abs_func<S> f;
+ return saturate_cast<T>(f(a - val));
+ }
+ };
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <typename T, typename S> struct TransformFunctorTraits< arithm::AbsDiffScalar<T, S> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ template <typename T, typename S>
+ void absDiffScalar(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream)
+ {
+ AbsDiffScalar<T, S> op(static_cast<S>(val));
+
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) dst, op, WithOutMask(), stream);
+ }
+
+ template void absDiffScalar<uchar, float>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+ template void absDiffScalar<schar, float>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+ template void absDiffScalar<ushort, float>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+ template void absDiffScalar<short, float>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+ template void absDiffScalar<int, float>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+ template void absDiffScalar<float, float>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+ template void absDiffScalar<double, double>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/simd_functions.hpp"
+
+#include "arithm_func_traits.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace arithm
+{
+ struct VAdd4 : binary_function<uint, uint, uint>
+ {
+ __device__ __forceinline__ uint operator ()(uint a, uint b) const
+ {
+ return vadd4(a, b);
+ }
+
+ __device__ __forceinline__ VAdd4() {}
+ __device__ __forceinline__ VAdd4(const VAdd4& other) {}
+ };
+
+ struct VAdd2 : binary_function<uint, uint, uint>
+ {
+ __device__ __forceinline__ uint operator ()(uint a, uint b) const
+ {
+ return vadd2(a, b);
+ }
+
+ __device__ __forceinline__ VAdd2() {}
+ __device__ __forceinline__ VAdd2(const VAdd2& other) {}
+ };
+
+ template <typename T, typename D> struct AddMat : binary_function<T, T, D>
+ {
+ __device__ __forceinline__ D operator ()(T a, T b) const
+ {
+ return saturate_cast<D>(a + b);
+ }
+
+ __device__ __forceinline__ AddMat() {}
+ __device__ __forceinline__ AddMat(const AddMat& other) {}
+ };
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <> struct TransformFunctorTraits< arithm::VAdd4 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
+ {
+ };
+
+ template <> struct TransformFunctorTraits< arithm::VAdd2 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
+ {
+ };
+
+ template <typename T, typename D> struct TransformFunctorTraits< arithm::AddMat<T, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ void addMat_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
+ {
+ cudev::transform(src1, src2, dst, VAdd4(), WithOutMask(), stream);
+ }
+
+ void addMat_v2(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
+ {
+ cudev::transform(src1, src2, dst, VAdd2(), WithOutMask(), stream);
+ }
+
+ template <typename T, typename D>
+ void addMat(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream)
+ {
+ if (mask.data)
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<D>) dst, AddMat<T, D>(), mask, stream);
+ else
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<D>) dst, AddMat<T, D>(), WithOutMask(), stream);
+ }
+
+ template void addMat<uchar, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<uchar, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<uchar, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<uchar, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<uchar, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<uchar, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<uchar, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ template void addMat<schar, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<schar, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<schar, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<schar, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<schar, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<schar, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<schar, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void addMat<ushort, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addMat<ushort, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<ushort, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<ushort, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<ushort, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<ushort, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<ushort, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void addMat<short, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addMat<short, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<short, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<short, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<short, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<short, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<short, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void addMat<int, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addMat<int, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addMat<int, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addMat<int, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<int, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<int, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<int, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void addMat<float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addMat<float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addMat<float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addMat<float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addMat<float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<float, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void addMat<double, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addMat<double, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addMat<double, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addMat<double, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addMat<double, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addMat<double, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addMat<double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/simd_functions.hpp"
+
+#include "arithm_func_traits.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace arithm
+{
+ template <typename T, typename S, typename D> struct AddScalar : unary_function<T, D>
+ {
+ S val;
+
+ explicit AddScalar(S val_) : val(val_) {}
+
+ __device__ __forceinline__ D operator ()(T a) const
+ {
+ return saturate_cast<D>(a + val);
+ }
+ };
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <typename T, typename S, typename D> struct TransformFunctorTraits< arithm::AddScalar<T, S, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ template <typename T, typename S, typename D>
+ void addScalar(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream)
+ {
+ AddScalar<T, S, D> op(static_cast<S>(val));
+
+ if (mask.data)
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<D>) dst, op, mask, stream);
+ else
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
+ }
+
+ template void addScalar<uchar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<uchar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<uchar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<uchar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<uchar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<uchar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<uchar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ template void addScalar<schar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<schar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<schar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<schar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<schar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<schar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<schar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void addScalar<ushort, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addScalar<ushort, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<ushort, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<ushort, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<ushort, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<ushort, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<ushort, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void addScalar<short, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addScalar<short, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<short, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<short, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<short, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<short, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<short, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void addScalar<int, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addScalar<int, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addScalar<int, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addScalar<int, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<int, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<int, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<int, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void addScalar<float, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addScalar<float, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addScalar<float, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addScalar<float, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addScalar<float, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<float, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<float, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void addScalar<double, double, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addScalar<double, double, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addScalar<double, double, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addScalar<double, double, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addScalar<double, double, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void addScalar<double, double, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void addScalar<double, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+
+#include "arithm_func_traits.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace arithm
+{
+ template <typename T> struct UseDouble_
+ {
+ enum {value = 0};
+ };
+ template <> struct UseDouble_<double>
+ {
+ enum {value = 1};
+ };
+ template <typename T1, typename T2, typename D> struct UseDouble
+ {
+ enum {value = (UseDouble_<T1>::value || UseDouble_<T2>::value || UseDouble_<D>::value)};
+ };
+
+ template <typename T1, typename T2, typename D, bool useDouble> struct AddWeighted_;
+ template <typename T1, typename T2, typename D> struct AddWeighted_<T1, T2, D, false> : binary_function<T1, T2, D>
+ {
+ float alpha;
+ float beta;
+ float gamma;
+
+ AddWeighted_(double alpha_, double beta_, double gamma_) : alpha(static_cast<float>(alpha_)), beta(static_cast<float>(beta_)), gamma(static_cast<float>(gamma_)) {}
+
+ __device__ __forceinline__ D operator ()(T1 a, T2 b) const
+ {
+ return saturate_cast<D>(a * alpha + b * beta + gamma);
+ }
+ };
+ template <typename T1, typename T2, typename D> struct AddWeighted_<T1, T2, D, true> : binary_function<T1, T2, D>
+ {
+ double alpha;
+ double beta;
+ double gamma;
+
+ AddWeighted_(double alpha_, double beta_, double gamma_) : alpha(alpha_), beta(beta_), gamma(gamma_) {}
+
+ __device__ __forceinline__ D operator ()(T1 a, T2 b) const
+ {
+ return saturate_cast<D>(a * alpha + b * beta + gamma);
+ }
+ };
+ template <typename T1, typename T2, typename D> struct AddWeighted : AddWeighted_<T1, T2, D, UseDouble<T1, T2, D>::value>
+ {
+ AddWeighted(double alpha_, double beta_, double gamma_) : AddWeighted_<T1, T2, D, UseDouble<T1, T2, D>::value>(alpha_, beta_, gamma_) {}
+ };
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <typename T1, typename T2, typename D, size_t src1_size, size_t src2_size, size_t dst_size> struct AddWeightedTraits : DefaultTransformFunctorTraits< arithm::AddWeighted<T1, T2, D> >
+ {
+ };
+ template <typename T1, typename T2, typename D, size_t src_size, size_t dst_size> struct AddWeightedTraits<T1, T2, D, src_size, src_size, dst_size> : arithm::ArithmFuncTraits<src_size, dst_size>
+ {
+ };
+
+ template <typename T1, typename T2, typename D> struct TransformFunctorTraits< arithm::AddWeighted<T1, T2, D> > : AddWeightedTraits<T1, T2, D, sizeof(T1), sizeof(T2), sizeof(D)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ template <typename T1, typename T2, typename D>
+ void addWeighted(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream)
+ {
+ AddWeighted<T1, T2, D> op(alpha, beta, gamma);
+
+ cudev::transform((PtrStepSz<T1>) src1, (PtrStepSz<T2>) src2, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
+ }
+
+ template void addWeighted<uchar, uchar, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, uchar, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, uchar, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, uchar, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, uchar, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, uchar, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, uchar, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<uchar, schar, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, schar, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, schar, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, schar, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, schar, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, schar, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, schar, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<uchar, ushort, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, ushort, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, ushort, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, ushort, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, ushort, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, ushort, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, ushort, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<uchar, short, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, short, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, short, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, short, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, short, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, short, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, short, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<uchar, int, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, int, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, int, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, int, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, int, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, int, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, int, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<uchar, float, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, float, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, float, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, float, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, float, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, float, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, float, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<uchar, double, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, double, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, double, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, double, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, double, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, double, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<uchar, double, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+
+
+ template void addWeighted<schar, schar, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, schar, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, schar, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, schar, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, schar, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, schar, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, schar, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<schar, ushort, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, ushort, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, ushort, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, ushort, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, ushort, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, ushort, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, ushort, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<schar, short, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, short, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, short, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, short, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, short, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, short, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, short, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<schar, int, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, int, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, int, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, int, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, int, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, int, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, int, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<schar, float, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, float, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, float, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, float, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, float, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, float, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, float, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<schar, double, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, double, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, double, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, double, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, double, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, double, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<schar, double, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+
+
+ template void addWeighted<ushort, ushort, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, ushort, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, ushort, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, ushort, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, ushort, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, ushort, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, ushort, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<ushort, short, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, short, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, short, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, short, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, short, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, short, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, short, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<ushort, int, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, int, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, int, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, int, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, int, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, int, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, int, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<ushort, float, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, float, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, float, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, float, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, float, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, float, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, float, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<ushort, double, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, double, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, double, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, double, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, double, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, double, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<ushort, double, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+
+
+ template void addWeighted<short, short, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, short, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, short, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, short, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, short, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, short, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, short, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<short, int, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, int, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, int, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, int, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, int, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, int, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, int, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<short, float, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, float, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, float, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, float, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, float, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, float, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, float, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<short, double, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, double, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, double, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, double, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, double, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, double, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<short, double, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+
+
+ template void addWeighted<int, int, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<int, int, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<int, int, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<int, int, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<int, int, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<int, int, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<int, int, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<int, float, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<int, float, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<int, float, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<int, float, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<int, float, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<int, float, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<int, float, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<int, double, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<int, double, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<int, double, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<int, double, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<int, double, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<int, double, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<int, double, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+
+
+ template void addWeighted<float, float, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<float, float, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<float, float, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<float, float, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<float, float, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<float, float, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<float, float, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+ template void addWeighted<float, double, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<float, double, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<float, double, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<float, double, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<float, double, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<float, double, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<float, double, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+
+
+
+ template void addWeighted<double, double, uchar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<double, double, schar>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<double, double, ushort>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<double, double, short>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<double, double, int>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<double, double, float>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+ template void addWeighted<double, double, double>(PtrStepSzb src1, double alpha, PtrStepSzb src2, double beta, double gamma, PtrStepSzb dst, cudaStream_t stream);
+}
+
+#endif /* CUDA_DISABLER */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __ARITHM_FUNC_TRAITS_HPP__
+#define __ARITHM_FUNC_TRAITS_HPP__
+
+#include <cstddef>
+
+namespace arithm
+{
+ template <size_t src_size, size_t dst_size> struct ArithmFuncTraits
+ {
+ enum { simple_block_dim_x = 32 };
+ enum { simple_block_dim_y = 8 };
+
+ enum { smart_block_dim_x = 32 };
+ enum { smart_block_dim_y = 8 };
+ enum { smart_shift = 1 };
+ };
+
+ template <> struct ArithmFuncTraits<1, 1>
+ {
+ enum { simple_block_dim_x = 32 };
+ enum { simple_block_dim_y = 8 };
+
+ enum { smart_block_dim_x = 32 };
+ enum { smart_block_dim_y = 8 };
+ enum { smart_shift = 4 };
+ };
+ template <> struct ArithmFuncTraits<1, 2>
+ {
+ enum { simple_block_dim_x = 32 };
+ enum { simple_block_dim_y = 8 };
+
+ enum { smart_block_dim_x = 32 };
+ enum { smart_block_dim_y = 8 };
+ enum { smart_shift = 4 };
+ };
+ template <> struct ArithmFuncTraits<1, 4>
+ {
+ enum { simple_block_dim_x = 32 };
+ enum { simple_block_dim_y = 8 };
+
+ enum { smart_block_dim_x = 32 };
+ enum { smart_block_dim_y = 8 };
+ enum { smart_shift = 4 };
+ };
+
+ template <> struct ArithmFuncTraits<2, 1>
+ {
+ enum { simple_block_dim_x = 32 };
+ enum { simple_block_dim_y = 8 };
+
+ enum { smart_block_dim_x = 32 };
+ enum { smart_block_dim_y = 8 };
+ enum { smart_shift = 4 };
+ };
+ template <> struct ArithmFuncTraits<2, 2>
+ {
+ enum { simple_block_dim_x = 32 };
+ enum { simple_block_dim_y = 8 };
+
+ enum { smart_block_dim_x = 32 };
+ enum { smart_block_dim_y = 8 };
+ enum { smart_shift = 4 };
+ };
+ template <> struct ArithmFuncTraits<2, 4>
+ {
+ enum { simple_block_dim_x = 32 };
+ enum { simple_block_dim_y = 8 };
+
+ enum { smart_block_dim_x = 32 };
+ enum { smart_block_dim_y = 8 };
+ enum { smart_shift = 4 };
+ };
+
+ template <> struct ArithmFuncTraits<4, 1>
+ {
+ enum { simple_block_dim_x = 32 };
+ enum { simple_block_dim_y = 8 };
+
+ enum { smart_block_dim_x = 32 };
+ enum { smart_block_dim_y = 8 };
+ enum { smart_shift = 4 };
+ };
+ template <> struct ArithmFuncTraits<4, 2>
+ {
+ enum { simple_block_dim_x = 32 };
+ enum { simple_block_dim_y = 8 };
+
+ enum { smart_block_dim_x = 32 };
+ enum { smart_block_dim_y = 8 };
+ enum { smart_shift = 4 };
+ };
+ template <> struct ArithmFuncTraits<4, 4>
+ {
+ enum { simple_block_dim_x = 32 };
+ enum { simple_block_dim_y = 8 };
+
+ enum { smart_block_dim_x = 32 };
+ enum { smart_block_dim_y = 8 };
+ enum { smart_shift = 4 };
+ };
+}
+
+#endif // __ARITHM_FUNC_TRAITS_HPP__
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/simd_functions.hpp"
+
+#include "arithm_func_traits.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <typename T> struct TransformFunctorTraits< bit_not<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+
+ template <typename T> struct TransformFunctorTraits< bit_and<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+
+ template <typename T> struct TransformFunctorTraits< bit_or<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+
+ template <typename T> struct TransformFunctorTraits< bit_xor<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ template <typename T> void bitMatNot(PtrStepSzb src, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream)
+ {
+ if (mask.data)
+ cudev::transform((PtrStepSz<T>) src, (PtrStepSz<T>) dst, bit_not<T>(), mask, stream);
+ else
+ cudev::transform((PtrStepSz<T>) src, (PtrStepSz<T>) dst, bit_not<T>(), WithOutMask(), stream);
+ }
+
+ template <typename T> void bitMatAnd(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream)
+ {
+ if (mask.data)
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<T>) dst, bit_and<T>(), mask, stream);
+ else
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<T>) dst, bit_and<T>(), WithOutMask(), stream);
+ }
+
+ template <typename T> void bitMatOr(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream)
+ {
+ if (mask.data)
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<T>) dst, bit_or<T>(), mask, stream);
+ else
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<T>) dst, bit_or<T>(), WithOutMask(), stream);
+ }
+
+ template <typename T> void bitMatXor(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream)
+ {
+ if (mask.data)
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<T>) dst, bit_xor<T>(), mask, stream);
+ else
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<T>) dst, bit_xor<T>(), WithOutMask(), stream);
+ }
+
+ template void bitMatNot<uchar>(PtrStepSzb src, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void bitMatNot<ushort>(PtrStepSzb src, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void bitMatNot<uint>(PtrStepSzb src, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ template void bitMatAnd<uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void bitMatAnd<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void bitMatAnd<uint>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ template void bitMatOr<uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void bitMatOr<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void bitMatOr<uint>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ template void bitMatXor<uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void bitMatXor<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void bitMatXor<uint>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/simd_functions.hpp"
+
+#include "arithm_func_traits.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <typename T> struct TransformFunctorTraits< binder2nd< bit_and<T> > > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+
+ template <typename T> struct TransformFunctorTraits< binder2nd< bit_or<T> > > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+
+ template <typename T> struct TransformFunctorTraits< binder2nd< bit_xor<T> > > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ template <typename T> void bitScalarAnd(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream)
+ {
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) dst, cv::gpu::cudev::bind2nd(bit_and<T>(), src2), WithOutMask(), stream);
+ }
+
+ template <typename T> void bitScalarOr(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream)
+ {
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) dst, cv::gpu::cudev::bind2nd(bit_or<T>(), src2), WithOutMask(), stream);
+ }
+
+ template <typename T> void bitScalarXor(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream)
+ {
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) dst, cv::gpu::cudev::bind2nd(bit_xor<T>(), src2), WithOutMask(), stream);
+ }
+
+ template void bitScalarAnd<uchar>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
+ template void bitScalarAnd<ushort>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
+ template void bitScalarAnd<int>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
+ template void bitScalarAnd<unsigned int>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
+
+ template void bitScalarOr<uchar>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
+ template void bitScalarOr<ushort>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
+ template void bitScalarOr<int>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
+ template void bitScalarOr<unsigned int>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
+
+ template void bitScalarXor<uchar>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
+ template void bitScalarXor<ushort>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
+ template void bitScalarXor<int>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
+ template void bitScalarXor<unsigned int>(PtrStepSzb src1, uint src2, PtrStepSzb dst, cudaStream_t stream);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/simd_functions.hpp"
+
+#include "arithm_func_traits.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace arithm
+{
+ struct VCmpEq4 : binary_function<uint, uint, uint>
+ {
+ __device__ __forceinline__ uint operator ()(uint a, uint b) const
+ {
+ return vcmpeq4(a, b);
+ }
+
+ __device__ __forceinline__ VCmpEq4() {}
+ __device__ __forceinline__ VCmpEq4(const VCmpEq4& other) {}
+ };
+ struct VCmpNe4 : binary_function<uint, uint, uint>
+ {
+ __device__ __forceinline__ uint operator ()(uint a, uint b) const
+ {
+ return vcmpne4(a, b);
+ }
+
+ __device__ __forceinline__ VCmpNe4() {}
+ __device__ __forceinline__ VCmpNe4(const VCmpNe4& other) {}
+ };
+ struct VCmpLt4 : binary_function<uint, uint, uint>
+ {
+ __device__ __forceinline__ uint operator ()(uint a, uint b) const
+ {
+ return vcmplt4(a, b);
+ }
+
+ __device__ __forceinline__ VCmpLt4() {}
+ __device__ __forceinline__ VCmpLt4(const VCmpLt4& other) {}
+ };
+ struct VCmpLe4 : binary_function<uint, uint, uint>
+ {
+ __device__ __forceinline__ uint operator ()(uint a, uint b) const
+ {
+ return vcmple4(a, b);
+ }
+
+ __device__ __forceinline__ VCmpLe4() {}
+ __device__ __forceinline__ VCmpLe4(const VCmpLe4& other) {}
+ };
+
+ template <class Op, typename T>
+ struct Cmp : binary_function<T, T, uchar>
+ {
+ __device__ __forceinline__ uchar operator()(T a, T b) const
+ {
+ Op op;
+ return -op(a, b);
+ }
+ };
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <> struct TransformFunctorTraits< arithm::VCmpEq4 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
+ {
+ };
+ template <> struct TransformFunctorTraits< arithm::VCmpNe4 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
+ {
+ };
+ template <> struct TransformFunctorTraits< arithm::VCmpLt4 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
+ {
+ };
+ template <> struct TransformFunctorTraits< arithm::VCmpLe4 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
+ {
+ };
+
+ template <class Op, typename T> struct TransformFunctorTraits< arithm::Cmp<Op, T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(uchar)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ void cmpMatEq_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
+ {
+ cudev::transform(src1, src2, dst, VCmpEq4(), WithOutMask(), stream);
+ }
+ void cmpMatNe_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
+ {
+ cudev::transform(src1, src2, dst, VCmpNe4(), WithOutMask(), stream);
+ }
+ void cmpMatLt_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
+ {
+ cudev::transform(src1, src2, dst, VCmpLt4(), WithOutMask(), stream);
+ }
+ void cmpMatLe_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
+ {
+ cudev::transform(src1, src2, dst, VCmpLe4(), WithOutMask(), stream);
+ }
+
+ template <template <typename> class Op, typename T>
+ void cmpMat(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream)
+ {
+ Cmp<Op<T>, T> op;
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, dst, op, WithOutMask(), stream);
+ }
+
+ template <typename T> void cmpMatEq(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream)
+ {
+ cmpMat<equal_to, T>(src1, src2, dst, stream);
+ }
+ template <typename T> void cmpMatNe(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream)
+ {
+ cmpMat<not_equal_to, T>(src1, src2, dst, stream);
+ }
+ template <typename T> void cmpMatLt(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream)
+ {
+ cmpMat<less, T>(src1, src2, dst, stream);
+ }
+ template <typename T> void cmpMatLe(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream)
+ {
+ cmpMat<less_equal, T>(src1, src2, dst, stream);
+ }
+
+ template void cmpMatEq<uchar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatEq<schar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatEq<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatEq<short >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatEq<int >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatEq<float >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatEq<double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+
+ template void cmpMatNe<uchar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatNe<schar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatNe<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatNe<short >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatNe<int >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatNe<float >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatNe<double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+
+ template void cmpMatLt<uchar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatLt<schar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatLt<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatLt<short >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatLt<int >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatLt<float >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatLt<double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+
+ template void cmpMatLe<uchar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatLe<schar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatLe<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatLe<short >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatLe<int >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatLe<float >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void cmpMatLe<double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/simd_functions.hpp"
+#include "opencv2/core/cuda/vec_math.hpp"
+
+#include "arithm_func_traits.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace arithm
+{
+ template <class Op, typename T>
+ struct Cmp : binary_function<T, T, uchar>
+ {
+ __device__ __forceinline__ uchar operator()(T a, T b) const
+ {
+ Op op;
+ return -op(a, b);
+ }
+ };
+
+#define TYPE_VEC(type, cn) typename TypeVec<type, cn>::vec_type
+
+ template <class Op, typename T, int cn> struct CmpScalar;
+ template <class Op, typename T>
+ struct CmpScalar<Op, T, 1> : unary_function<T, uchar>
+ {
+ T val;
+
+ __host__ explicit CmpScalar(T val_) : val(val_) {}
+
+ __device__ __forceinline__ uchar operator()(T src) const
+ {
+ Cmp<Op, T> op;
+ return op(src, val);
+ }
+ };
+ template <class Op, typename T>
+ struct CmpScalar<Op, T, 2> : unary_function<TYPE_VEC(T, 2), TYPE_VEC(uchar, 2)>
+ {
+ TYPE_VEC(T, 2) val;
+
+ __host__ explicit CmpScalar(TYPE_VEC(T, 2) val_) : val(val_) {}
+
+ __device__ __forceinline__ TYPE_VEC(uchar, 2) operator()(const TYPE_VEC(T, 2) & src) const
+ {
+ Cmp<Op, T> op;
+ return VecTraits<TYPE_VEC(uchar, 2)>::make(op(src.x, val.x), op(src.y, val.y));
+ }
+ };
+ template <class Op, typename T>
+ struct CmpScalar<Op, T, 3> : unary_function<TYPE_VEC(T, 3), TYPE_VEC(uchar, 3)>
+ {
+ TYPE_VEC(T, 3) val;
+
+ __host__ explicit CmpScalar(TYPE_VEC(T, 3) val_) : val(val_) {}
+
+ __device__ __forceinline__ TYPE_VEC(uchar, 3) operator()(const TYPE_VEC(T, 3) & src) const
+ {
+ Cmp<Op, T> op;
+ return VecTraits<TYPE_VEC(uchar, 3)>::make(op(src.x, val.x), op(src.y, val.y), op(src.z, val.z));
+ }
+ };
+ template <class Op, typename T>
+ struct CmpScalar<Op, T, 4> : unary_function<TYPE_VEC(T, 4), TYPE_VEC(uchar, 4)>
+ {
+ TYPE_VEC(T, 4) val;
+
+ __host__ explicit CmpScalar(TYPE_VEC(T, 4) val_) : val(val_) {}
+
+ __device__ __forceinline__ TYPE_VEC(uchar, 4) operator()(const TYPE_VEC(T, 4) & src) const
+ {
+ Cmp<Op, T> op;
+ return VecTraits<TYPE_VEC(uchar, 4)>::make(op(src.x, val.x), op(src.y, val.y), op(src.z, val.z), op(src.w, val.w));
+ }
+ };
+
+#undef TYPE_VEC
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <class Op, typename T> struct TransformFunctorTraits< arithm::CmpScalar<Op, T, 1> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(uchar)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ template <template <typename> class Op, typename T, int cn>
+ void cmpScalar(PtrStepSzb src, double val[4], PtrStepSzb dst, cudaStream_t stream)
+ {
+ typedef typename TypeVec<T, cn>::vec_type src_t;
+ typedef typename TypeVec<uchar, cn>::vec_type dst_t;
+
+ T sval[] = {static_cast<T>(val[0]), static_cast<T>(val[1]), static_cast<T>(val[2]), static_cast<T>(val[3])};
+ src_t val1 = VecTraits<src_t>::make(sval);
+
+ CmpScalar<Op<T>, T, cn> op(val1);
+ cudev::transform((PtrStepSz<src_t>) src, (PtrStepSz<dst_t>) dst, op, WithOutMask(), stream);
+ }
+
+ template <typename T> void cmpScalarEq(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream)
+ {
+ typedef void (*func_t)(PtrStepSzb src, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ static const func_t funcs[] =
+ {
+ 0,
+ cmpScalar<equal_to, T, 1>,
+ cmpScalar<equal_to, T, 2>,
+ cmpScalar<equal_to, T, 3>,
+ cmpScalar<equal_to, T, 4>
+ };
+
+ funcs[cn](src, val, dst, stream);
+ }
+ template <typename T> void cmpScalarNe(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream)
+ {
+ typedef void (*func_t)(PtrStepSzb src, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ static const func_t funcs[] =
+ {
+ 0,
+ cmpScalar<not_equal_to, T, 1>,
+ cmpScalar<not_equal_to, T, 2>,
+ cmpScalar<not_equal_to, T, 3>,
+ cmpScalar<not_equal_to, T, 4>
+ };
+
+ funcs[cn](src, val, dst, stream);
+ }
+ template <typename T> void cmpScalarLt(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream)
+ {
+ typedef void (*func_t)(PtrStepSzb src, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ static const func_t funcs[] =
+ {
+ 0,
+ cmpScalar<less, T, 1>,
+ cmpScalar<less, T, 2>,
+ cmpScalar<less, T, 3>,
+ cmpScalar<less, T, 4>
+ };
+
+ funcs[cn](src, val, dst, stream);
+ }
+ template <typename T> void cmpScalarLe(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream)
+ {
+ typedef void (*func_t)(PtrStepSzb src, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ static const func_t funcs[] =
+ {
+ 0,
+ cmpScalar<less_equal, T, 1>,
+ cmpScalar<less_equal, T, 2>,
+ cmpScalar<less_equal, T, 3>,
+ cmpScalar<less_equal, T, 4>
+ };
+
+ funcs[cn](src, val, dst, stream);
+ }
+ template <typename T> void cmpScalarGt(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream)
+ {
+ typedef void (*func_t)(PtrStepSzb src, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ static const func_t funcs[] =
+ {
+ 0,
+ cmpScalar<greater, T, 1>,
+ cmpScalar<greater, T, 2>,
+ cmpScalar<greater, T, 3>,
+ cmpScalar<greater, T, 4>
+ };
+
+ funcs[cn](src, val, dst, stream);
+ }
+ template <typename T> void cmpScalarGe(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream)
+ {
+ typedef void (*func_t)(PtrStepSzb src, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ static const func_t funcs[] =
+ {
+ 0,
+ cmpScalar<greater_equal, T, 1>,
+ cmpScalar<greater_equal, T, 2>,
+ cmpScalar<greater_equal, T, 3>,
+ cmpScalar<greater_equal, T, 4>
+ };
+
+ funcs[cn](src, val, dst, stream);
+ }
+
+ template void cmpScalarEq<uchar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarEq<schar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarEq<ushort>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarEq<short >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarEq<int >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarEq<float >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarEq<double>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+
+ template void cmpScalarNe<uchar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarNe<schar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarNe<ushort>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarNe<short >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarNe<int >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarNe<float >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarNe<double>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+
+ template void cmpScalarLt<uchar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarLt<schar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarLt<ushort>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarLt<short >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarLt<int >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarLt<float >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarLt<double>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+
+ template void cmpScalarLe<uchar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarLe<schar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarLe<ushort>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarLe<short >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarLe<int >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarLe<float >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarLe<double>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+
+ template void cmpScalarGt<uchar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarGt<schar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarGt<ushort>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarGt<short >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarGt<int >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarGt<float >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarGt<double>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+
+ template void cmpScalarGe<uchar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarGe<schar >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarGe<ushort>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarGe<short >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarGe<int >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarGe<float >(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+ template void cmpScalarGe<double>(PtrStepSzb src, int cn, double val[4], PtrStepSzb dst, cudaStream_t stream);
+}
+
+#endif // CUDA_DISABLER
static const caller_t callers[5] =
{
- CopyMakeBorderDispatcher<BrdReflect101, vec_type>::call,
- CopyMakeBorderDispatcher<BrdReplicate, vec_type>::call,
CopyMakeBorderDispatcher<BrdConstant, vec_type>::call,
+ CopyMakeBorderDispatcher<BrdReplicate, vec_type>::call,
CopyMakeBorderDispatcher<BrdReflect, vec_type>::call,
- CopyMakeBorderDispatcher<BrdWrap, vec_type>::call
+ CopyMakeBorderDispatcher<BrdWrap, vec_type>::call,
+ CopyMakeBorderDispatcher<BrdReflect101, vec_type>::call
};
callers[borderMode](PtrStepSz<vec_type>(src), PtrStepSz<vec_type>(dst), top, left, borderValue, stream);
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/vec_traits.hpp"
+#include "opencv2/core/cuda/vec_math.hpp"
+#include "opencv2/core/cuda/reduce.hpp"
+#include "opencv2/core/cuda/emulation.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace countNonZero
+{
+ __device__ unsigned int blocks_finished = 0;
+
+ template <int BLOCK_SIZE, typename T>
+ __global__ void kernel(const PtrStepSz<T> src, unsigned int* count, const int twidth, const int theight)
+ {
+ __shared__ unsigned int scount[BLOCK_SIZE];
+
+ const int x0 = blockIdx.x * blockDim.x * twidth + threadIdx.x;
+ const int y0 = blockIdx.y * blockDim.y * theight + threadIdx.y;
+
+ const int tid = threadIdx.y * blockDim.x + threadIdx.x;
+
+ unsigned int mycount = 0;
+
+ for (int i = 0, y = y0; i < theight && y < src.rows; ++i, y += blockDim.y)
+ {
+ const T* ptr = src.ptr(y);
+
+ for (int j = 0, x = x0; j < twidth && x < src.cols; ++j, x += blockDim.x)
+ {
+ const T srcVal = ptr[x];
+
+ mycount += (srcVal != 0);
+ }
+ }
+
+ cudev::reduce<BLOCK_SIZE>(scount, mycount, tid, plus<unsigned int>());
+
+ #if __CUDA_ARCH__ >= 200
+ if (tid == 0)
+ ::atomicAdd(count, mycount);
+ #else
+ __shared__ bool is_last;
+ const int bid = blockIdx.y * gridDim.x + blockIdx.x;
+
+ if (tid == 0)
+ {
+ count[bid] = mycount;
+
+ __threadfence();
+
+ unsigned int ticket = ::atomicInc(&blocks_finished, gridDim.x * gridDim.y);
+ is_last = (ticket == gridDim.x * gridDim.y - 1);
+ }
+
+ __syncthreads();
+
+ if (is_last)
+ {
+ mycount = tid < gridDim.x * gridDim.y ? count[tid] : 0;
+
+ cudev::reduce<BLOCK_SIZE>(scount, mycount, tid, plus<unsigned int>());
+
+ if (tid == 0)
+ {
+ count[0] = mycount;
+
+ blocks_finished = 0;
+ }
+ }
+ #endif
+ }
+
+ const int threads_x = 32;
+ const int threads_y = 8;
+
+ void getLaunchCfg(int cols, int rows, dim3& block, dim3& grid)
+ {
+ block = dim3(threads_x, threads_y);
+
+ grid = dim3(divUp(cols, block.x * block.y),
+ divUp(rows, block.y * block.x));
+
+ grid.x = ::min(grid.x, block.x);
+ grid.y = ::min(grid.y, block.y);
+ }
+
+ void getBufSize(int cols, int rows, int& bufcols, int& bufrows)
+ {
+ dim3 block, grid;
+ getLaunchCfg(cols, rows, block, grid);
+
+ bufcols = grid.x * grid.y * sizeof(int);
+ bufrows = 1;
+ }
+
+ template <typename T>
+ int run(const PtrStepSzb src, PtrStep<unsigned int> buf)
+ {
+ dim3 block, grid;
+ getLaunchCfg(src.cols, src.rows, block, grid);
+
+ const int twidth = divUp(divUp(src.cols, grid.x), block.x);
+ const int theight = divUp(divUp(src.rows, grid.y), block.y);
+
+ unsigned int* count_buf = buf.ptr(0);
+
+ cudaSafeCall( cudaMemset(count_buf, 0, sizeof(unsigned int)) );
+
+ kernel<threads_x * threads_y><<<grid, block>>>((PtrStepSz<T>) src, count_buf, twidth, theight);
+ cudaSafeCall( cudaGetLastError() );
+
+ cudaSafeCall( cudaDeviceSynchronize() );
+
+ unsigned int count;
+ cudaSafeCall(cudaMemcpy(&count, count_buf, sizeof(unsigned int), cudaMemcpyDeviceToHost));
+
+ return count;
+ }
+
+ template int run<uchar >(const PtrStepSzb src, PtrStep<unsigned int> buf);
+ template int run<schar >(const PtrStepSzb src, PtrStep<unsigned int> buf);
+ template int run<ushort>(const PtrStepSzb src, PtrStep<unsigned int> buf);
+ template int run<short >(const PtrStepSzb src, PtrStep<unsigned int> buf);
+ template int run<int >(const PtrStepSzb src, PtrStep<unsigned int> buf);
+ template int run<float >(const PtrStepSzb src, PtrStep<unsigned int> buf);
+ template int run<double>(const PtrStepSzb src, PtrStep<unsigned int> buf);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/simd_functions.hpp"
+
+#include "arithm_func_traits.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace arithm
+{
+ template <typename T, typename S, typename D> struct DivInv : unary_function<T, D>
+ {
+ S val;
+
+ explicit DivInv(S val_) : val(val_) {}
+
+ __device__ __forceinline__ D operator ()(T a) const
+ {
+ return a != 0 ? saturate_cast<D>(val / a) : 0;
+ }
+ };
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <typename T, typename S, typename D> struct TransformFunctorTraits< arithm::DivInv<T, S, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ template <typename T, typename S, typename D>
+ void divInv(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream)
+ {
+ DivInv<T, S, D> op(static_cast<S>(val));
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
+ }
+
+ template void divInv<uchar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<uchar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<uchar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<uchar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<uchar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<uchar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<uchar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+
+ template void divInv<schar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<schar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<schar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<schar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<schar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<schar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<schar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+
+ //template void divInv<ushort, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divInv<ushort, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<ushort, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<ushort, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<ushort, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<ushort, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<ushort, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+
+ //template void divInv<short, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divInv<short, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<short, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<short, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<short, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<short, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<short, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+
+ //template void divInv<int, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divInv<int, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divInv<int, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divInv<int, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<int, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<int, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<int, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+
+ //template void divInv<float, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divInv<float, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divInv<float, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divInv<float, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divInv<float, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<float, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<float, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+
+ //template void divInv<double, double, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divInv<double, double, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divInv<double, double, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divInv<double, double, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divInv<double, double, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divInv<double, double, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divInv<double, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/simd_functions.hpp"
+
+#include "arithm_func_traits.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace arithm
+{
+ struct Div_8uc4_32f : binary_function<uint, float, uint>
+ {
+ __device__ __forceinline__ uint operator ()(uint a, float b) const
+ {
+ uint res = 0;
+
+ if (b != 0)
+ {
+ b = 1.0f / b;
+ res |= (saturate_cast<uchar>((0xffu & (a )) * b) );
+ res |= (saturate_cast<uchar>((0xffu & (a >> 8)) * b) << 8);
+ res |= (saturate_cast<uchar>((0xffu & (a >> 16)) * b) << 16);
+ res |= (saturate_cast<uchar>((0xffu & (a >> 24)) * b) << 24);
+ }
+
+ return res;
+ }
+ };
+
+ struct Div_16sc4_32f : binary_function<short4, float, short4>
+ {
+ __device__ __forceinline__ short4 operator ()(short4 a, float b) const
+ {
+ return b != 0 ? make_short4(saturate_cast<short>(a.x / b), saturate_cast<short>(a.y / b),
+ saturate_cast<short>(a.z / b), saturate_cast<short>(a.w / b))
+ : make_short4(0,0,0,0);
+ }
+ };
+
+ template <typename T, typename D> struct Div : binary_function<T, T, D>
+ {
+ __device__ __forceinline__ D operator ()(T a, T b) const
+ {
+ return b != 0 ? saturate_cast<D>(a / b) : 0;
+ }
+
+ __device__ __forceinline__ Div() {}
+ __device__ __forceinline__ Div(const Div& other) {}
+ };
+ template <typename T> struct Div<T, float> : binary_function<T, T, float>
+ {
+ __device__ __forceinline__ float operator ()(T a, T b) const
+ {
+ return b != 0 ? static_cast<float>(a) / b : 0;
+ }
+
+ __device__ __forceinline__ Div() {}
+ __device__ __forceinline__ Div(const Div& other) {}
+ };
+ template <typename T> struct Div<T, double> : binary_function<T, T, double>
+ {
+ __device__ __forceinline__ double operator ()(T a, T b) const
+ {
+ return b != 0 ? static_cast<double>(a) / b : 0;
+ }
+
+ __device__ __forceinline__ Div() {}
+ __device__ __forceinline__ Div(const Div& other) {}
+ };
+
+ template <typename T, typename S, typename D> struct DivScale : binary_function<T, T, D>
+ {
+ S scale;
+
+ explicit DivScale(S scale_) : scale(scale_) {}
+
+ __device__ __forceinline__ D operator ()(T a, T b) const
+ {
+ return b != 0 ? saturate_cast<D>(scale * a / b) : 0;
+ }
+ };
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <> struct TransformFunctorTraits<arithm::Div_8uc4_32f> : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
+ {
+ };
+
+ template <typename T, typename D> struct TransformFunctorTraits< arithm::Div<T, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
+ {
+ };
+
+ template <typename T, typename S, typename D> struct TransformFunctorTraits< arithm::DivScale<T, S, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ void divMat_8uc4_32f(PtrStepSz<uint> src1, PtrStepSzf src2, PtrStepSz<uint> dst, cudaStream_t stream)
+ {
+ cudev::transform(src1, src2, dst, Div_8uc4_32f(), WithOutMask(), stream);
+ }
+
+ void divMat_16sc4_32f(PtrStepSz<short4> src1, PtrStepSzf src2, PtrStepSz<short4> dst, cudaStream_t stream)
+ {
+ cudev::transform(src1, src2, dst, Div_16sc4_32f(), WithOutMask(), stream);
+ }
+
+ template <typename T, typename S, typename D>
+ void divMat(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream)
+ {
+ if (scale == 1)
+ {
+ Div<T, D> op;
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
+ }
+ else
+ {
+ DivScale<T, S, D> op(static_cast<S>(scale));
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
+ }
+ }
+
+ template void divMat<uchar, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<uchar, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<uchar, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<uchar, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<uchar, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<uchar, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<uchar, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+
+ template void divMat<schar, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<schar, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<schar, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<schar, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<schar, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<schar, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<schar, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+
+ //template void divMat<ushort, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void divMat<ushort, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<ushort, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<ushort, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<ushort, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<ushort, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<ushort, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+
+ //template void divMat<short, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void divMat<short, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<short, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<short, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<short, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<short, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<short, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+
+ //template void divMat<int, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void divMat<int, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void divMat<int, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void divMat<int, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<int, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<int, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<int, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+
+ //template void divMat<float, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void divMat<float, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void divMat<float, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void divMat<float, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void divMat<float, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<float, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<float, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+
+ //template void divMat<double, double, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void divMat<double, double, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void divMat<double, double, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void divMat<double, double, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void divMat<double, double, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void divMat<double, double, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void divMat<double, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/simd_functions.hpp"
+
+#include "arithm_func_traits.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace arithm
+{
+ template <typename T, typename S, typename D> struct DivScalar : unary_function<T, D>
+ {
+ S val;
+
+ explicit DivScalar(S val_) : val(val_) {}
+
+ __device__ __forceinline__ D operator ()(T a) const
+ {
+ return saturate_cast<D>(a / val);
+ }
+ };
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <typename T, typename S, typename D> struct TransformFunctorTraits< arithm::DivScalar<T, S, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ template <typename T, typename S, typename D>
+ void divScalar(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream)
+ {
+ DivScalar<T, S, D> op(static_cast<S>(val));
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
+ }
+
+ template void divScalar<uchar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<uchar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<uchar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<uchar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<uchar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<uchar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<uchar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+
+ template void divScalar<schar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<schar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<schar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<schar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<schar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<schar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<schar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+
+ //template void divScalar<ushort, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divScalar<ushort, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<ushort, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<ushort, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<ushort, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<ushort, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<ushort, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+
+ //template void divScalar<short, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divScalar<short, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<short, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<short, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<short, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<short, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<short, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+
+ //template void divScalar<int, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divScalar<int, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divScalar<int, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divScalar<int, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<int, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<int, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<int, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+
+ //template void divScalar<float, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divScalar<float, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divScalar<float, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divScalar<float, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divScalar<float, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<float, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<float, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+
+ //template void divScalar<double, double, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divScalar<double, double, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divScalar<double, double, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divScalar<double, double, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divScalar<double, double, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void divScalar<double, double, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void divScalar<double, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/simd_functions.hpp"
+#include "opencv2/core/cuda/limits.hpp"
+#include "opencv2/core/cuda/type_traits.hpp"
+
+#include "arithm_func_traits.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+//////////////////////////////////////////////////////////////////////////
+// absMat
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <typename T> struct TransformFunctorTraits< abs_func<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ template <typename T>
+ void absMat(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream)
+ {
+ cudev::transform((PtrStepSz<T>) src, (PtrStepSz<T>) dst, abs_func<T>(), WithOutMask(), stream);
+ }
+
+ template void absMat<uchar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void absMat<schar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void absMat<ushort>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void absMat<short>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void absMat<int>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void absMat<float>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void absMat<double>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+}
+
+//////////////////////////////////////////////////////////////////////////
+// sqrMat
+
+namespace arithm
+{
+ template <typename T> struct Sqr : unary_function<T, T>
+ {
+ __device__ __forceinline__ T operator ()(T x) const
+ {
+ return saturate_cast<T>(x * x);
+ }
+
+ __device__ __forceinline__ Sqr() {}
+ __device__ __forceinline__ Sqr(const Sqr& other) {}
+ };
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <typename T> struct TransformFunctorTraits< arithm::Sqr<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ template <typename T>
+ void sqrMat(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream)
+ {
+ cudev::transform((PtrStepSz<T>) src, (PtrStepSz<T>) dst, Sqr<T>(), WithOutMask(), stream);
+ }
+
+ template void sqrMat<uchar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void sqrMat<schar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void sqrMat<ushort>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void sqrMat<short>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void sqrMat<int>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void sqrMat<float>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void sqrMat<double>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+}
+
+//////////////////////////////////////////////////////////////////////////
+// sqrtMat
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <typename T> struct TransformFunctorTraits< sqrt_func<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ template <typename T>
+ void sqrtMat(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream)
+ {
+ cudev::transform((PtrStepSz<T>) src, (PtrStepSz<T>) dst, sqrt_func<T>(), WithOutMask(), stream);
+ }
+
+ template void sqrtMat<uchar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void sqrtMat<schar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void sqrtMat<ushort>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void sqrtMat<short>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void sqrtMat<int>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void sqrtMat<float>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void sqrtMat<double>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+}
+
+//////////////////////////////////////////////////////////////////////////
+// logMat
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <typename T> struct TransformFunctorTraits< log_func<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ template <typename T>
+ void logMat(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream)
+ {
+ cudev::transform((PtrStepSz<T>) src, (PtrStepSz<T>) dst, log_func<T>(), WithOutMask(), stream);
+ }
+
+ template void logMat<uchar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void logMat<schar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void logMat<ushort>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void logMat<short>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void logMat<int>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void logMat<float>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void logMat<double>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+}
+
+//////////////////////////////////////////////////////////////////////////
+// expMat
+
+namespace arithm
+{
+ template <typename T> struct Exp : unary_function<T, T>
+ {
+ __device__ __forceinline__ T operator ()(T x) const
+ {
+ exp_func<T> f;
+ return saturate_cast<T>(f(x));
+ }
+
+ __device__ __forceinline__ Exp() {}
+ __device__ __forceinline__ Exp(const Exp& other) {}
+ };
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <typename T> struct TransformFunctorTraits< arithm::Exp<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ template <typename T>
+ void expMat(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream)
+ {
+ cudev::transform((PtrStepSz<T>) src, (PtrStepSz<T>) dst, Exp<T>(), WithOutMask(), stream);
+ }
+
+ template void expMat<uchar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void expMat<schar>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void expMat<ushort>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void expMat<short>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void expMat<int>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void expMat<float>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ template void expMat<double>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+}
+
+//////////////////////////////////////////////////////////////////////////
+// pow
+
+namespace arithm
+{
+ template<typename T, bool Signed = numeric_limits<T>::is_signed> struct PowOp : unary_function<T, T>
+ {
+ float power;
+
+ PowOp(double power_) : power(static_cast<float>(power_)) {}
+
+ __device__ __forceinline__ T operator()(T e) const
+ {
+ return saturate_cast<T>(__powf((float)e, power));
+ }
+ };
+ template<typename T> struct PowOp<T, true> : unary_function<T, T>
+ {
+ float power;
+
+ PowOp(double power_) : power(static_cast<float>(power_)) {}
+
+ __device__ __forceinline__ T operator()(T e) const
+ {
+ T res = saturate_cast<T>(__powf((float)e, power));
+
+ if ((e < 0) && (1 & static_cast<int>(power)))
+ res *= -1;
+
+ return res;
+ }
+ };
+ template<> struct PowOp<float> : unary_function<float, float>
+ {
+ float power;
+
+ PowOp(double power_) : power(static_cast<float>(power_)) {}
+
+ __device__ __forceinline__ float operator()(float e) const
+ {
+ return __powf(::fabs(e), power);
+ }
+ };
+ template<> struct PowOp<double> : unary_function<double, double>
+ {
+ double power;
+
+ PowOp(double power_) : power(power_) {}
+
+ __device__ __forceinline__ double operator()(double e) const
+ {
+ return ::pow(::fabs(e), power);
+ }
+ };
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <typename T> struct TransformFunctorTraits< arithm::PowOp<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ template<typename T>
+ void pow(PtrStepSzb src, double power, PtrStepSzb dst, cudaStream_t stream)
+ {
+ cudev::transform((PtrStepSz<T>) src, (PtrStepSz<T>) dst, PowOp<T>(power), WithOutMask(), stream);
+ }
+
+ template void pow<uchar>(PtrStepSzb src, double power, PtrStepSzb dst, cudaStream_t stream);
+ template void pow<schar>(PtrStepSzb src, double power, PtrStepSzb dst, cudaStream_t stream);
+ template void pow<short>(PtrStepSzb src, double power, PtrStepSzb dst, cudaStream_t stream);
+ template void pow<ushort>(PtrStepSzb src, double power, PtrStepSzb dst, cudaStream_t stream);
+ template void pow<int>(PtrStepSzb src, double power, PtrStepSzb dst, cudaStream_t stream);
+ template void pow<float>(PtrStepSzb src, double power, PtrStepSzb dst, cudaStream_t stream);
+ template void pow<double>(PtrStepSzb src, double power, PtrStepSzb dst, cudaStream_t stream);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/vec_traits.hpp"
+#include "opencv2/core/cuda/vec_math.hpp"
+#include "opencv2/core/cuda/reduce.hpp"
+#include "opencv2/core/cuda/emulation.hpp"
+#include "opencv2/core/cuda/limits.hpp"
+#include "opencv2/core/cuda/utility.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace minMax
+{
+ __device__ unsigned int blocks_finished = 0;
+
+ // To avoid shared bank conflicts we convert each value into value of
+ // appropriate type (32 bits minimum)
+ template <typename T> struct MinMaxTypeTraits;
+ template <> struct MinMaxTypeTraits<uchar> { typedef int best_type; };
+ template <> struct MinMaxTypeTraits<schar> { typedef int best_type; };
+ template <> struct MinMaxTypeTraits<ushort> { typedef int best_type; };
+ template <> struct MinMaxTypeTraits<short> { typedef int best_type; };
+ template <> struct MinMaxTypeTraits<int> { typedef int best_type; };
+ template <> struct MinMaxTypeTraits<float> { typedef float best_type; };
+ template <> struct MinMaxTypeTraits<double> { typedef double best_type; };
+
+ template <int BLOCK_SIZE, typename R>
+ struct GlobalReduce
+ {
+ static __device__ void run(R& mymin, R& mymax, R* minval, R* maxval, int tid, int bid, R* sminval, R* smaxval)
+ {
+ #if __CUDA_ARCH__ >= 200
+ if (tid == 0)
+ {
+ Emulation::glob::atomicMin(minval, mymin);
+ Emulation::glob::atomicMax(maxval, mymax);
+ }
+ #else
+ __shared__ bool is_last;
+
+ if (tid == 0)
+ {
+ minval[bid] = mymin;
+ maxval[bid] = mymax;
+
+ __threadfence();
+
+ unsigned int ticket = ::atomicAdd(&blocks_finished, 1);
+ is_last = (ticket == gridDim.x * gridDim.y - 1);
+ }
+
+ __syncthreads();
+
+ if (is_last)
+ {
+ int idx = ::min(tid, gridDim.x * gridDim.y - 1);
+
+ mymin = minval[idx];
+ mymax = maxval[idx];
+
+ const minimum<R> minOp;
+ const maximum<R> maxOp;
+ cudev::reduce<BLOCK_SIZE>(smem_tuple(sminval, smaxval), thrust::tie(mymin, mymax), tid, thrust::make_tuple(minOp, maxOp));
+
+ if (tid == 0)
+ {
+ minval[0] = mymin;
+ maxval[0] = mymax;
+
+ blocks_finished = 0;
+ }
+ }
+ #endif
+ }
+ };
+
+ template <int BLOCK_SIZE, typename T, typename R, class Mask>
+ __global__ void kernel(const PtrStepSz<T> src, const Mask mask, R* minval, R* maxval, const int twidth, const int theight)
+ {
+ __shared__ R sminval[BLOCK_SIZE];
+ __shared__ R smaxval[BLOCK_SIZE];
+
+ const int x0 = blockIdx.x * blockDim.x * twidth + threadIdx.x;
+ const int y0 = blockIdx.y * blockDim.y * theight + threadIdx.y;
+
+ const int tid = threadIdx.y * blockDim.x + threadIdx.x;
+ const int bid = blockIdx.y * gridDim.x + blockIdx.x;
+
+ R mymin = numeric_limits<R>::max();
+ R mymax = -numeric_limits<R>::max();
+
+ const minimum<R> minOp;
+ const maximum<R> maxOp;
+
+ for (int i = 0, y = y0; i < theight && y < src.rows; ++i, y += blockDim.y)
+ {
+ const T* ptr = src.ptr(y);
+
+ for (int j = 0, x = x0; j < twidth && x < src.cols; ++j, x += blockDim.x)
+ {
+ if (mask(y, x))
+ {
+ const R srcVal = ptr[x];
+
+ mymin = minOp(mymin, srcVal);
+ mymax = maxOp(mymax, srcVal);
+ }
+ }
+ }
+
+ cudev::reduce<BLOCK_SIZE>(smem_tuple(sminval, smaxval), thrust::tie(mymin, mymax), tid, thrust::make_tuple(minOp, maxOp));
+
+ GlobalReduce<BLOCK_SIZE, R>::run(mymin, mymax, minval, maxval, tid, bid, sminval, smaxval);
+ }
+
+ const int threads_x = 32;
+ const int threads_y = 8;
+
+ void getLaunchCfg(int cols, int rows, dim3& block, dim3& grid)
+ {
+ block = dim3(threads_x, threads_y);
+
+ grid = dim3(divUp(cols, block.x * block.y),
+ divUp(rows, block.y * block.x));
+
+ grid.x = ::min(grid.x, block.x);
+ grid.y = ::min(grid.y, block.y);
+ }
+
+ void getBufSize(int cols, int rows, int& bufcols, int& bufrows)
+ {
+ dim3 block, grid;
+ getLaunchCfg(cols, rows, block, grid);
+
+ bufcols = grid.x * grid.y * sizeof(double);
+ bufrows = 2;
+ }
+
+ __global__ void setDefaultKernel(int* minval_buf, int* maxval_buf)
+ {
+ *minval_buf = numeric_limits<int>::max();
+ *maxval_buf = numeric_limits<int>::min();
+ }
+ __global__ void setDefaultKernel(float* minval_buf, float* maxval_buf)
+ {
+ *minval_buf = numeric_limits<float>::max();
+ *maxval_buf = -numeric_limits<float>::max();
+ }
+ __global__ void setDefaultKernel(double* minval_buf, double* maxval_buf)
+ {
+ *minval_buf = numeric_limits<double>::max();
+ *maxval_buf = -numeric_limits<double>::max();
+ }
+
+ template <typename R>
+ void setDefault(R* minval_buf, R* maxval_buf)
+ {
+ setDefaultKernel<<<1, 1>>>(minval_buf, maxval_buf);
+ }
+
+ template <typename T>
+ void run(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf)
+ {
+ typedef typename MinMaxTypeTraits<T>::best_type R;
+
+ dim3 block, grid;
+ getLaunchCfg(src.cols, src.rows, block, grid);
+
+ const int twidth = divUp(divUp(src.cols, grid.x), block.x);
+ const int theight = divUp(divUp(src.rows, grid.y), block.y);
+
+ R* minval_buf = (R*) buf.ptr(0);
+ R* maxval_buf = (R*) buf.ptr(1);
+
+ setDefault(minval_buf, maxval_buf);
+
+ if (mask.data)
+ kernel<threads_x * threads_y><<<grid, block>>>((PtrStepSz<T>) src, SingleMask(mask), minval_buf, maxval_buf, twidth, theight);
+ else
+ kernel<threads_x * threads_y><<<grid, block>>>((PtrStepSz<T>) src, WithOutMask(), minval_buf, maxval_buf, twidth, theight);
+
+ cudaSafeCall( cudaGetLastError() );
+
+ cudaSafeCall( cudaDeviceSynchronize() );
+
+ R minval_, maxval_;
+ cudaSafeCall( cudaMemcpy(&minval_, minval_buf, sizeof(R), cudaMemcpyDeviceToHost) );
+ cudaSafeCall( cudaMemcpy(&maxval_, maxval_buf, sizeof(R), cudaMemcpyDeviceToHost) );
+ *minval = minval_;
+ *maxval = maxval_;
+ }
+
+ template void run<uchar >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf);
+ template void run<schar >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf);
+ template void run<ushort>(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf);
+ template void run<short >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf);
+ template void run<int >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf);
+ template void run<float >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf);
+ template void run<double>(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/simd_functions.hpp"
+
+#include "arithm_func_traits.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+//////////////////////////////////////////////////////////////////////////
+// min
+
+namespace arithm
+{
+ struct VMin4 : binary_function<uint, uint, uint>
+ {
+ __device__ __forceinline__ uint operator ()(uint a, uint b) const
+ {
+ return vmin4(a, b);
+ }
+
+ __device__ __forceinline__ VMin4() {}
+ __device__ __forceinline__ VMin4(const VMin4& other) {}
+ };
+
+ struct VMin2 : binary_function<uint, uint, uint>
+ {
+ __device__ __forceinline__ uint operator ()(uint a, uint b) const
+ {
+ return vmin2(a, b);
+ }
+
+ __device__ __forceinline__ VMin2() {}
+ __device__ __forceinline__ VMin2(const VMin2& other) {}
+ };
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <> struct TransformFunctorTraits< arithm::VMin4 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
+ {
+ };
+
+ template <> struct TransformFunctorTraits< arithm::VMin2 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
+ {
+ };
+
+ template <typename T> struct TransformFunctorTraits< minimum<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+
+ template <typename T> struct TransformFunctorTraits< binder2nd< minimum<T> > > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ void minMat_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
+ {
+ cudev::transform(src1, src2, dst, VMin4(), WithOutMask(), stream);
+ }
+
+ void minMat_v2(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
+ {
+ cudev::transform(src1, src2, dst, VMin2(), WithOutMask(), stream);
+ }
+
+ template <typename T> void minMat(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream)
+ {
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<T>) dst, minimum<T>(), WithOutMask(), stream);
+ }
+
+ template void minMat<uchar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void minMat<schar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void minMat<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void minMat<short >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void minMat<int >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void minMat<float >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void minMat<double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+
+ template <typename T> void minScalar(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream)
+ {
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) dst, cv::gpu::cudev::bind2nd(minimum<T>(), src2), WithOutMask(), stream);
+ }
+
+ template void minScalar<uchar >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+ template void minScalar<schar >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+ template void minScalar<ushort>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+ template void minScalar<short >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+ template void minScalar<int >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+ template void minScalar<float >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+ template void minScalar<double>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+}
+
+//////////////////////////////////////////////////////////////////////////
+// max
+
+namespace arithm
+{
+ struct VMax4 : binary_function<uint, uint, uint>
+ {
+ __device__ __forceinline__ uint operator ()(uint a, uint b) const
+ {
+ return vmax4(a, b);
+ }
+
+ __device__ __forceinline__ VMax4() {}
+ __device__ __forceinline__ VMax4(const VMax4& other) {}
+ };
+
+ struct VMax2 : binary_function<uint, uint, uint>
+ {
+ __device__ __forceinline__ uint operator ()(uint a, uint b) const
+ {
+ return vmax2(a, b);
+ }
+
+ __device__ __forceinline__ VMax2() {}
+ __device__ __forceinline__ VMax2(const VMax2& other) {}
+ };
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <> struct TransformFunctorTraits< arithm::VMax4 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
+ {
+ };
+
+ template <> struct TransformFunctorTraits< arithm::VMax2 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
+ {
+ };
+
+ template <typename T> struct TransformFunctorTraits< maximum<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+
+ template <typename T> struct TransformFunctorTraits< binder2nd< maximum<T> > > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ void maxMat_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
+ {
+ cudev::transform(src1, src2, dst, VMax4(), WithOutMask(), stream);
+ }
+
+ void maxMat_v2(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
+ {
+ cudev::transform(src1, src2, dst, VMax2(), WithOutMask(), stream);
+ }
+
+ template <typename T> void maxMat(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream)
+ {
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<T>) dst, maximum<T>(), WithOutMask(), stream);
+ }
+
+ template void maxMat<uchar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void maxMat<schar >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void maxMat<ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void maxMat<short >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void maxMat<int >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void maxMat<float >(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+ template void maxMat<double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
+
+ template <typename T> void maxScalar(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream)
+ {
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) dst, cv::gpu::cudev::bind2nd(maximum<T>(), src2), WithOutMask(), stream);
+ }
+
+ template void maxScalar<uchar >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+ template void maxScalar<schar >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+ template void maxScalar<ushort>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+ template void maxScalar<short >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+ template void maxScalar<int >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+ template void maxScalar<float >(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+ template void maxScalar<double>(PtrStepSzb src1, double src2, PtrStepSzb dst, cudaStream_t stream);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/vec_traits.hpp"
+#include "opencv2/core/cuda/vec_math.hpp"
+#include "opencv2/core/cuda/reduce.hpp"
+#include "opencv2/core/cuda/emulation.hpp"
+#include "opencv2/core/cuda/limits.hpp"
+#include "opencv2/core/cuda/utility.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace minMaxLoc
+{
+ // To avoid shared bank conflicts we convert each value into value of
+ // appropriate type (32 bits minimum)
+ template <typename T> struct MinMaxTypeTraits;
+ template <> struct MinMaxTypeTraits<unsigned char> { typedef int best_type; };
+ template <> struct MinMaxTypeTraits<signed char> { typedef int best_type; };
+ template <> struct MinMaxTypeTraits<unsigned short> { typedef int best_type; };
+ template <> struct MinMaxTypeTraits<short> { typedef int best_type; };
+ template <> struct MinMaxTypeTraits<int> { typedef int best_type; };
+ template <> struct MinMaxTypeTraits<float> { typedef float best_type; };
+ template <> struct MinMaxTypeTraits<double> { typedef double best_type; };
+
+ template <int BLOCK_SIZE, typename T, class Mask>
+ __global__ void kernel_pass_1(const PtrStepSz<T> src, const Mask mask, T* minval, T* maxval, unsigned int* minloc, unsigned int* maxloc, const int twidth, const int theight)
+ {
+ typedef typename MinMaxTypeTraits<T>::best_type work_type;
+
+ __shared__ work_type sminval[BLOCK_SIZE];
+ __shared__ work_type smaxval[BLOCK_SIZE];
+ __shared__ unsigned int sminloc[BLOCK_SIZE];
+ __shared__ unsigned int smaxloc[BLOCK_SIZE];
+
+ const int x0 = blockIdx.x * blockDim.x * twidth + threadIdx.x;
+ const int y0 = blockIdx.y * blockDim.y * theight + threadIdx.y;
+
+ const int tid = threadIdx.y * blockDim.x + threadIdx.x;
+ const int bid = blockIdx.y * gridDim.x + blockIdx.x;
+
+ work_type mymin = numeric_limits<work_type>::max();
+ work_type mymax = -numeric_limits<work_type>::max();
+ unsigned int myminloc = 0;
+ unsigned int mymaxloc = 0;
+
+ for (int i = 0, y = y0; i < theight && y < src.rows; ++i, y += blockDim.y)
+ {
+ const T* ptr = src.ptr(y);
+
+ for (int j = 0, x = x0; j < twidth && x < src.cols; ++j, x += blockDim.x)
+ {
+ if (mask(y, x))
+ {
+ const work_type srcVal = ptr[x];
+
+ if (srcVal < mymin)
+ {
+ mymin = srcVal;
+ myminloc = y * src.cols + x;
+ }
+
+ if (srcVal > mymax)
+ {
+ mymax = srcVal;
+ mymaxloc = y * src.cols + x;
+ }
+ }
+ }
+ }
+
+ reduceKeyVal<BLOCK_SIZE>(smem_tuple(sminval, smaxval), thrust::tie(mymin, mymax),
+ smem_tuple(sminloc, smaxloc), thrust::tie(myminloc, mymaxloc),
+ tid,
+ thrust::make_tuple(less<work_type>(), greater<work_type>()));
+
+ if (tid == 0)
+ {
+ minval[bid] = (T) mymin;
+ maxval[bid] = (T) mymax;
+ minloc[bid] = myminloc;
+ maxloc[bid] = mymaxloc;
+ }
+ }
+ template <int BLOCK_SIZE, typename T>
+ __global__ void kernel_pass_2(T* minval, T* maxval, unsigned int* minloc, unsigned int* maxloc, int count)
+ {
+ typedef typename MinMaxTypeTraits<T>::best_type work_type;
+
+ __shared__ work_type sminval[BLOCK_SIZE];
+ __shared__ work_type smaxval[BLOCK_SIZE];
+ __shared__ unsigned int sminloc[BLOCK_SIZE];
+ __shared__ unsigned int smaxloc[BLOCK_SIZE];
+
+ unsigned int idx = ::min(threadIdx.x, count - 1);
+
+ work_type mymin = minval[idx];
+ work_type mymax = maxval[idx];
+ unsigned int myminloc = minloc[idx];
+ unsigned int mymaxloc = maxloc[idx];
+
+ reduceKeyVal<BLOCK_SIZE>(smem_tuple(sminval, smaxval), thrust::tie(mymin, mymax),
+ smem_tuple(sminloc, smaxloc), thrust::tie(myminloc, mymaxloc),
+ threadIdx.x,
+ thrust::make_tuple(less<work_type>(), greater<work_type>()));
+
+ if (threadIdx.x == 0)
+ {
+ minval[0] = (T) mymin;
+ maxval[0] = (T) mymax;
+ minloc[0] = myminloc;
+ maxloc[0] = mymaxloc;
+ }
+ }
+
+ const int threads_x = 32;
+ const int threads_y = 8;
+
+ void getLaunchCfg(int cols, int rows, dim3& block, dim3& grid)
+ {
+ block = dim3(threads_x, threads_y);
+
+ grid = dim3(divUp(cols, block.x * block.y),
+ divUp(rows, block.y * block.x));
+
+ grid.x = ::min(grid.x, block.x);
+ grid.y = ::min(grid.y, block.y);
+ }
+
+ void getBufSize(int cols, int rows, size_t elem_size, int& b1cols, int& b1rows, int& b2cols, int& b2rows)
+ {
+ dim3 block, grid;
+ getLaunchCfg(cols, rows, block, grid);
+
+ // For values
+ b1cols = (int)(grid.x * grid.y * elem_size);
+ b1rows = 2;
+
+ // For locations
+ b2cols = grid.x * grid.y * sizeof(int);
+ b2rows = 2;
+ }
+
+ template <typename T>
+ void run(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, int* minloc, int* maxloc, PtrStepb valbuf, PtrStep<unsigned int> locbuf)
+ {
+ dim3 block, grid;
+ getLaunchCfg(src.cols, src.rows, block, grid);
+
+ const int twidth = divUp(divUp(src.cols, grid.x), block.x);
+ const int theight = divUp(divUp(src.rows, grid.y), block.y);
+
+ T* minval_buf = (T*) valbuf.ptr(0);
+ T* maxval_buf = (T*) valbuf.ptr(1);
+ unsigned int* minloc_buf = locbuf.ptr(0);
+ unsigned int* maxloc_buf = locbuf.ptr(1);
+
+ if (mask.data)
+ kernel_pass_1<threads_x * threads_y><<<grid, block>>>((PtrStepSz<T>) src, SingleMask(mask), minval_buf, maxval_buf, minloc_buf, maxloc_buf, twidth, theight);
+ else
+ kernel_pass_1<threads_x * threads_y><<<grid, block>>>((PtrStepSz<T>) src, WithOutMask(), minval_buf, maxval_buf, minloc_buf, maxloc_buf, twidth, theight);
+
+ cudaSafeCall( cudaGetLastError() );
+
+ kernel_pass_2<threads_x * threads_y><<<1, threads_x * threads_y>>>(minval_buf, maxval_buf, minloc_buf, maxloc_buf, grid.x * grid.y);
+ cudaSafeCall( cudaGetLastError() );
+
+ cudaSafeCall( cudaDeviceSynchronize() );
+
+ T minval_, maxval_;
+ cudaSafeCall( cudaMemcpy(&minval_, minval_buf, sizeof(T), cudaMemcpyDeviceToHost) );
+ cudaSafeCall( cudaMemcpy(&maxval_, maxval_buf, sizeof(T), cudaMemcpyDeviceToHost) );
+ *minval = minval_;
+ *maxval = maxval_;
+
+ unsigned int minloc_, maxloc_;
+ cudaSafeCall( cudaMemcpy(&minloc_, minloc_buf, sizeof(unsigned int), cudaMemcpyDeviceToHost) );
+ cudaSafeCall( cudaMemcpy(&maxloc_, maxloc_buf, sizeof(unsigned int), cudaMemcpyDeviceToHost) );
+ minloc[1] = minloc_ / src.cols; minloc[0] = minloc_ - minloc[1] * src.cols;
+ maxloc[1] = maxloc_ / src.cols; maxloc[0] = maxloc_ - maxloc[1] * src.cols;
+ }
+
+ template void run<unsigned char >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, int* minloc, int* maxloc, PtrStepb valbuf, PtrStep<unsigned int> locbuf);
+ template void run<signed char >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, int* minloc, int* maxloc, PtrStepb valbuf, PtrStep<unsigned int> locbuf);
+ template void run<unsigned short>(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, int* minloc, int* maxloc, PtrStepb valbuf, PtrStep<unsigned int> locbuf);
+ template void run<short >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, int* minloc, int* maxloc, PtrStepb valbuf, PtrStep<unsigned int> locbuf);
+ template void run<int >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, int* minloc, int* maxloc, PtrStepb valbuf, PtrStep<unsigned int> locbuf);
+ template void run<float >(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, int* minloc, int* maxloc, PtrStepb valbuf, PtrStep<unsigned int> locbuf);
+ template void run<double>(const PtrStepSzb src, const PtrStepb mask, double* minval, double* maxval, int* minloc, int* maxloc, PtrStepb valbuf, PtrStep<unsigned int> locbuf);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/simd_functions.hpp"
+
+#include "arithm_func_traits.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace arithm
+{
+ struct Mul_8uc4_32f : binary_function<uint, float, uint>
+ {
+ __device__ __forceinline__ uint operator ()(uint a, float b) const
+ {
+ uint res = 0;
+
+ res |= (saturate_cast<uchar>((0xffu & (a )) * b) );
+ res |= (saturate_cast<uchar>((0xffu & (a >> 8)) * b) << 8);
+ res |= (saturate_cast<uchar>((0xffu & (a >> 16)) * b) << 16);
+ res |= (saturate_cast<uchar>((0xffu & (a >> 24)) * b) << 24);
+
+ return res;
+ }
+
+ __device__ __forceinline__ Mul_8uc4_32f() {}
+ __device__ __forceinline__ Mul_8uc4_32f(const Mul_8uc4_32f& other) {}
+ };
+
+ struct Mul_16sc4_32f : binary_function<short4, float, short4>
+ {
+ __device__ __forceinline__ short4 operator ()(short4 a, float b) const
+ {
+ return make_short4(saturate_cast<short>(a.x * b), saturate_cast<short>(a.y * b),
+ saturate_cast<short>(a.z * b), saturate_cast<short>(a.w * b));
+ }
+
+ __device__ __forceinline__ Mul_16sc4_32f() {}
+ __device__ __forceinline__ Mul_16sc4_32f(const Mul_16sc4_32f& other) {}
+ };
+
+ template <typename T, typename D> struct Mul : binary_function<T, T, D>
+ {
+ __device__ __forceinline__ D operator ()(T a, T b) const
+ {
+ return saturate_cast<D>(a * b);
+ }
+
+ __device__ __forceinline__ Mul() {}
+ __device__ __forceinline__ Mul(const Mul& other) {}
+ };
+
+ template <typename T, typename S, typename D> struct MulScale : binary_function<T, T, D>
+ {
+ S scale;
+
+ explicit MulScale(S scale_) : scale(scale_) {}
+
+ __device__ __forceinline__ D operator ()(T a, T b) const
+ {
+ return saturate_cast<D>(scale * a * b);
+ }
+ };
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <> struct TransformFunctorTraits<arithm::Mul_8uc4_32f> : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
+ {
+ };
+
+ template <typename T, typename D> struct TransformFunctorTraits< arithm::Mul<T, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
+ {
+ };
+
+ template <typename T, typename S, typename D> struct TransformFunctorTraits< arithm::MulScale<T, S, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ void mulMat_8uc4_32f(PtrStepSz<uint> src1, PtrStepSzf src2, PtrStepSz<uint> dst, cudaStream_t stream)
+ {
+ cudev::transform(src1, src2, dst, Mul_8uc4_32f(), WithOutMask(), stream);
+ }
+
+ void mulMat_16sc4_32f(PtrStepSz<short4> src1, PtrStepSzf src2, PtrStepSz<short4> dst, cudaStream_t stream)
+ {
+ cudev::transform(src1, src2, dst, Mul_16sc4_32f(), WithOutMask(), stream);
+ }
+
+ template <typename T, typename S, typename D>
+ void mulMat(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream)
+ {
+ if (scale == 1)
+ {
+ Mul<T, D> op;
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
+ }
+ else
+ {
+ MulScale<T, S, D> op(static_cast<S>(scale));
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
+ }
+ }
+
+ template void mulMat<uchar, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<uchar, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<uchar, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<uchar, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<uchar, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<uchar, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<uchar, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+
+ template void mulMat<schar, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<schar, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<schar, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<schar, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<schar, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<schar, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<schar, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+
+ //template void mulMat<ushort, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void mulMat<ushort, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<ushort, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<ushort, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<ushort, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<ushort, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<ushort, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+
+ //template void mulMat<short, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void mulMat<short, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<short, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<short, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<short, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<short, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<short, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+
+ //template void mulMat<int, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void mulMat<int, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void mulMat<int, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void mulMat<int, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<int, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<int, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<int, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+
+ //template void mulMat<float, float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void mulMat<float, float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void mulMat<float, float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void mulMat<float, float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void mulMat<float, float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<float, float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<float, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+
+ //template void mulMat<double, double, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void mulMat<double, double, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void mulMat<double, double, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void mulMat<double, double, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void mulMat<double, double, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ //template void mulMat<double, double, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+ template void mulMat<double, double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, double scale, cudaStream_t stream);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/simd_functions.hpp"
+
+#include "arithm_func_traits.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace arithm
+{
+ template <typename T, typename S, typename D> struct MulScalar : unary_function<T, D>
+ {
+ S val;
+
+ explicit MulScalar(S val_) : val(val_) {}
+
+ __device__ __forceinline__ D operator ()(T a) const
+ {
+ return saturate_cast<D>(a * val);
+ }
+ };
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <typename T, typename S, typename D> struct TransformFunctorTraits< arithm::MulScalar<T, S, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ template <typename T, typename S, typename D>
+ void mulScalar(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream)
+ {
+ MulScalar<T, S, D> op(static_cast<S>(val));
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
+ }
+
+ template void mulScalar<uchar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<uchar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<uchar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<uchar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<uchar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<uchar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<uchar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+
+ template void mulScalar<schar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<schar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<schar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<schar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<schar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<schar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<schar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+
+ //template void mulScalar<ushort, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void mulScalar<ushort, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<ushort, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<ushort, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<ushort, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<ushort, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<ushort, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+
+ //template void mulScalar<short, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void mulScalar<short, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<short, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<short, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<short, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<short, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<short, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+
+ //template void mulScalar<int, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void mulScalar<int, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void mulScalar<int, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void mulScalar<int, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<int, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<int, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<int, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+
+ //template void mulScalar<float, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void mulScalar<float, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void mulScalar<float, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void mulScalar<float, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void mulScalar<float, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<float, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<float, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+
+ //template void mulScalar<double, double, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void mulScalar<double, double, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void mulScalar<double, double, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void mulScalar<double, double, short>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void mulScalar<double, double, int>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ //template void mulScalar<double, double, float>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+ template void mulScalar<double, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, cudaStream_t stream);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "cvconfig.h"
+
+#ifdef HAVE_CUFFT
+
+#include <cufft.h>
+
+#include "opencv2/core/cuda/common.hpp"
+
+namespace cv { namespace gpu { namespace cudev
+{
+ //////////////////////////////////////////////////////////////////////////
+ // mulSpectrums
+
+ __global__ void mulSpectrumsKernel(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, PtrStepSz<cufftComplex> c)
+ {
+ const int x = blockIdx.x * blockDim.x + threadIdx.x;
+ const int y = blockIdx.y * blockDim.y + threadIdx.y;
+
+ if (x < c.cols && y < c.rows)
+ {
+ c.ptr(y)[x] = cuCmulf(a.ptr(y)[x], b.ptr(y)[x]);
+ }
+ }
+
+
+ void mulSpectrums(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, PtrStepSz<cufftComplex> c, cudaStream_t stream)
+ {
+ dim3 threads(256);
+ dim3 grid(divUp(c.cols, threads.x), divUp(c.rows, threads.y));
+
+ mulSpectrumsKernel<<<grid, threads, 0, stream>>>(a, b, c);
+ cudaSafeCall( cudaGetLastError() );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+
+
+ //////////////////////////////////////////////////////////////////////////
+ // mulSpectrums_CONJ
+
+ __global__ void mulSpectrumsKernel_CONJ(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, PtrStepSz<cufftComplex> c)
+ {
+ const int x = blockIdx.x * blockDim.x + threadIdx.x;
+ const int y = blockIdx.y * blockDim.y + threadIdx.y;
+
+ if (x < c.cols && y < c.rows)
+ {
+ c.ptr(y)[x] = cuCmulf(a.ptr(y)[x], cuConjf(b.ptr(y)[x]));
+ }
+ }
+
+
+ void mulSpectrums_CONJ(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, PtrStepSz<cufftComplex> c, cudaStream_t stream)
+ {
+ dim3 threads(256);
+ dim3 grid(divUp(c.cols, threads.x), divUp(c.rows, threads.y));
+
+ mulSpectrumsKernel_CONJ<<<grid, threads, 0, stream>>>(a, b, c);
+ cudaSafeCall( cudaGetLastError() );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+
+
+ //////////////////////////////////////////////////////////////////////////
+ // mulAndScaleSpectrums
+
+ __global__ void mulAndScaleSpectrumsKernel(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, float scale, PtrStepSz<cufftComplex> c)
+ {
+ const int x = blockIdx.x * blockDim.x + threadIdx.x;
+ const int y = blockIdx.y * blockDim.y + threadIdx.y;
+
+ if (x < c.cols && y < c.rows)
+ {
+ cufftComplex v = cuCmulf(a.ptr(y)[x], b.ptr(y)[x]);
+ c.ptr(y)[x] = make_cuFloatComplex(cuCrealf(v) * scale, cuCimagf(v) * scale);
+ }
+ }
+
+
+ void mulAndScaleSpectrums(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, float scale, PtrStepSz<cufftComplex> c, cudaStream_t stream)
+ {
+ dim3 threads(256);
+ dim3 grid(divUp(c.cols, threads.x), divUp(c.rows, threads.y));
+
+ mulAndScaleSpectrumsKernel<<<grid, threads, 0, stream>>>(a, b, scale, c);
+ cudaSafeCall( cudaGetLastError() );
+
+ if (stream)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+
+
+ //////////////////////////////////////////////////////////////////////////
+ // mulAndScaleSpectrums_CONJ
+
+ __global__ void mulAndScaleSpectrumsKernel_CONJ(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, float scale, PtrStepSz<cufftComplex> c)
+ {
+ const int x = blockIdx.x * blockDim.x + threadIdx.x;
+ const int y = blockIdx.y * blockDim.y + threadIdx.y;
+
+ if (x < c.cols && y < c.rows)
+ {
+ cufftComplex v = cuCmulf(a.ptr(y)[x], cuConjf(b.ptr(y)[x]));
+ c.ptr(y)[x] = make_cuFloatComplex(cuCrealf(v) * scale, cuCimagf(v) * scale);
+ }
+ }
+
+
+ void mulAndScaleSpectrums_CONJ(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, float scale, PtrStepSz<cufftComplex> c, cudaStream_t stream)
+ {
+ dim3 threads(256);
+ dim3 grid(divUp(c.cols, threads.x), divUp(c.rows, threads.y));
+
+ mulAndScaleSpectrumsKernel_CONJ<<<grid, threads, 0, stream>>>(a, b, scale, c);
+ cudaSafeCall( cudaGetLastError() );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+}}} // namespace cv { namespace gpu { namespace cudev
+
+#endif // HAVE_CUFFT
+
+#endif /* CUDA_DISABLER */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/vec_traits.hpp"
+#include "opencv2/core/cuda/vec_math.hpp"
+#include "opencv2/core/cuda/reduce.hpp"
+#include "opencv2/core/cuda/limits.hpp"
+
+#include "unroll_detail.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace reduce
+{
+ struct Sum
+ {
+ template <typename T>
+ __device__ __forceinline__ T startValue() const
+ {
+ return VecTraits<T>::all(0);
+ }
+
+ template <typename T>
+ __device__ __forceinline__ T operator ()(T a, T b) const
+ {
+ return a + b;
+ }
+
+ template <typename T>
+ __device__ __forceinline__ T result(T r, double) const
+ {
+ return r;
+ }
+
+ __device__ __forceinline__ Sum() {}
+ __device__ __forceinline__ Sum(const Sum&) {}
+ };
+
+ struct Avg
+ {
+ template <typename T>
+ __device__ __forceinline__ T startValue() const
+ {
+ return VecTraits<T>::all(0);
+ }
+
+ template <typename T>
+ __device__ __forceinline__ T operator ()(T a, T b) const
+ {
+ return a + b;
+ }
+
+ template <typename T>
+ __device__ __forceinline__ typename TypeVec<double, VecTraits<T>::cn>::vec_type result(T r, double sz) const
+ {
+ return r / sz;
+ }
+
+ __device__ __forceinline__ Avg() {}
+ __device__ __forceinline__ Avg(const Avg&) {}
+ };
+
+ struct Min
+ {
+ template <typename T>
+ __device__ __forceinline__ T startValue() const
+ {
+ return VecTraits<T>::all(numeric_limits<typename VecTraits<T>::elem_type>::max());
+ }
+
+ template <typename T>
+ __device__ __forceinline__ T operator ()(T a, T b) const
+ {
+ minimum<T> minOp;
+ return minOp(a, b);
+ }
+
+ template <typename T>
+ __device__ __forceinline__ T result(T r, double) const
+ {
+ return r;
+ }
+
+ __device__ __forceinline__ Min() {}
+ __device__ __forceinline__ Min(const Min&) {}
+ };
+
+ struct Max
+ {
+ template <typename T>
+ __device__ __forceinline__ T startValue() const
+ {
+ return VecTraits<T>::all(-numeric_limits<typename VecTraits<T>::elem_type>::max());
+ }
+
+ template <typename T>
+ __device__ __forceinline__ T operator ()(T a, T b) const
+ {
+ maximum<T> maxOp;
+ return maxOp(a, b);
+ }
+
+ template <typename T>
+ __device__ __forceinline__ T result(T r, double) const
+ {
+ return r;
+ }
+
+ __device__ __forceinline__ Max() {}
+ __device__ __forceinline__ Max(const Max&) {}
+ };
+
+ ///////////////////////////////////////////////////////////
+
+ template <typename T, typename S, typename D, class Op>
+ __global__ void rowsKernel(const PtrStepSz<T> src, D* dst, const Op op)
+ {
+ __shared__ S smem[16 * 16];
+
+ const int x = blockIdx.x * 16 + threadIdx.x;
+
+ S myVal = op.template startValue<S>();
+
+ if (x < src.cols)
+ {
+ for (int y = threadIdx.y; y < src.rows; y += 16)
+ {
+ S srcVal = src(y, x);
+ myVal = op(myVal, srcVal);
+ }
+ }
+
+ smem[threadIdx.x * 16 + threadIdx.y] = myVal;
+
+ __syncthreads();
+
+ volatile S* srow = smem + threadIdx.y * 16;
+
+ myVal = srow[threadIdx.x];
+ cudev::reduce<16>(srow, myVal, threadIdx.x, op);
+
+ if (threadIdx.x == 0)
+ srow[0] = myVal;
+
+ __syncthreads();
+
+ if (threadIdx.y == 0 && x < src.cols)
+ dst[x] = (D) op.result(smem[threadIdx.x * 16], src.rows);
+ }
+
+ template <typename T, typename S, typename D, class Op>
+ void rowsCaller(PtrStepSz<T> src, D* dst, cudaStream_t stream)
+ {
+ const dim3 block(16, 16);
+ const dim3 grid(divUp(src.cols, block.x));
+
+ Op op;
+ rowsKernel<T, S, D, Op><<<grid, block, 0, stream>>>(src, dst, op);
+ cudaSafeCall( cudaGetLastError() );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+
+ template <typename T, typename S, typename D>
+ void rows(PtrStepSzb src, void* dst, int op, cudaStream_t stream)
+ {
+ typedef void (*func_t)(PtrStepSz<T> src, D* dst, cudaStream_t stream);
+ static const func_t funcs[] =
+ {
+ rowsCaller<T, S, D, Sum>,
+ rowsCaller<T, S, D, Avg>,
+ rowsCaller<T, S, D, Max>,
+ rowsCaller<T, S, D, Min>
+ };
+
+ funcs[op]((PtrStepSz<T>) src, (D*) dst, stream);
+ }
+
+ template void rows<unsigned char, int, unsigned char>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
+ template void rows<unsigned char, int, int>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
+ template void rows<unsigned char, float, float>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
+ template void rows<unsigned char, double, double>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
+
+ template void rows<unsigned short, int, unsigned short>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
+ template void rows<unsigned short, int, int>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
+ template void rows<unsigned short, float, float>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
+ template void rows<unsigned short, double, double>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
+
+ template void rows<short, int, short>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
+ template void rows<short, int, int>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
+ template void rows<short, float, float>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
+ template void rows<short, double, double>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
+
+ template void rows<int, int, int>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
+ template void rows<int, float, float>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
+ template void rows<int, double, double>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
+
+ template void rows<float, float, float>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
+ template void rows<float, double, double>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
+
+ template void rows<double, double, double>(PtrStepSzb src, void* dst, int op, cudaStream_t stream);
+
+ ///////////////////////////////////////////////////////////
+
+ template <int BLOCK_SIZE, typename T, typename S, typename D, int cn, class Op>
+ __global__ void colsKernel(const PtrStepSz<typename TypeVec<T, cn>::vec_type> src, typename TypeVec<D, cn>::vec_type* dst, const Op op)
+ {
+ typedef typename TypeVec<T, cn>::vec_type src_type;
+ typedef typename TypeVec<S, cn>::vec_type work_type;
+ typedef typename TypeVec<D, cn>::vec_type dst_type;
+
+ __shared__ S smem[BLOCK_SIZE * cn];
+
+ const int y = blockIdx.x;
+
+ const src_type* srcRow = src.ptr(y);
+
+ work_type myVal = op.template startValue<work_type>();
+
+ for (int x = threadIdx.x; x < src.cols; x += BLOCK_SIZE)
+ myVal = op(myVal, saturate_cast<work_type>(srcRow[x]));
+
+ cudev::reduce<BLOCK_SIZE>(detail::Unroll<cn>::template smem_tuple<BLOCK_SIZE>(smem), detail::Unroll<cn>::tie(myVal), threadIdx.x, detail::Unroll<cn>::op(op));
+
+ if (threadIdx.x == 0)
+ dst[y] = saturate_cast<dst_type>(op.result(myVal, src.cols));
+ }
+
+ template <typename T, typename S, typename D, int cn, class Op> void colsCaller(PtrStepSzb src, void* dst, cudaStream_t stream)
+ {
+ const int BLOCK_SIZE = 256;
+
+ const dim3 block(BLOCK_SIZE);
+ const dim3 grid(src.rows);
+
+ Op op;
+ colsKernel<BLOCK_SIZE, T, S, D, cn, Op><<<grid, block, 0, stream>>>((PtrStepSz<typename TypeVec<T, cn>::vec_type>) src, (typename TypeVec<D, cn>::vec_type*) dst, op);
+ cudaSafeCall( cudaGetLastError() );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+
+ }
+
+ template <typename T, typename S, typename D> void cols(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream)
+ {
+ typedef void (*func_t)(PtrStepSzb src, void* dst, cudaStream_t stream);
+ static const func_t funcs[5][4] =
+ {
+ {0,0,0,0},
+ {colsCaller<T, S, D, 1, Sum>, colsCaller<T, S, D, 1, Avg>, colsCaller<T, S, D, 1, Max>, colsCaller<T, S, D, 1, Min>},
+ {colsCaller<T, S, D, 2, Sum>, colsCaller<T, S, D, 2, Avg>, colsCaller<T, S, D, 2, Max>, colsCaller<T, S, D, 2, Min>},
+ {colsCaller<T, S, D, 3, Sum>, colsCaller<T, S, D, 3, Avg>, colsCaller<T, S, D, 3, Max>, colsCaller<T, S, D, 3, Min>},
+ {colsCaller<T, S, D, 4, Sum>, colsCaller<T, S, D, 4, Avg>, colsCaller<T, S, D, 4, Max>, colsCaller<T, S, D, 4, Min>},
+ };
+
+ funcs[cn][op](src, dst, stream);
+ }
+
+ template void cols<unsigned char, int, unsigned char>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
+ template void cols<unsigned char, int, int>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
+ template void cols<unsigned char, float, float>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
+ template void cols<unsigned char, double, double>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
+
+ template void cols<unsigned short, int, unsigned short>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
+ template void cols<unsigned short, int, int>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
+ template void cols<unsigned short, float, float>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
+ template void cols<unsigned short, double, double>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
+
+ template void cols<short, int, short>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
+ template void cols<short, int, int>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
+ template void cols<short, float, float>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
+ template void cols<short, double, double>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
+
+ template void cols<int, int, int>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
+ template void cols<int, float, float>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
+ template void cols<int, double, double>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
+
+ template void cols<float, float, float>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
+ template void cols<float, double, double>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
+
+ template void cols<double, double, double>(PtrStepSzb src, void* dst, int cn, int op, cudaStream_t stream);
+}
+
+#endif /* CUDA_DISABLER */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/simd_functions.hpp"
+
+#include "arithm_func_traits.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace arithm
+{
+ struct VSub4 : binary_function<uint, uint, uint>
+ {
+ __device__ __forceinline__ uint operator ()(uint a, uint b) const
+ {
+ return vsub4(a, b);
+ }
+
+ __device__ __forceinline__ VSub4() {}
+ __device__ __forceinline__ VSub4(const VSub4& other) {}
+ };
+
+ struct VSub2 : binary_function<uint, uint, uint>
+ {
+ __device__ __forceinline__ uint operator ()(uint a, uint b) const
+ {
+ return vsub2(a, b);
+ }
+
+ __device__ __forceinline__ VSub2() {}
+ __device__ __forceinline__ VSub2(const VSub2& other) {}
+ };
+
+ template <typename T, typename D> struct SubMat : binary_function<T, T, D>
+ {
+ __device__ __forceinline__ D operator ()(T a, T b) const
+ {
+ return saturate_cast<D>(a - b);
+ }
+
+ __device__ __forceinline__ SubMat() {}
+ __device__ __forceinline__ SubMat(const SubMat& other) {}
+ };
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <> struct TransformFunctorTraits< arithm::VSub4 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
+ {
+ };
+
+ template <> struct TransformFunctorTraits< arithm::VSub2 > : arithm::ArithmFuncTraits<sizeof(uint), sizeof(uint)>
+ {
+ };
+
+ template <typename T, typename D> struct TransformFunctorTraits< arithm::SubMat<T, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ void subMat_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
+ {
+ cudev::transform(src1, src2, dst, VSub4(), WithOutMask(), stream);
+ }
+
+ void subMat_v2(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream)
+ {
+ cudev::transform(src1, src2, dst, VSub2(), WithOutMask(), stream);
+ }
+
+ template <typename T, typename D>
+ void subMat(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream)
+ {
+ if (mask.data)
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<D>) dst, SubMat<T, D>(), mask, stream);
+ else
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<T>) src2, (PtrStepSz<D>) dst, SubMat<T, D>(), WithOutMask(), stream);
+ }
+
+ template void subMat<uchar, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<uchar, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<uchar, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<uchar, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<uchar, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<uchar, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<uchar, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ template void subMat<schar, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<schar, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<schar, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<schar, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<schar, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<schar, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<schar, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void subMat<ushort, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subMat<ushort, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<ushort, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<ushort, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<ushort, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<ushort, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<ushort, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void subMat<short, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subMat<short, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<short, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<short, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<short, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<short, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<short, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void subMat<int, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subMat<int, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subMat<int, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subMat<int, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<int, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<int, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<int, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void subMat<float, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subMat<float, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subMat<float, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subMat<float, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subMat<float, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<float, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<float, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void subMat<double, uchar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subMat<double, schar>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subMat<double, ushort>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subMat<double, short>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subMat<double, int>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subMat<double, float>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subMat<double, double>(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/simd_functions.hpp"
+
+#include "arithm_func_traits.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace arithm
+{
+ template <typename T, typename S, typename D> struct SubScalar : unary_function<T, D>
+ {
+ S val;
+
+ explicit SubScalar(S val_) : val(val_) {}
+
+ __device__ __forceinline__ D operator ()(T a) const
+ {
+ return saturate_cast<D>(a - val);
+ }
+ };
+}
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <typename T, typename S, typename D> struct TransformFunctorTraits< arithm::SubScalar<T, S, D> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(D)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ template <typename T, typename S, typename D>
+ void subScalar(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream)
+ {
+ SubScalar<T, S, D> op(static_cast<S>(val));
+
+ if (mask.data)
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<D>) dst, op, mask, stream);
+ else
+ cudev::transform((PtrStepSz<T>) src1, (PtrStepSz<D>) dst, op, WithOutMask(), stream);
+ }
+
+ template void subScalar<uchar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<uchar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<uchar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<uchar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<uchar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<uchar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<uchar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ template void subScalar<schar, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<schar, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<schar, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<schar, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<schar, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<schar, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<schar, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void subScalar<ushort, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subScalar<ushort, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<ushort, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<ushort, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<ushort, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<ushort, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<ushort, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void subScalar<short, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subScalar<short, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<short, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<short, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<short, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<short, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<short, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void subScalar<int, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subScalar<int, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subScalar<int, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subScalar<int, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<int, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<int, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<int, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void subScalar<float, float, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subScalar<float, float, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subScalar<float, float, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subScalar<float, float, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subScalar<float, float, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<float, float, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<float, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+
+ //template void subScalar<double, double, uchar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subScalar<double, double, schar>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subScalar<double, double, ushort>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subScalar<double, double, short>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subScalar<double, double, int>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ //template void subScalar<double, double, float>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+ template void subScalar<double, double, double>(PtrStepSzb src1, double val, PtrStepSzb dst, PtrStepb mask, cudaStream_t stream);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/vec_traits.hpp"
+#include "opencv2/core/cuda/vec_math.hpp"
+#include "opencv2/core/cuda/reduce.hpp"
+#include "opencv2/core/cuda/emulation.hpp"
+#include "opencv2/core/cuda/utility.hpp"
+
+#include "unroll_detail.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace sum
+{
+ __device__ unsigned int blocks_finished = 0;
+
+ template <typename R, int cn> struct AtomicAdd;
+ template <typename R> struct AtomicAdd<R, 1>
+ {
+ static __device__ void run(R* ptr, R val)
+ {
+ Emulation::glob::atomicAdd(ptr, val);
+ }
+ };
+ template <typename R> struct AtomicAdd<R, 2>
+ {
+ typedef typename TypeVec<R, 2>::vec_type val_type;
+
+ static __device__ void run(R* ptr, val_type val)
+ {
+ Emulation::glob::atomicAdd(ptr, val.x);
+ Emulation::glob::atomicAdd(ptr + 1, val.y);
+ }
+ };
+ template <typename R> struct AtomicAdd<R, 3>
+ {
+ typedef typename TypeVec<R, 3>::vec_type val_type;
+
+ static __device__ void run(R* ptr, val_type val)
+ {
+ Emulation::glob::atomicAdd(ptr, val.x);
+ Emulation::glob::atomicAdd(ptr + 1, val.y);
+ Emulation::glob::atomicAdd(ptr + 2, val.z);
+ }
+ };
+ template <typename R> struct AtomicAdd<R, 4>
+ {
+ typedef typename TypeVec<R, 4>::vec_type val_type;
+
+ static __device__ void run(R* ptr, val_type val)
+ {
+ Emulation::glob::atomicAdd(ptr, val.x);
+ Emulation::glob::atomicAdd(ptr + 1, val.y);
+ Emulation::glob::atomicAdd(ptr + 2, val.z);
+ Emulation::glob::atomicAdd(ptr + 3, val.w);
+ }
+ };
+
+ template <int BLOCK_SIZE, typename R, int cn>
+ struct GlobalReduce
+ {
+ typedef typename TypeVec<R, cn>::vec_type result_type;
+
+ static __device__ void run(result_type& sum, result_type* result, int tid, int bid, R* smem)
+ {
+ #if __CUDA_ARCH__ >= 200
+ if (tid == 0)
+ AtomicAdd<R, cn>::run((R*) result, sum);
+ #else
+ __shared__ bool is_last;
+
+ if (tid == 0)
+ {
+ result[bid] = sum;
+
+ __threadfence();
+
+ unsigned int ticket = ::atomicAdd(&blocks_finished, 1);
+ is_last = (ticket == gridDim.x * gridDim.y - 1);
+ }
+
+ __syncthreads();
+
+ if (is_last)
+ {
+ sum = tid < gridDim.x * gridDim.y ? result[tid] : VecTraits<result_type>::all(0);
+
+ cudev::reduce<BLOCK_SIZE>(detail::Unroll<cn>::template smem_tuple<BLOCK_SIZE>(smem), detail::Unroll<cn>::tie(sum), tid, detail::Unroll<cn>::op(plus<R>()));
+
+ if (tid == 0)
+ {
+ result[0] = sum;
+ blocks_finished = 0;
+ }
+ }
+ #endif
+ }
+ };
+
+ template <int BLOCK_SIZE, typename src_type, typename result_type, class Mask, class Op>
+ __global__ void kernel(const PtrStepSz<src_type> src, result_type* result, const Mask mask, const Op op, const int twidth, const int theight)
+ {
+ typedef typename VecTraits<src_type>::elem_type T;
+ typedef typename VecTraits<result_type>::elem_type R;
+ const int cn = VecTraits<src_type>::cn;
+
+ __shared__ R smem[BLOCK_SIZE * cn];
+
+ const int x0 = blockIdx.x * blockDim.x * twidth + threadIdx.x;
+ const int y0 = blockIdx.y * blockDim.y * theight + threadIdx.y;
+
+ const int tid = threadIdx.y * blockDim.x + threadIdx.x;
+ const int bid = blockIdx.y * gridDim.x + blockIdx.x;
+
+ result_type sum = VecTraits<result_type>::all(0);
+
+ for (int i = 0, y = y0; i < theight && y < src.rows; ++i, y += blockDim.y)
+ {
+ const src_type* ptr = src.ptr(y);
+
+ for (int j = 0, x = x0; j < twidth && x < src.cols; ++j, x += blockDim.x)
+ {
+ if (mask(y, x))
+ {
+ const src_type srcVal = ptr[x];
+ sum = sum + op(saturate_cast<result_type>(srcVal));
+ }
+ }
+ }
+
+ cudev::reduce<BLOCK_SIZE>(detail::Unroll<cn>::template smem_tuple<BLOCK_SIZE>(smem), detail::Unroll<cn>::tie(sum), tid, detail::Unroll<cn>::op(plus<R>()));
+
+ GlobalReduce<BLOCK_SIZE, R, cn>::run(sum, result, tid, bid, smem);
+ }
+
+ const int threads_x = 32;
+ const int threads_y = 8;
+
+ void getLaunchCfg(int cols, int rows, dim3& block, dim3& grid)
+ {
+ block = dim3(threads_x, threads_y);
+
+ grid = dim3(divUp(cols, block.x * block.y),
+ divUp(rows, block.y * block.x));
+
+ grid.x = ::min(grid.x, block.x);
+ grid.y = ::min(grid.y, block.y);
+ }
+
+ void getBufSize(int cols, int rows, int cn, int& bufcols, int& bufrows)
+ {
+ dim3 block, grid;
+ getLaunchCfg(cols, rows, block, grid);
+
+ bufcols = grid.x * grid.y * sizeof(double) * cn;
+ bufrows = 1;
+ }
+
+ template <typename T, typename R, int cn, template <typename> class Op>
+ void caller(PtrStepSzb src_, void* buf_, double* out, PtrStepSzb mask)
+ {
+ typedef typename TypeVec<T, cn>::vec_type src_type;
+ typedef typename TypeVec<R, cn>::vec_type result_type;
+
+ PtrStepSz<src_type> src(src_);
+ result_type* buf = (result_type*) buf_;
+
+ dim3 block, grid;
+ getLaunchCfg(src.cols, src.rows, block, grid);
+
+ const int twidth = divUp(divUp(src.cols, grid.x), block.x);
+ const int theight = divUp(divUp(src.rows, grid.y), block.y);
+
+ Op<result_type> op;
+
+ if (mask.data)
+ kernel<threads_x * threads_y><<<grid, block>>>(src, buf, SingleMask(mask), op, twidth, theight);
+ else
+ kernel<threads_x * threads_y><<<grid, block>>>(src, buf, WithOutMask(), op, twidth, theight);
+ cudaSafeCall( cudaGetLastError() );
+
+ cudaSafeCall( cudaDeviceSynchronize() );
+
+ R result[4] = {0, 0, 0, 0};
+ cudaSafeCall( cudaMemcpy(&result, buf, sizeof(result_type), cudaMemcpyDeviceToHost) );
+
+ out[0] = result[0];
+ out[1] = result[1];
+ out[2] = result[2];
+ out[3] = result[3];
+ }
+
+ template <typename T> struct SumType;
+ template <> struct SumType<uchar> { typedef unsigned int R; };
+ template <> struct SumType<schar> { typedef int R; };
+ template <> struct SumType<ushort> { typedef unsigned int R; };
+ template <> struct SumType<short> { typedef int R; };
+ template <> struct SumType<int> { typedef int R; };
+ template <> struct SumType<float> { typedef float R; };
+ template <> struct SumType<double> { typedef double R; };
+
+ template <typename T, int cn>
+ void run(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask)
+ {
+ typedef typename SumType<T>::R R;
+ caller<T, R, cn, identity>(src, buf, out, mask);
+ }
+
+ template void run<uchar, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<uchar, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<uchar, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<uchar, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template void run<schar, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<schar, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<schar, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<schar, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template void run<ushort, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<ushort, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<ushort, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<ushort, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template void run<short, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<short, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<short, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<short, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template void run<int, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<int, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<int, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<int, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template void run<float, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<float, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<float, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<float, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template void run<double, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<double, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<double, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void run<double, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template <typename T, int cn>
+ void runAbs(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask)
+ {
+ typedef typename SumType<T>::R R;
+ caller<T, R, cn, abs_func>(src, buf, out, mask);
+ }
+
+ template void runAbs<uchar, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<uchar, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<uchar, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<uchar, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template void runAbs<schar, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<schar, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<schar, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<schar, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template void runAbs<ushort, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<ushort, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<ushort, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<ushort, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template void runAbs<short, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<short, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<short, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<short, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template void runAbs<int, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<int, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<int, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<int, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template void runAbs<float, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<float, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<float, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<float, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template void runAbs<double, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<double, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<double, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runAbs<double, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template <typename T> struct Sqr : unary_function<T, T>
+ {
+ __device__ __forceinline__ T operator ()(T x) const
+ {
+ return x * x;
+ }
+ };
+
+ template <typename T, int cn>
+ void runSqr(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask)
+ {
+ caller<T, double, cn, Sqr>(src, buf, out, mask);
+ }
+
+ template void runSqr<uchar, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<uchar, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<uchar, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<uchar, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template void runSqr<schar, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<schar, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<schar, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<schar, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template void runSqr<ushort, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<ushort, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<ushort, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<ushort, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template void runSqr<short, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<short, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<short, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<short, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template void runSqr<int, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<int, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<int, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<int, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template void runSqr<float, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<float, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<float, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<float, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+
+ template void runSqr<double, 1>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<double, 2>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<double, 3>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+ template void runSqr<double, 4>(PtrStepSzb src, void* buf, double* out, PtrStepSzb mask);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/simd_functions.hpp"
+
+#include "arithm_func_traits.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <typename T> struct TransformFunctorTraits< thresh_binary_func<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+
+ template <typename T> struct TransformFunctorTraits< thresh_binary_inv_func<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+
+ template <typename T> struct TransformFunctorTraits< thresh_trunc_func<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+
+ template <typename T> struct TransformFunctorTraits< thresh_to_zero_func<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+
+ template <typename T> struct TransformFunctorTraits< thresh_to_zero_inv_func<T> > : arithm::ArithmFuncTraits<sizeof(T), sizeof(T)>
+ {
+ };
+}}}
+
+namespace arithm
+{
+ template <template <typename> class Op, typename T>
+ void threshold_caller(PtrStepSz<T> src, PtrStepSz<T> dst, T thresh, T maxVal, cudaStream_t stream)
+ {
+ Op<T> op(thresh, maxVal);
+ cudev::transform(src, dst, op, WithOutMask(), stream);
+ }
+
+ template <typename T>
+ void threshold(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream)
+ {
+ typedef void (*caller_t)(PtrStepSz<T> src, PtrStepSz<T> dst, T thresh, T maxVal, cudaStream_t stream);
+
+ static const caller_t callers[] =
+ {
+ threshold_caller<thresh_binary_func, T>,
+ threshold_caller<thresh_binary_inv_func, T>,
+ threshold_caller<thresh_trunc_func, T>,
+ threshold_caller<thresh_to_zero_func, T>,
+ threshold_caller<thresh_to_zero_inv_func, T>
+ };
+
+ callers[type]((PtrStepSz<T>) src, (PtrStepSz<T>) dst, static_cast<T>(thresh), static_cast<T>(maxVal), stream);
+ }
+
+ template void threshold<uchar>(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream);
+ template void threshold<schar>(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream);
+ template void threshold<ushort>(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream);
+ template void threshold<short>(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream);
+ template void threshold<int>(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream);
+ template void threshold<float>(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream);
+ template void threshold<double>(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace arithm
+{
+ const int TRANSPOSE_TILE_DIM = 16;
+ const int TRANSPOSE_BLOCK_ROWS = 16;
+
+ template <typename T>
+ __global__ void transposeKernel(const PtrStepSz<T> src, PtrStep<T> dst)
+ {
+ __shared__ T tile[TRANSPOSE_TILE_DIM][TRANSPOSE_TILE_DIM + 1];
+
+ int blockIdx_x, blockIdx_y;
+
+ // do diagonal reordering
+ if (gridDim.x == gridDim.y)
+ {
+ blockIdx_y = blockIdx.x;
+ blockIdx_x = (blockIdx.x + blockIdx.y) % gridDim.x;
+ }
+ else
+ {
+ int bid = blockIdx.x + gridDim.x * blockIdx.y;
+ blockIdx_y = bid % gridDim.y;
+ blockIdx_x = ((bid / gridDim.y) + blockIdx_y) % gridDim.x;
+ }
+
+ int xIndex = blockIdx_x * TRANSPOSE_TILE_DIM + threadIdx.x;
+ int yIndex = blockIdx_y * TRANSPOSE_TILE_DIM + threadIdx.y;
+
+ if (xIndex < src.cols)
+ {
+ for (int i = 0; i < TRANSPOSE_TILE_DIM; i += TRANSPOSE_BLOCK_ROWS)
+ {
+ if (yIndex + i < src.rows)
+ {
+ tile[threadIdx.y + i][threadIdx.x] = src(yIndex + i, xIndex);
+ }
+ }
+ }
+
+ __syncthreads();
+
+ xIndex = blockIdx_y * TRANSPOSE_TILE_DIM + threadIdx.x;
+ yIndex = blockIdx_x * TRANSPOSE_TILE_DIM + threadIdx.y;
+
+ if (xIndex < src.rows)
+ {
+ for (int i = 0; i < TRANSPOSE_TILE_DIM; i += TRANSPOSE_BLOCK_ROWS)
+ {
+ if (yIndex + i < src.cols)
+ {
+ dst(yIndex + i, xIndex) = tile[threadIdx.x][threadIdx.y + i];
+ }
+ }
+ }
+ }
+
+ template <typename T> void transpose(PtrStepSz<T> src, PtrStepSz<T> dst, cudaStream_t stream)
+ {
+ const dim3 block(TRANSPOSE_TILE_DIM, TRANSPOSE_TILE_DIM);
+ const dim3 grid(divUp(src.cols, block.x), divUp(src.rows, block.y));
+
+ transposeKernel<<<grid, block, 0, stream>>>(src, dst);
+ cudaSafeCall( cudaGetLastError() );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+
+ template void transpose<int>(PtrStepSz<int> src, PtrStepSz<int> dst, cudaStream_t stream);
+ template void transpose<double>(PtrStepSz<double> src, PtrStepSz<double> dst, cudaStream_t stream);
+}
+
+#endif // CUDA_DISABLER
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __UNROLL_DETAIL_HPP__
+#define __UNROLL_DETAIL_HPP__
+
+#include <thrust/tuple.h>
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/vec_traits.hpp"
+
+namespace detail
+{
+ template <int cn> struct Unroll;
+ template <> struct Unroll<1>
+ {
+ template <int BLOCK_SIZE, typename R>
+ static __device__ __forceinline__ volatile R* smem_tuple(R* smem)
+ {
+ return smem;
+ }
+
+ template <typename R>
+ static __device__ __forceinline__ R& tie(R& val)
+ {
+ return val;
+ }
+
+ template <class Op>
+ static __device__ __forceinline__ const Op& op(const Op& op)
+ {
+ return op;
+ }
+ };
+ template <> struct Unroll<2>
+ {
+ template <int BLOCK_SIZE, typename R>
+ static __device__ __forceinline__ thrust::tuple<volatile R*, volatile R*> smem_tuple(R* smem)
+ {
+ return cv::gpu::cudev::smem_tuple(smem, smem + BLOCK_SIZE);
+ }
+
+ template <typename R>
+ static __device__ __forceinline__ thrust::tuple<typename cv::gpu::cudev::VecTraits<R>::elem_type&, typename cv::gpu::cudev::VecTraits<R>::elem_type&> tie(R& val)
+ {
+ return thrust::tie(val.x, val.y);
+ }
+
+ template <class Op>
+ static __device__ __forceinline__ const thrust::tuple<Op, Op> op(const Op& op)
+ {
+ return thrust::make_tuple(op, op);
+ }
+ };
+ template <> struct Unroll<3>
+ {
+ template <int BLOCK_SIZE, typename R>
+ static __device__ __forceinline__ thrust::tuple<volatile R*, volatile R*, volatile R*> smem_tuple(R* smem)
+ {
+ return cv::gpu::cudev::smem_tuple(smem, smem + BLOCK_SIZE, smem + 2 * BLOCK_SIZE);
+ }
+
+ template <typename R>
+ static __device__ __forceinline__ thrust::tuple<typename cv::gpu::cudev::VecTraits<R>::elem_type&, typename cv::gpu::cudev::VecTraits<R>::elem_type&, typename cv::gpu::cudev::VecTraits<R>::elem_type&> tie(R& val)
+ {
+ return thrust::tie(val.x, val.y, val.z);
+ }
+
+ template <class Op>
+ static __device__ __forceinline__ const thrust::tuple<Op, Op, Op> op(const Op& op)
+ {
+ return thrust::make_tuple(op, op, op);
+ }
+ };
+ template <> struct Unroll<4>
+ {
+ template <int BLOCK_SIZE, typename R>
+ static __device__ __forceinline__ thrust::tuple<volatile R*, volatile R*, volatile R*, volatile R*> smem_tuple(R* smem)
+ {
+ return cv::gpu::cudev::smem_tuple(smem, smem + BLOCK_SIZE, smem + 2 * BLOCK_SIZE, smem + 3 * BLOCK_SIZE);
+ }
+
+ template <typename R>
+ static __device__ __forceinline__ thrust::tuple<typename cv::gpu::cudev::VecTraits<R>::elem_type&, typename cv::gpu::cudev::VecTraits<R>::elem_type&, typename cv::gpu::cudev::VecTraits<R>::elem_type&, typename cv::gpu::cudev::VecTraits<R>::elem_type&> tie(R& val)
+ {
+ return thrust::tie(val.x, val.y, val.z, val.w);
+ }
+
+ template <class Op>
+ static __device__ __forceinline__ const thrust::tuple<Op, Op, Op, Op> op(const Op& op)
+ {
+ return thrust::make_tuple(op, op, op, op);
+ }
+ };
+}
+
+#endif // __UNROLL_DETAIL_HPP__
void cv::gpu::add(const GpuMat&, const GpuMat&, GpuMat&, const GpuMat&, int, Stream&) { throw_no_cuda(); }
void cv::gpu::add(const GpuMat&, const Scalar&, GpuMat&, const GpuMat&, int, Stream&) { throw_no_cuda(); }
+
void cv::gpu::subtract(const GpuMat&, const GpuMat&, GpuMat&, const GpuMat&, int, Stream&) { throw_no_cuda(); }
void cv::gpu::subtract(const GpuMat&, const Scalar&, GpuMat&, const GpuMat&, int, Stream&) { throw_no_cuda(); }
+
void cv::gpu::multiply(const GpuMat&, const GpuMat&, GpuMat&, double, int, Stream&) { throw_no_cuda(); }
void cv::gpu::multiply(const GpuMat&, const Scalar&, GpuMat&, double, int, Stream&) { throw_no_cuda(); }
+
void cv::gpu::divide(const GpuMat&, const GpuMat&, GpuMat&, double, int, Stream&) { throw_no_cuda(); }
void cv::gpu::divide(const GpuMat&, const Scalar&, GpuMat&, double, int, Stream&) { throw_no_cuda(); }
void cv::gpu::divide(double, const GpuMat&, GpuMat&, int, Stream&) { throw_no_cuda(); }
+
void cv::gpu::absdiff(const GpuMat&, const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
void cv::gpu::absdiff(const GpuMat&, const Scalar&, GpuMat&, Stream&) { throw_no_cuda(); }
+
void cv::gpu::abs(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+
void cv::gpu::sqr(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+
void cv::gpu::sqrt(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+
void cv::gpu::exp(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+
void cv::gpu::log(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+
+void cv::gpu::pow(const GpuMat&, double, GpuMat&, Stream&) { throw_no_cuda(); }
+
void cv::gpu::compare(const GpuMat&, const GpuMat&, GpuMat&, int, Stream&) { throw_no_cuda(); }
void cv::gpu::compare(const GpuMat&, Scalar, GpuMat&, int, Stream&) { throw_no_cuda(); }
+
void cv::gpu::bitwise_not(const GpuMat&, GpuMat&, const GpuMat&, Stream&) { throw_no_cuda(); }
+
void cv::gpu::bitwise_or(const GpuMat&, const GpuMat&, GpuMat&, const GpuMat&, Stream&) { throw_no_cuda(); }
void cv::gpu::bitwise_or(const GpuMat&, const Scalar&, GpuMat&, Stream&) { throw_no_cuda(); }
+
void cv::gpu::bitwise_and(const GpuMat&, const GpuMat&, GpuMat&, const GpuMat&, Stream&) { throw_no_cuda(); }
void cv::gpu::bitwise_and(const GpuMat&, const Scalar&, GpuMat&, Stream&) { throw_no_cuda(); }
+
void cv::gpu::bitwise_xor(const GpuMat&, const GpuMat&, GpuMat&, const GpuMat&, Stream&) { throw_no_cuda(); }
void cv::gpu::bitwise_xor(const GpuMat&, const Scalar&, GpuMat&, Stream&) { throw_no_cuda(); }
+
void cv::gpu::rshift(const GpuMat&, Scalar_<int>, GpuMat&, Stream&) { throw_no_cuda(); }
+
void cv::gpu::lshift(const GpuMat&, Scalar_<int>, GpuMat&, Stream&) { throw_no_cuda(); }
+
void cv::gpu::min(const GpuMat&, const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
void cv::gpu::min(const GpuMat&, double, GpuMat&, Stream&) { throw_no_cuda(); }
+
void cv::gpu::max(const GpuMat&, const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
void cv::gpu::max(const GpuMat&, double, GpuMat&, Stream&) { throw_no_cuda(); }
-double cv::gpu::threshold(const GpuMat&, GpuMat&, double, double, int, Stream&) {throw_no_cuda(); return 0.0;}
-void cv::gpu::pow(const GpuMat&, double, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::alphaComp(const GpuMat&, const GpuMat&, GpuMat&, int, Stream&) { throw_no_cuda(); }
+
void cv::gpu::addWeighted(const GpuMat&, double, const GpuMat&, double, double, GpuMat&, int, Stream&) { throw_no_cuda(); }
+double cv::gpu::threshold(const GpuMat&, GpuMat&, double, double, int, Stream&) {throw_no_cuda(); return 0.0;}
+
+void cv::gpu::magnitude(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+void cv::gpu::magnitude(const GpuMat&, const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+
+void cv::gpu::magnitudeSqr(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+void cv::gpu::magnitudeSqr(const GpuMat&, const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+
+void cv::gpu::phase(const GpuMat&, const GpuMat&, GpuMat&, bool, Stream&) { throw_no_cuda(); }
+
+void cv::gpu::cartToPolar(const GpuMat&, const GpuMat&, GpuMat&, GpuMat&, bool, Stream&) { throw_no_cuda(); }
+
+void cv::gpu::polarToCart(const GpuMat&, const GpuMat&, GpuMat&, GpuMat&, bool, Stream&) { throw_no_cuda(); }
+
#else
////////////////////////////////////////////////////////////////////////
namespace arithm
{
- void cmpMatEq_v4(PtrStepSz<unsigned int> src1, PtrStepSz<unsigned int> src2, PtrStepSz<unsigned int> dst, cudaStream_t stream);
- void cmpMatNe_v4(PtrStepSz<unsigned int> src1, PtrStepSz<unsigned int> src2, PtrStepSz<unsigned int> dst, cudaStream_t stream);
- void cmpMatLt_v4(PtrStepSz<unsigned int> src1, PtrStepSz<unsigned int> src2, PtrStepSz<unsigned int> dst, cudaStream_t stream);
- void cmpMatLe_v4(PtrStepSz<unsigned int> src1, PtrStepSz<unsigned int> src2, PtrStepSz<unsigned int> dst, cudaStream_t stream);
+ void cmpMatEq_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream);
+ void cmpMatNe_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream);
+ void cmpMatLt_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream);
+ void cmpMatLe_v4(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream);
template <typename T> void cmpMatEq(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
template <typename T> void cmpMatNe(PtrStepSzb src1, PtrStepSzb src2, PtrStepSzb dst, cudaStream_t stream);
{cmpMatEq<double> , cmpMatNe<double> , cmpMatLt<double> , cmpMatLe<double> }
};
- typedef void (*func_v4_t)(PtrStepSz<unsigned int> src1, PtrStepSz<unsigned int> src2, PtrStepSz<unsigned int> dst, cudaStream_t stream);
+ typedef void (*func_v4_t)(PtrStepSz<uint> src1, PtrStepSz<uint> src2, PtrStepSz<uint> dst, cudaStream_t stream);
static const func_v4_t funcs_v4[] =
{
cmpMatEq_v4, cmpMatNe_v4, cmpMatLt_v4, cmpMatLe_v4
}
////////////////////////////////////////////////////////////////////////
-// threshold
-
-namespace arithm
-{
- template <typename T>
- void threshold(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream);
-}
-
-double cv::gpu::threshold(const GpuMat& src, GpuMat& dst, double thresh, double maxVal, int type, Stream& s)
-{
- const int depth = src.depth();
-
- CV_Assert( src.channels() == 1 && depth <= CV_64F );
- CV_Assert( type <= THRESH_TOZERO_INV );
-
- if (depth == CV_64F)
- {
- if (!deviceSupports(NATIVE_DOUBLE))
- CV_Error(cv::Error::StsUnsupportedFormat, "The device doesn't support double");
- }
-
- dst.create(src.size(), src.type());
-
- cudaStream_t stream = StreamAccessor::getStream(s);
-
- if (src.type() == CV_32FC1 && type == THRESH_TRUNC)
- {
- NppStreamHandler h(stream);
-
- NppiSize sz;
- sz.width = src.cols;
- sz.height = src.rows;
-
- nppSafeCall( nppiThreshold_32f_C1R(src.ptr<Npp32f>(), static_cast<int>(src.step),
- dst.ptr<Npp32f>(), static_cast<int>(dst.step), sz, static_cast<Npp32f>(thresh), NPP_CMP_GREATER) );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
- else
- {
- typedef void (*func_t)(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream);
- static const func_t funcs[] =
- {
- arithm::threshold<unsigned char>,
- arithm::threshold<signed char>,
- arithm::threshold<unsigned short>,
- arithm::threshold<short>,
- arithm::threshold<int>,
- arithm::threshold<float>,
- arithm::threshold<double>
- };
-
- if (depth != CV_32F && depth != CV_64F)
- {
- thresh = cvFloor(thresh);
- maxVal = cvRound(maxVal);
- }
-
- funcs[depth](src, dst, thresh, maxVal, type, stream);
- }
-
- return thresh;
-}
-
-////////////////////////////////////////////////////////////////////////
// pow
namespace arithm
}
////////////////////////////////////////////////////////////////////////
-// alphaComp
-
-namespace
-{
- template <int DEPTH> struct NppAlphaCompFunc
- {
- typedef typename NppTypeTraits<DEPTH>::npp_t npp_t;
-
- typedef NppStatus (*func_t)(const npp_t* pSrc1, int nSrc1Step, const npp_t* pSrc2, int nSrc2Step, npp_t* pDst, int nDstStep, NppiSize oSizeROI, NppiAlphaOp eAlphaOp);
- };
-
- template <int DEPTH, typename NppAlphaCompFunc<DEPTH>::func_t func> struct NppAlphaComp
- {
- typedef typename NppTypeTraits<DEPTH>::npp_t npp_t;
-
- static void call(const GpuMat& img1, const GpuMat& img2, GpuMat& dst, NppiAlphaOp eAlphaOp, cudaStream_t stream)
- {
- NppStreamHandler h(stream);
-
- NppiSize oSizeROI;
- oSizeROI.width = img1.cols;
- oSizeROI.height = img2.rows;
-
- nppSafeCall( func(img1.ptr<npp_t>(), static_cast<int>(img1.step), img2.ptr<npp_t>(), static_cast<int>(img2.step),
- dst.ptr<npp_t>(), static_cast<int>(dst.step), oSizeROI, eAlphaOp) );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
- };
-}
-
-void cv::gpu::alphaComp(const GpuMat& img1, const GpuMat& img2, GpuMat& dst, int alpha_op, Stream& stream)
-{
- static const NppiAlphaOp npp_alpha_ops[] = {
- NPPI_OP_ALPHA_OVER,
- NPPI_OP_ALPHA_IN,
- NPPI_OP_ALPHA_OUT,
- NPPI_OP_ALPHA_ATOP,
- NPPI_OP_ALPHA_XOR,
- NPPI_OP_ALPHA_PLUS,
- NPPI_OP_ALPHA_OVER_PREMUL,
- NPPI_OP_ALPHA_IN_PREMUL,
- NPPI_OP_ALPHA_OUT_PREMUL,
- NPPI_OP_ALPHA_ATOP_PREMUL,
- NPPI_OP_ALPHA_XOR_PREMUL,
- NPPI_OP_ALPHA_PLUS_PREMUL,
- NPPI_OP_ALPHA_PREMUL
- };
-
- typedef void (*func_t)(const GpuMat& img1, const GpuMat& img2, GpuMat& dst, NppiAlphaOp eAlphaOp, cudaStream_t stream);
-
- static const func_t funcs[] =
- {
- NppAlphaComp<CV_8U, nppiAlphaComp_8u_AC4R>::call,
- 0,
- NppAlphaComp<CV_16U, nppiAlphaComp_16u_AC4R>::call,
- 0,
- NppAlphaComp<CV_32S, nppiAlphaComp_32s_AC4R>::call,
- NppAlphaComp<CV_32F, nppiAlphaComp_32f_AC4R>::call
- };
-
- CV_Assert( img1.type() == CV_8UC4 || img1.type() == CV_16UC4 || img1.type() == CV_32SC4 || img1.type() == CV_32FC4 );
- CV_Assert( img1.size() == img2.size() && img1.type() == img2.type() );
-
- dst.create(img1.size(), img1.type());
-
- const func_t func = funcs[img1.depth()];
-
- func(img1, img2, dst, npp_alpha_ops[alpha_op], StreamAccessor::getStream(stream));
-}
-
-////////////////////////////////////////////////////////////////////////
// addWeighted
namespace arithm
func(src1_, alpha, src2_, beta, gamma, dst_, StreamAccessor::getStream(stream));
}
+////////////////////////////////////////////////////////////////////////
+// threshold
+
+namespace arithm
+{
+ template <typename T>
+ void threshold(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream);
+}
+
+double cv::gpu::threshold(const GpuMat& src, GpuMat& dst, double thresh, double maxVal, int type, Stream& s)
+{
+ const int depth = src.depth();
+
+ CV_Assert( src.channels() == 1 && depth <= CV_64F );
+ CV_Assert( type <= 4/*THRESH_TOZERO_INV*/ );
+
+ if (depth == CV_64F)
+ {
+ if (!deviceSupports(NATIVE_DOUBLE))
+ CV_Error(cv::Error::StsUnsupportedFormat, "The device doesn't support double");
+ }
+
+ dst.create(src.size(), src.type());
+
+ cudaStream_t stream = StreamAccessor::getStream(s);
+
+ if (src.type() == CV_32FC1 && type == 2/*THRESH_TRUNC*/)
+ {
+ NppStreamHandler h(stream);
+
+ NppiSize sz;
+ sz.width = src.cols;
+ sz.height = src.rows;
+
+ nppSafeCall( nppiThreshold_32f_C1R(src.ptr<Npp32f>(), static_cast<int>(src.step),
+ dst.ptr<Npp32f>(), static_cast<int>(dst.step), sz, static_cast<Npp32f>(thresh), NPP_CMP_GREATER) );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+ else
+ {
+ typedef void (*func_t)(PtrStepSzb src, PtrStepSzb dst, double thresh, double maxVal, int type, cudaStream_t stream);
+ static const func_t funcs[] =
+ {
+ arithm::threshold<unsigned char>,
+ arithm::threshold<signed char>,
+ arithm::threshold<unsigned short>,
+ arithm::threshold<short>,
+ arithm::threshold<int>,
+ arithm::threshold<float>,
+ arithm::threshold<double>
+ };
+
+ if (depth != CV_32F && depth != CV_64F)
+ {
+ thresh = cvFloor(thresh);
+ maxVal = cvRound(maxVal);
+ }
+
+ funcs[depth](src, dst, thresh, maxVal, type, stream);
+ }
+
+ return thresh;
+}
+
+////////////////////////////////////////////////////////////////////////
+// NPP magnitide
+
+namespace
+{
+ typedef NppStatus (*nppMagnitude_t)(const Npp32fc* pSrc, int nSrcStep, Npp32f* pDst, int nDstStep, NppiSize oSizeROI);
+
+ inline void npp_magnitude(const GpuMat& src, GpuMat& dst, nppMagnitude_t func, cudaStream_t stream)
+ {
+ CV_Assert(src.type() == CV_32FC2);
+
+ dst.create(src.size(), CV_32FC1);
+
+ NppiSize sz;
+ sz.width = src.cols;
+ sz.height = src.rows;
+
+ NppStreamHandler h(stream);
+
+ nppSafeCall( func(src.ptr<Npp32fc>(), static_cast<int>(src.step), dst.ptr<Npp32f>(), static_cast<int>(dst.step), sz) );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+}
+
+void cv::gpu::magnitude(const GpuMat& src, GpuMat& dst, Stream& stream)
+{
+ npp_magnitude(src, dst, nppiMagnitude_32fc32f_C1R, StreamAccessor::getStream(stream));
+}
+
+void cv::gpu::magnitudeSqr(const GpuMat& src, GpuMat& dst, Stream& stream)
+{
+ npp_magnitude(src, dst, nppiMagnitudeSqr_32fc32f_C1R, StreamAccessor::getStream(stream));
+}
+
+////////////////////////////////////////////////////////////////////////
+// Polar <-> Cart
+
+namespace cv { namespace gpu { namespace cudev
+{
+ namespace mathfunc
+ {
+ void cartToPolar_gpu(PtrStepSzf x, PtrStepSzf y, PtrStepSzf mag, bool magSqr, PtrStepSzf angle, bool angleInDegrees, cudaStream_t stream);
+ void polarToCart_gpu(PtrStepSzf mag, PtrStepSzf angle, PtrStepSzf x, PtrStepSzf y, bool angleInDegrees, cudaStream_t stream);
+ }
+}}}
+
+namespace
+{
+ inline void cartToPolar_caller(const GpuMat& x, const GpuMat& y, GpuMat* mag, bool magSqr, GpuMat* angle, bool angleInDegrees, cudaStream_t stream)
+ {
+ using namespace ::cv::gpu::cudev::mathfunc;
+
+ CV_Assert(x.size() == y.size() && x.type() == y.type());
+ CV_Assert(x.depth() == CV_32F);
+
+ if (mag)
+ mag->create(x.size(), x.type());
+ if (angle)
+ angle->create(x.size(), x.type());
+
+ GpuMat x1cn = x.reshape(1);
+ GpuMat y1cn = y.reshape(1);
+ GpuMat mag1cn = mag ? mag->reshape(1) : GpuMat();
+ GpuMat angle1cn = angle ? angle->reshape(1) : GpuMat();
+
+ cartToPolar_gpu(x1cn, y1cn, mag1cn, magSqr, angle1cn, angleInDegrees, stream);
+ }
+
+ inline void polarToCart_caller(const GpuMat& mag, const GpuMat& angle, GpuMat& x, GpuMat& y, bool angleInDegrees, cudaStream_t stream)
+ {
+ using namespace ::cv::gpu::cudev::mathfunc;
+
+ CV_Assert((mag.empty() || mag.size() == angle.size()) && mag.type() == angle.type());
+ CV_Assert(mag.depth() == CV_32F);
+
+ x.create(mag.size(), mag.type());
+ y.create(mag.size(), mag.type());
+
+ GpuMat mag1cn = mag.reshape(1);
+ GpuMat angle1cn = angle.reshape(1);
+ GpuMat x1cn = x.reshape(1);
+ GpuMat y1cn = y.reshape(1);
+
+ polarToCart_gpu(mag1cn, angle1cn, x1cn, y1cn, angleInDegrees, stream);
+ }
+}
+
+void cv::gpu::magnitude(const GpuMat& x, const GpuMat& y, GpuMat& dst, Stream& stream)
+{
+ cartToPolar_caller(x, y, &dst, false, 0, false, StreamAccessor::getStream(stream));
+}
+
+void cv::gpu::magnitudeSqr(const GpuMat& x, const GpuMat& y, GpuMat& dst, Stream& stream)
+{
+ cartToPolar_caller(x, y, &dst, true, 0, false, StreamAccessor::getStream(stream));
+}
+
+void cv::gpu::phase(const GpuMat& x, const GpuMat& y, GpuMat& angle, bool angleInDegrees, Stream& stream)
+{
+ cartToPolar_caller(x, y, 0, false, &angle, angleInDegrees, StreamAccessor::getStream(stream));
+}
+
+void cv::gpu::cartToPolar(const GpuMat& x, const GpuMat& y, GpuMat& mag, GpuMat& angle, bool angleInDegrees, Stream& stream)
+{
+ cartToPolar_caller(x, y, &mag, false, &angle, angleInDegrees, StreamAccessor::getStream(stream));
+}
+
+void cv::gpu::polarToCart(const GpuMat& magnitude, const GpuMat& angle, GpuMat& x, GpuMat& y, bool angleInDegrees, Stream& stream)
+{
+ polarToCart_caller(magnitude, angle, x, y, angleInDegrees, StreamAccessor::getStream(stream));
+}
+
#endif
//
//M*/
-#include "opencv2/core/cuda_devptrs.hpp"
+#include "precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_PRECOMP_H__
+#define __OPENCV_PRECOMP_H__
+
+#include <limits>
+
+#include "cvconfig.h"
+
+#include "opencv2/gpuarithm.hpp"
+#include "opencv2/core/utility.hpp"
+
+#include "opencv2/core/gpu_private.hpp"
+
+#include "opencv2/opencv_modules.hpp"
+
+#ifdef HAVE_OPENCV_GPULEGACY
+# include "opencv2/gpulegacy.hpp"
+# include "opencv2/gpulegacy/private.hpp"
+#endif
+
+#ifdef HAVE_CUBLAS
+# include <cublas.h>
+#endif
+
+#ifdef HAVE_CUFFT
+# include <cufft.h>
+#endif
+
+#endif /* __OPENCV_PRECOMP_H__ */
#if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
-void cv::gpu::meanStdDev(const GpuMat&, Scalar&, Scalar&) { throw_no_cuda(); }
-void cv::gpu::meanStdDev(const GpuMat&, Scalar&, Scalar&, GpuMat&) { throw_no_cuda(); }
double cv::gpu::norm(const GpuMat&, int) { throw_no_cuda(); return 0.0; }
double cv::gpu::norm(const GpuMat&, int, GpuMat&) { throw_no_cuda(); return 0.0; }
double cv::gpu::norm(const GpuMat&, int, const GpuMat&, GpuMat&) { throw_no_cuda(); return 0.0; }
double cv::gpu::norm(const GpuMat&, const GpuMat&, int) { throw_no_cuda(); return 0.0; }
+
Scalar cv::gpu::sum(const GpuMat&) { throw_no_cuda(); return Scalar(); }
Scalar cv::gpu::sum(const GpuMat&, GpuMat&) { throw_no_cuda(); return Scalar(); }
Scalar cv::gpu::sum(const GpuMat&, const GpuMat&, GpuMat&) { throw_no_cuda(); return Scalar(); }
+
Scalar cv::gpu::absSum(const GpuMat&) { throw_no_cuda(); return Scalar(); }
Scalar cv::gpu::absSum(const GpuMat&, GpuMat&) { throw_no_cuda(); return Scalar(); }
Scalar cv::gpu::absSum(const GpuMat&, const GpuMat&, GpuMat&) { throw_no_cuda(); return Scalar(); }
+
Scalar cv::gpu::sqrSum(const GpuMat&) { throw_no_cuda(); return Scalar(); }
Scalar cv::gpu::sqrSum(const GpuMat&, GpuMat&) { throw_no_cuda(); return Scalar(); }
Scalar cv::gpu::sqrSum(const GpuMat&, const GpuMat&, GpuMat&) { throw_no_cuda(); return Scalar(); }
+
void cv::gpu::minMax(const GpuMat&, double*, double*, const GpuMat&) { throw_no_cuda(); }
void cv::gpu::minMax(const GpuMat&, double*, double*, const GpuMat&, GpuMat&) { throw_no_cuda(); }
+
void cv::gpu::minMaxLoc(const GpuMat&, double*, double*, Point*, Point*, const GpuMat&) { throw_no_cuda(); }
void cv::gpu::minMaxLoc(const GpuMat&, double*, double*, Point*, Point*, const GpuMat&, GpuMat&, GpuMat&) { throw_no_cuda(); }
+
int cv::gpu::countNonZero(const GpuMat&) { throw_no_cuda(); return 0; }
int cv::gpu::countNonZero(const GpuMat&, GpuMat&) { throw_no_cuda(); return 0; }
+
void cv::gpu::reduce(const GpuMat&, GpuMat&, int, int, int, Stream&) { throw_no_cuda(); }
+void cv::gpu::meanStdDev(const GpuMat&, Scalar&, Scalar&) { throw_no_cuda(); }
+void cv::gpu::meanStdDev(const GpuMat&, Scalar&, Scalar&, GpuMat&) { throw_no_cuda(); }
+
+void cv::gpu::rectStdDev(const GpuMat&, const GpuMat&, GpuMat&, const Rect&, Stream&) { throw_no_cuda(); }
+
+void cv::gpu::normalize(const GpuMat&, GpuMat&, double, double, int, int, const GpuMat&) { throw_no_cuda(); }
+void cv::gpu::normalize(const GpuMat&, GpuMat&, double, double, int, int, const GpuMat&, GpuMat&, GpuMat&) { throw_no_cuda(); }
+
#else
-#include "opencv2/core/utility.hpp"
namespace
{
};
}
-
-////////////////////////////////////////////////////////////////////////
-// meanStdDev
-
-void cv::gpu::meanStdDev(const GpuMat& src, Scalar& mean, Scalar& stddev)
-{
- GpuMat buf;
- meanStdDev(src, mean, stddev, buf);
-}
-
-void cv::gpu::meanStdDev(const GpuMat& src, Scalar& mean, Scalar& stddev, GpuMat& buf)
-{
- CV_Assert(src.type() == CV_8UC1);
-
- if (!deviceSupports(FEATURE_SET_COMPUTE_13))
- CV_Error(cv::Error::StsNotImplemented, "Not sufficient compute capebility");
-
- NppiSize sz;
- sz.width = src.cols;
- sz.height = src.rows;
-
- DeviceBuffer dbuf(2);
-
- int bufSize;
-#if (CUDA_VERSION <= 4020)
- nppSafeCall( nppiMeanStdDev8uC1RGetBufferHostSize(sz, &bufSize) );
-#else
- nppSafeCall( nppiMeanStdDevGetBufferHostSize_8u_C1R(sz, &bufSize) );
-#endif
-
- ensureSizeIsEnough(1, bufSize, CV_8UC1, buf);
-
- nppSafeCall( nppiMean_StdDev_8u_C1R(src.ptr<Npp8u>(), static_cast<int>(src.step), sz, buf.ptr<Npp8u>(), dbuf, (double*)dbuf + 1) );
-
- cudaSafeCall( cudaDeviceSynchronize() );
-
- double* ptrs[2] = {mean.val, stddev.val};
- dbuf.download(ptrs);
-}
-
////////////////////////////////////////////////////////////////////////
// norm
double cv::gpu::norm(const GpuMat& src, int normType)
{
GpuMat buf;
- return norm(src, normType, GpuMat(), buf);
+ return gpu::norm(src, normType, GpuMat(), buf);
}
double cv::gpu::norm(const GpuMat& src, int normType, GpuMat& buf)
{
- return norm(src, normType, GpuMat(), buf);
+ return gpu::norm(src, normType, GpuMat(), buf);
}
double cv::gpu::norm(const GpuMat& src, int normType, const GpuMat& mask, GpuMat& buf)
GpuMat src_single_channel = src.reshape(1);
if (normType == NORM_L1)
- return absSum(src_single_channel, mask, buf)[0];
+ return gpu::absSum(src_single_channel, mask, buf)[0];
if (normType == NORM_L2)
- return std::sqrt(sqrSum(src_single_channel, mask, buf)[0]);
+ return std::sqrt(gpu::sqrSum(src_single_channel, mask, buf)[0]);
// NORM_INF
double min_val, max_val;
- minMax(src_single_channel, &min_val, &max_val, mask, buf);
+ gpu::minMax(src_single_channel, &min_val, &max_val, mask, buf);
return std::max(std::abs(min_val), std::abs(max_val));
}
Scalar cv::gpu::sum(const GpuMat& src)
{
GpuMat buf;
- return sum(src, GpuMat(), buf);
+ return gpu::sum(src, GpuMat(), buf);
}
Scalar cv::gpu::sum(const GpuMat& src, GpuMat& buf)
{
- return sum(src, GpuMat(), buf);
+ return gpu::sum(src, GpuMat(), buf);
}
Scalar cv::gpu::sum(const GpuMat& src, const GpuMat& mask, GpuMat& buf)
Scalar cv::gpu::absSum(const GpuMat& src)
{
GpuMat buf;
- return absSum(src, GpuMat(), buf);
+ return gpu::absSum(src, GpuMat(), buf);
}
Scalar cv::gpu::absSum(const GpuMat& src, GpuMat& buf)
{
- return absSum(src, GpuMat(), buf);
+ return gpu::absSum(src, GpuMat(), buf);
}
Scalar cv::gpu::absSum(const GpuMat& src, const GpuMat& mask, GpuMat& buf)
Scalar cv::gpu::sqrSum(const GpuMat& src)
{
GpuMat buf;
- return sqrSum(src, GpuMat(), buf);
+ return gpu::sqrSum(src, GpuMat(), buf);
}
Scalar cv::gpu::sqrSum(const GpuMat& src, GpuMat& buf)
{
- return sqrSum(src, GpuMat(), buf);
+ return gpu::sqrSum(src, GpuMat(), buf);
}
Scalar cv::gpu::sqrSum(const GpuMat& src, const GpuMat& mask, GpuMat& buf)
void cv::gpu::minMax(const GpuMat& src, double* minVal, double* maxVal, const GpuMat& mask)
{
GpuMat buf;
- minMax(src, minVal, maxVal, mask, buf);
+ gpu::minMax(src, minVal, maxVal, mask, buf);
}
void cv::gpu::minMax(const GpuMat& src, double* minVal, double* maxVal, const GpuMat& mask, GpuMat& buf)
void cv::gpu::minMaxLoc(const GpuMat& src, double* minVal, double* maxVal, Point* minLoc, Point* maxLoc, const GpuMat& mask)
{
GpuMat valBuf, locBuf;
- minMaxLoc(src, minVal, maxVal, minLoc, maxLoc, mask, valBuf, locBuf);
+ gpu::minMaxLoc(src, minVal, maxVal, minLoc, maxLoc, mask, valBuf, locBuf);
}
void cv::gpu::minMaxLoc(const GpuMat& src, double* minVal, double* maxVal, Point* minLoc, Point* maxLoc,
}
}
+////////////////////////////////////////////////////////////////////////
+// meanStdDev
+
+void cv::gpu::meanStdDev(const GpuMat& src, Scalar& mean, Scalar& stddev)
+{
+ GpuMat buf;
+ meanStdDev(src, mean, stddev, buf);
+}
+
+void cv::gpu::meanStdDev(const GpuMat& src, Scalar& mean, Scalar& stddev, GpuMat& buf)
+{
+ CV_Assert(src.type() == CV_8UC1);
+
+ if (!deviceSupports(FEATURE_SET_COMPUTE_13))
+ CV_Error(cv::Error::StsNotImplemented, "Not sufficient compute capebility");
+
+ NppiSize sz;
+ sz.width = src.cols;
+ sz.height = src.rows;
+
+ DeviceBuffer dbuf(2);
+
+ int bufSize;
+#if (CUDA_VERSION <= 4020)
+ nppSafeCall( nppiMeanStdDev8uC1RGetBufferHostSize(sz, &bufSize) );
+#else
+ nppSafeCall( nppiMeanStdDevGetBufferHostSize_8u_C1R(sz, &bufSize) );
+#endif
+
+ ensureSizeIsEnough(1, bufSize, CV_8UC1, buf);
+
+ nppSafeCall( nppiMean_StdDev_8u_C1R(src.ptr<Npp8u>(), static_cast<int>(src.step), sz, buf.ptr<Npp8u>(), dbuf, (double*)dbuf + 1) );
+
+ cudaSafeCall( cudaDeviceSynchronize() );
+
+ double* ptrs[2] = {mean.val, stddev.val};
+ dbuf.download(ptrs);
+}
+
+//////////////////////////////////////////////////////////////////////////////
+// rectStdDev
+
+void cv::gpu::rectStdDev(const GpuMat& src, const GpuMat& sqr, GpuMat& dst, const Rect& rect, Stream& s)
+{
+ CV_Assert(src.type() == CV_32SC1 && sqr.type() == CV_64FC1);
+
+ dst.create(src.size(), CV_32FC1);
+
+ NppiSize sz;
+ sz.width = src.cols;
+ sz.height = src.rows;
+
+ NppiRect nppRect;
+ nppRect.height = rect.height;
+ nppRect.width = rect.width;
+ nppRect.x = rect.x;
+ nppRect.y = rect.y;
+
+ cudaStream_t stream = StreamAccessor::getStream(s);
+
+ NppStreamHandler h(stream);
+
+ nppSafeCall( nppiRectStdDev_32s32f_C1R(src.ptr<Npp32s>(), static_cast<int>(src.step), sqr.ptr<Npp64f>(), static_cast<int>(sqr.step),
+ dst.ptr<Npp32f>(), static_cast<int>(dst.step), sz, nppRect) );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+}
+
+////////////////////////////////////////////////////////////////////////
+// normalize
+
+void cv::gpu::normalize(const GpuMat& src, GpuMat& dst, double a, double b, int norm_type, int dtype, const GpuMat& mask)
+{
+ GpuMat norm_buf;
+ GpuMat cvt_buf;
+ normalize(src, dst, a, b, norm_type, dtype, mask, norm_buf, cvt_buf);
+}
+
+void cv::gpu::normalize(const GpuMat& src, GpuMat& dst, double a, double b, int norm_type, int dtype, const GpuMat& mask, GpuMat& norm_buf, GpuMat& cvt_buf)
+{
+ double scale = 1, shift = 0;
+ if (norm_type == NORM_MINMAX)
+ {
+ double smin = 0, smax = 0;
+ double dmin = std::min(a, b), dmax = std::max(a, b);
+ gpu::minMax(src, &smin, &smax, mask, norm_buf);
+ scale = (dmax - dmin) * (smax - smin > std::numeric_limits<double>::epsilon() ? 1.0 / (smax - smin) : 0.0);
+ shift = dmin - smin * scale;
+ }
+ else if (norm_type == NORM_L2 || norm_type == NORM_L1 || norm_type == NORM_INF)
+ {
+ scale = gpu::norm(src, norm_type, mask, norm_buf);
+ scale = scale > std::numeric_limits<double>::epsilon() ? a / scale : 0.0;
+ shift = 0;
+ }
+ else
+ {
+ CV_Error(cv::Error::StsBadArg, "Unknown/unsupported norm type");
+ }
+
+ if (mask.empty())
+ {
+ src.convertTo(dst, dtype, scale, shift);
+ }
+ else
+ {
+ src.convertTo(cvt_buf, dtype, scale, shift);
+ cvt_buf.copyTo(dst, mask);
+ }
+}
+
#endif
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+#ifdef HAVE_CUDA
+
+using namespace cvtest;
+
+//////////////////////////////////////////////////////////////////////////////
+// GEMM
+
+#ifdef HAVE_CUBLAS
+
+CV_FLAGS(GemmFlags, 0, cv::GEMM_1_T, cv::GEMM_2_T, cv::GEMM_3_T);
+#define ALL_GEMM_FLAGS testing::Values(GemmFlags(0), GemmFlags(cv::GEMM_1_T), GemmFlags(cv::GEMM_2_T), GemmFlags(cv::GEMM_3_T), GemmFlags(cv::GEMM_1_T | cv::GEMM_2_T), GemmFlags(cv::GEMM_1_T | cv::GEMM_3_T), GemmFlags(cv::GEMM_1_T | cv::GEMM_2_T | cv::GEMM_3_T))
+
+PARAM_TEST_CASE(GEMM, cv::gpu::DeviceInfo, cv::Size, MatType, GemmFlags, UseRoi)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ int type;
+ int flags;
+ bool useRoi;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ type = GET_PARAM(2);
+ flags = GET_PARAM(3);
+ useRoi = GET_PARAM(4);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(GEMM, Accuracy)
+{
+ cv::Mat src1 = randomMat(size, type, -10.0, 10.0);
+ cv::Mat src2 = randomMat(size, type, -10.0, 10.0);
+ cv::Mat src3 = randomMat(size, type, -10.0, 10.0);
+ double alpha = randomDouble(-10.0, 10.0);
+ double beta = randomDouble(-10.0, 10.0);
+
+ if (CV_MAT_DEPTH(type) == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
+ {
+ try
+ {
+ cv::gpu::GpuMat dst;
+ cv::gpu::gemm(loadMat(src1), loadMat(src2), alpha, loadMat(src3), beta, dst, flags);
+ }
+ catch (const cv::Exception& e)
+ {
+ ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
+ }
+ }
+ else if (type == CV_64FC2 && flags != 0)
+ {
+ try
+ {
+ cv::gpu::GpuMat dst;
+ cv::gpu::gemm(loadMat(src1), loadMat(src2), alpha, loadMat(src3), beta, dst, flags);
+ }
+ catch (const cv::Exception& e)
+ {
+ ASSERT_EQ(cv::Error::StsNotImplemented, e.code);
+ }
+ }
+ else
+ {
+ cv::gpu::GpuMat dst = createMat(size, type, useRoi);
+ cv::gpu::gemm(loadMat(src1, useRoi), loadMat(src2, useRoi), alpha, loadMat(src3, useRoi), beta, dst, flags);
+
+ cv::Mat dst_gold;
+ cv::gemm(src1, src2, alpha, src3, beta, dst_gold, flags);
+
+ EXPECT_MAT_NEAR(dst_gold, dst, CV_MAT_DEPTH(type) == CV_32F ? 1e-1 : 1e-10);
+ }
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, GEMM, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ testing::Values(MatType(CV_32FC1), MatType(CV_32FC2), MatType(CV_64FC1), MatType(CV_64FC2)),
+ ALL_GEMM_FLAGS,
+ WHOLE_SUBMAT));
+
+///////////////////////////////////////////////////////////////////////////////////////////////////////
+// Integral
+
+PARAM_TEST_CASE(Integral, cv::gpu::DeviceInfo, cv::Size, UseRoi)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ bool useRoi;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ useRoi = GET_PARAM(2);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(Integral, Accuracy)
+{
+ cv::Mat src = randomMat(size, CV_8UC1);
+
+ cv::gpu::GpuMat dst = createMat(cv::Size(src.cols + 1, src.rows + 1), CV_32SC1, useRoi);
+ cv::gpu::integral(loadMat(src, useRoi), dst);
+
+ cv::Mat dst_gold;
+ cv::integral(src, dst_gold, CV_32S);
+
+ EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Integral, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ WHOLE_SUBMAT));
+
+////////////////////////////////////////////////////////////////////////////
+// MulSpectrums
+
+CV_FLAGS(DftFlags, 0, cv::DFT_INVERSE, cv::DFT_SCALE, cv::DFT_ROWS, cv::DFT_COMPLEX_OUTPUT, cv::DFT_REAL_OUTPUT)
+
+PARAM_TEST_CASE(MulSpectrums, cv::gpu::DeviceInfo, cv::Size, DftFlags)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ int flag;
+
+ cv::Mat a, b;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ flag = GET_PARAM(2);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+
+ a = randomMat(size, CV_32FC2);
+ b = randomMat(size, CV_32FC2);
+ }
+};
+
+GPU_TEST_P(MulSpectrums, Simple)
+{
+ cv::gpu::GpuMat c;
+ cv::gpu::mulSpectrums(loadMat(a), loadMat(b), c, flag, false);
+
+ cv::Mat c_gold;
+ cv::mulSpectrums(a, b, c_gold, flag, false);
+
+ EXPECT_MAT_NEAR(c_gold, c, 1e-2);
+}
+
+GPU_TEST_P(MulSpectrums, Scaled)
+{
+ float scale = 1.f / size.area();
+
+ cv::gpu::GpuMat c;
+ cv::gpu::mulAndScaleSpectrums(loadMat(a), loadMat(b), c, flag, scale, false);
+
+ cv::Mat c_gold;
+ cv::mulSpectrums(a, b, c_gold, flag, false);
+ c_gold.convertTo(c_gold, c_gold.type(), scale);
+
+ EXPECT_MAT_NEAR(c_gold, c, 1e-2);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, MulSpectrums, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ testing::Values(DftFlags(0), DftFlags(cv::DFT_ROWS))));
+
+////////////////////////////////////////////////////////////////////////////
+// Dft
+
+struct Dft : testing::TestWithParam<cv::gpu::DeviceInfo>
+{
+ cv::gpu::DeviceInfo devInfo;
+
+ virtual void SetUp()
+ {
+ devInfo = GetParam();
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+namespace
+{
+ void testC2C(const std::string& hint, int cols, int rows, int flags, bool inplace)
+ {
+ SCOPED_TRACE(hint);
+
+ cv::Mat a = randomMat(cv::Size(cols, rows), CV_32FC2, 0.0, 10.0);
+
+ cv::Mat b_gold;
+ cv::dft(a, b_gold, flags);
+
+ cv::gpu::GpuMat d_b;
+ cv::gpu::GpuMat d_b_data;
+ if (inplace)
+ {
+ d_b_data.create(1, a.size().area(), CV_32FC2);
+ d_b = cv::gpu::GpuMat(a.rows, a.cols, CV_32FC2, d_b_data.ptr(), a.cols * d_b_data.elemSize());
+ }
+ cv::gpu::dft(loadMat(a), d_b, cv::Size(cols, rows), flags);
+
+ EXPECT_TRUE(!inplace || d_b.ptr() == d_b_data.ptr());
+ ASSERT_EQ(CV_32F, d_b.depth());
+ ASSERT_EQ(2, d_b.channels());
+ EXPECT_MAT_NEAR(b_gold, cv::Mat(d_b), rows * cols * 1e-4);
+ }
+}
+
+GPU_TEST_P(Dft, C2C)
+{
+ int cols = randomInt(2, 100);
+ int rows = randomInt(2, 100);
+
+ for (int i = 0; i < 2; ++i)
+ {
+ bool inplace = i != 0;
+
+ testC2C("no flags", cols, rows, 0, inplace);
+ testC2C("no flags 0 1", cols, rows + 1, 0, inplace);
+ testC2C("no flags 1 0", cols, rows + 1, 0, inplace);
+ testC2C("no flags 1 1", cols + 1, rows, 0, inplace);
+ testC2C("DFT_INVERSE", cols, rows, cv::DFT_INVERSE, inplace);
+ testC2C("DFT_ROWS", cols, rows, cv::DFT_ROWS, inplace);
+ testC2C("single col", 1, rows, 0, inplace);
+ testC2C("single row", cols, 1, 0, inplace);
+ testC2C("single col inversed", 1, rows, cv::DFT_INVERSE, inplace);
+ testC2C("single row inversed", cols, 1, cv::DFT_INVERSE, inplace);
+ testC2C("single row DFT_ROWS", cols, 1, cv::DFT_ROWS, inplace);
+ testC2C("size 1 2", 1, 2, 0, inplace);
+ testC2C("size 2 1", 2, 1, 0, inplace);
+ }
+}
+
+namespace
+{
+ void testR2CThenC2R(const std::string& hint, int cols, int rows, bool inplace)
+ {
+ SCOPED_TRACE(hint);
+
+ cv::Mat a = randomMat(cv::Size(cols, rows), CV_32FC1, 0.0, 10.0);
+
+ cv::gpu::GpuMat d_b, d_c;
+ cv::gpu::GpuMat d_b_data, d_c_data;
+ if (inplace)
+ {
+ if (a.cols == 1)
+ {
+ d_b_data.create(1, (a.rows / 2 + 1) * a.cols, CV_32FC2);
+ d_b = cv::gpu::GpuMat(a.rows / 2 + 1, a.cols, CV_32FC2, d_b_data.ptr(), a.cols * d_b_data.elemSize());
+ }
+ else
+ {
+ d_b_data.create(1, a.rows * (a.cols / 2 + 1), CV_32FC2);
+ d_b = cv::gpu::GpuMat(a.rows, a.cols / 2 + 1, CV_32FC2, d_b_data.ptr(), (a.cols / 2 + 1) * d_b_data.elemSize());
+ }
+ d_c_data.create(1, a.size().area(), CV_32F);
+ d_c = cv::gpu::GpuMat(a.rows, a.cols, CV_32F, d_c_data.ptr(), a.cols * d_c_data.elemSize());
+ }
+
+ cv::gpu::dft(loadMat(a), d_b, cv::Size(cols, rows), 0);
+ cv::gpu::dft(d_b, d_c, cv::Size(cols, rows), cv::DFT_REAL_OUTPUT | cv::DFT_SCALE);
+
+ EXPECT_TRUE(!inplace || d_b.ptr() == d_b_data.ptr());
+ EXPECT_TRUE(!inplace || d_c.ptr() == d_c_data.ptr());
+ ASSERT_EQ(CV_32F, d_c.depth());
+ ASSERT_EQ(1, d_c.channels());
+
+ cv::Mat c(d_c);
+ EXPECT_MAT_NEAR(a, c, rows * cols * 1e-5);
+ }
+}
+
+GPU_TEST_P(Dft, R2CThenC2R)
+{
+ int cols = randomInt(2, 100);
+ int rows = randomInt(2, 100);
+
+ testR2CThenC2R("sanity", cols, rows, false);
+ testR2CThenC2R("sanity 0 1", cols, rows + 1, false);
+ testR2CThenC2R("sanity 1 0", cols + 1, rows, false);
+ testR2CThenC2R("sanity 1 1", cols + 1, rows + 1, false);
+ testR2CThenC2R("single col", 1, rows, false);
+ testR2CThenC2R("single col 1", 1, rows + 1, false);
+ testR2CThenC2R("single row", cols, 1, false);
+ testR2CThenC2R("single row 1", cols + 1, 1, false);
+
+ testR2CThenC2R("sanity", cols, rows, true);
+ testR2CThenC2R("sanity 0 1", cols, rows + 1, true);
+ testR2CThenC2R("sanity 1 0", cols + 1, rows, true);
+ testR2CThenC2R("sanity 1 1", cols + 1, rows + 1, true);
+ testR2CThenC2R("single row", cols, 1, true);
+ testR2CThenC2R("single row 1", cols + 1, 1, true);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Dft, ALL_DEVICES);
+
+////////////////////////////////////////////////////////
+// Convolve
+
+namespace
+{
+ void convolveDFT(const cv::Mat& A, const cv::Mat& B, cv::Mat& C, bool ccorr = false)
+ {
+ // reallocate the output array if needed
+ C.create(std::abs(A.rows - B.rows) + 1, std::abs(A.cols - B.cols) + 1, A.type());
+ cv::Size dftSize;
+
+ // compute the size of DFT transform
+ dftSize.width = cv::getOptimalDFTSize(A.cols + B.cols - 1);
+ dftSize.height = cv::getOptimalDFTSize(A.rows + B.rows - 1);
+
+ // allocate temporary buffers and initialize them with 0s
+ cv::Mat tempA(dftSize, A.type(), cv::Scalar::all(0));
+ cv::Mat tempB(dftSize, B.type(), cv::Scalar::all(0));
+
+ // copy A and B to the top-left corners of tempA and tempB, respectively
+ cv::Mat roiA(tempA, cv::Rect(0, 0, A.cols, A.rows));
+ A.copyTo(roiA);
+ cv::Mat roiB(tempB, cv::Rect(0, 0, B.cols, B.rows));
+ B.copyTo(roiB);
+
+ // now transform the padded A & B in-place;
+ // use "nonzeroRows" hint for faster processing
+ cv::dft(tempA, tempA, 0, A.rows);
+ cv::dft(tempB, tempB, 0, B.rows);
+
+ // multiply the spectrums;
+ // the function handles packed spectrum representations well
+ cv::mulSpectrums(tempA, tempB, tempA, 0, ccorr);
+
+ // transform the product back from the frequency domain.
+ // Even though all the result rows will be non-zero,
+ // you need only the first C.rows of them, and thus you
+ // pass nonzeroRows == C.rows
+ cv::dft(tempA, tempA, cv::DFT_INVERSE + cv::DFT_SCALE, C.rows);
+
+ // now copy the result back to C.
+ tempA(cv::Rect(0, 0, C.cols, C.rows)).copyTo(C);
+ }
+
+ IMPLEMENT_PARAM_CLASS(KSize, int)
+ IMPLEMENT_PARAM_CLASS(Ccorr, bool)
+}
+
+PARAM_TEST_CASE(Convolve, cv::gpu::DeviceInfo, cv::Size, KSize, Ccorr)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ int ksize;
+ bool ccorr;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ ksize = GET_PARAM(2);
+ ccorr = GET_PARAM(3);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(Convolve, Accuracy)
+{
+ cv::Mat src = randomMat(size, CV_32FC1, 0.0, 100.0);
+ cv::Mat kernel = randomMat(cv::Size(ksize, ksize), CV_32FC1, 0.0, 1.0);
+
+ cv::gpu::GpuMat dst;
+ cv::gpu::convolve(loadMat(src), loadMat(kernel), dst, ccorr);
+
+ cv::Mat dst_gold;
+ convolveDFT(src, kernel, dst_gold, ccorr);
+
+ EXPECT_MAT_NEAR(dst, dst_gold, 1e-1);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Convolve, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ testing::Values(KSize(3), KSize(7), KSize(11), KSize(17), KSize(19), KSize(23), KSize(45)),
+ testing::Values(Ccorr(false), Ccorr(true))));
+
+#endif // HAVE_CUBLAS
+
+#endif // HAVE_CUDA
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+#ifdef HAVE_CUDA
+
+using namespace cvtest;
+
+////////////////////////////////////////////////////////////////////////////////
+// Merge
+
+PARAM_TEST_CASE(Merge, cv::gpu::DeviceInfo, cv::Size, MatDepth, Channels, UseRoi)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ int depth;
+ int channels;
+ bool useRoi;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ depth = GET_PARAM(2);
+ channels = GET_PARAM(3);
+ useRoi = GET_PARAM(4);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(Merge, Accuracy)
+{
+ std::vector<cv::Mat> src;
+ src.reserve(channels);
+ for (int i = 0; i < channels; ++i)
+ src.push_back(cv::Mat(size, depth, cv::Scalar::all(i)));
+
+ std::vector<cv::gpu::GpuMat> d_src;
+ for (int i = 0; i < channels; ++i)
+ d_src.push_back(loadMat(src[i], useRoi));
+
+ if (depth == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
+ {
+ try
+ {
+ cv::gpu::GpuMat dst;
+ cv::gpu::merge(d_src, dst);
+ }
+ catch (const cv::Exception& e)
+ {
+ ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
+ }
+ }
+ else
+ {
+ cv::gpu::GpuMat dst;
+ cv::gpu::merge(d_src, dst);
+
+ cv::Mat dst_gold;
+ cv::merge(src, dst_gold);
+
+ EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
+ }
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Merge, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ ALL_DEPTH,
+ testing::Values(1, 2, 3, 4),
+ WHOLE_SUBMAT));
+
+////////////////////////////////////////////////////////////////////////////////
+// Split
+
+PARAM_TEST_CASE(Split, cv::gpu::DeviceInfo, cv::Size, MatDepth, Channels, UseRoi)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ int depth;
+ int channels;
+ bool useRoi;
+
+ int type;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ depth = GET_PARAM(2);
+ channels = GET_PARAM(3);
+ useRoi = GET_PARAM(4);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+
+ type = CV_MAKE_TYPE(depth, channels);
+ }
+};
+
+GPU_TEST_P(Split, Accuracy)
+{
+ cv::Mat src = randomMat(size, type);
+
+ if (depth == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
+ {
+ try
+ {
+ std::vector<cv::gpu::GpuMat> dst;
+ cv::gpu::split(loadMat(src), dst);
+ }
+ catch (const cv::Exception& e)
+ {
+ ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
+ }
+ }
+ else
+ {
+ std::vector<cv::gpu::GpuMat> dst;
+ cv::gpu::split(loadMat(src, useRoi), dst);
+
+ std::vector<cv::Mat> dst_gold;
+ cv::split(src, dst_gold);
+
+ ASSERT_EQ(dst_gold.size(), dst.size());
+
+ for (size_t i = 0; i < dst_gold.size(); ++i)
+ {
+ EXPECT_MAT_NEAR(dst_gold[i], dst[i], 0.0);
+ }
+ }
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Split, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ ALL_DEPTH,
+ testing::Values(1, 2, 3, 4),
+ WHOLE_SUBMAT));
+
+////////////////////////////////////////////////////////////////////////////////
+// Transpose
+
+PARAM_TEST_CASE(Transpose, cv::gpu::DeviceInfo, cv::Size, MatType, UseRoi)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ int type;
+ bool useRoi;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ type = GET_PARAM(2);
+ useRoi = GET_PARAM(3);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(Transpose, Accuracy)
+{
+ cv::Mat src = randomMat(size, type);
+
+ if (CV_MAT_DEPTH(type) == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
+ {
+ try
+ {
+ cv::gpu::GpuMat dst;
+ cv::gpu::transpose(loadMat(src), dst);
+ }
+ catch (const cv::Exception& e)
+ {
+ ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
+ }
+ }
+ else
+ {
+ cv::gpu::GpuMat dst = createMat(cv::Size(size.height, size.width), type, useRoi);
+ cv::gpu::transpose(loadMat(src, useRoi), dst);
+
+ cv::Mat dst_gold;
+ cv::transpose(src, dst_gold);
+
+ EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
+ }
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Transpose, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ testing::Values(MatType(CV_8UC1),
+ MatType(CV_8UC4),
+ MatType(CV_16UC2),
+ MatType(CV_16SC2),
+ MatType(CV_32SC1),
+ MatType(CV_32SC2),
+ MatType(CV_64FC1)),
+ WHOLE_SUBMAT));
+
+////////////////////////////////////////////////////////////////////////////////
+// Flip
+
+enum {FLIP_BOTH = 0, FLIP_X = 1, FLIP_Y = -1};
+CV_ENUM(FlipCode, FLIP_BOTH, FLIP_X, FLIP_Y)
+#define ALL_FLIP_CODES testing::Values(FlipCode(FLIP_BOTH), FlipCode(FLIP_X), FlipCode(FLIP_Y))
+
+PARAM_TEST_CASE(Flip, cv::gpu::DeviceInfo, cv::Size, MatType, FlipCode, UseRoi)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ int type;
+ int flip_code;
+ bool useRoi;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ type = GET_PARAM(2);
+ flip_code = GET_PARAM(3);
+ useRoi = GET_PARAM(4);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(Flip, Accuracy)
+{
+ cv::Mat src = randomMat(size, type);
+
+ cv::gpu::GpuMat dst = createMat(size, type, useRoi);
+ cv::gpu::flip(loadMat(src, useRoi), dst, flip_code);
+
+ cv::Mat dst_gold;
+ cv::flip(src, dst_gold, flip_code);
+
+ EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Flip, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ testing::Values(MatType(CV_8UC1),
+ MatType(CV_8UC3),
+ MatType(CV_8UC4),
+ MatType(CV_16UC1),
+ MatType(CV_16UC3),
+ MatType(CV_16UC4),
+ MatType(CV_32SC1),
+ MatType(CV_32SC3),
+ MatType(CV_32SC4),
+ MatType(CV_32FC1),
+ MatType(CV_32FC3),
+ MatType(CV_32FC4)),
+ ALL_FLIP_CODES,
+ WHOLE_SUBMAT));
+
+////////////////////////////////////////////////////////////////////////////////
+// LUT
+
+PARAM_TEST_CASE(LUT, cv::gpu::DeviceInfo, cv::Size, MatType, UseRoi)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ int type;
+ bool useRoi;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ type = GET_PARAM(2);
+ useRoi = GET_PARAM(3);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(LUT, OneChannel)
+{
+ cv::Mat src = randomMat(size, type);
+ cv::Mat lut = randomMat(cv::Size(256, 1), CV_8UC1);
+
+ cv::gpu::GpuMat dst = createMat(size, CV_MAKE_TYPE(lut.depth(), src.channels()));
+ cv::gpu::LUT(loadMat(src, useRoi), lut, dst);
+
+ cv::Mat dst_gold;
+ cv::LUT(src, lut, dst_gold);
+
+ EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
+}
+
+GPU_TEST_P(LUT, MultiChannel)
+{
+ cv::Mat src = randomMat(size, type);
+ cv::Mat lut = randomMat(cv::Size(256, 1), CV_MAKE_TYPE(CV_8U, src.channels()));
+
+ cv::gpu::GpuMat dst = createMat(size, CV_MAKE_TYPE(lut.depth(), src.channels()), useRoi);
+ cv::gpu::LUT(loadMat(src, useRoi), lut, dst);
+
+ cv::Mat dst_gold;
+ cv::LUT(src, lut, dst_gold);
+
+ EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, LUT, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ testing::Values(MatType(CV_8UC1), MatType(CV_8UC3)),
+ WHOLE_SUBMAT));
+
+//////////////////////////////////////////////////////////////////////////////
+// CopyMakeBorder
+
+namespace
+{
+ IMPLEMENT_PARAM_CLASS(Border, int)
+}
+
+PARAM_TEST_CASE(CopyMakeBorder, cv::gpu::DeviceInfo, cv::Size, MatType, Border, BorderType, UseRoi)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ int type;
+ int border;
+ int borderType;
+ bool useRoi;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ type = GET_PARAM(2);
+ border = GET_PARAM(3);
+ borderType = GET_PARAM(4);
+ useRoi = GET_PARAM(5);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(CopyMakeBorder, Accuracy)
+{
+ cv::Mat src = randomMat(size, type);
+ cv::Scalar val = randomScalar(0, 255);
+
+ cv::gpu::GpuMat dst = createMat(cv::Size(size.width + 2 * border, size.height + 2 * border), type, useRoi);
+ cv::gpu::copyMakeBorder(loadMat(src, useRoi), dst, border, border, border, border, borderType, val);
+
+ cv::Mat dst_gold;
+ cv::copyMakeBorder(src, dst_gold, border, border, border, border, borderType, val);
+
+ EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, CopyMakeBorder, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ testing::Values(MatType(CV_8UC1),
+ MatType(CV_8UC3),
+ MatType(CV_8UC4),
+ MatType(CV_16UC1),
+ MatType(CV_16UC3),
+ MatType(CV_16UC4),
+ MatType(CV_32FC1),
+ MatType(CV_32FC3),
+ MatType(CV_32FC4)),
+ testing::Values(Border(1), Border(10), Border(50)),
+ ALL_BORDER_TYPES,
+ WHOLE_SUBMAT));
+
+#endif // HAVE_CUDA
using namespace cvtest;
////////////////////////////////////////////////////////////////////////////////
-// Merge
-
-PARAM_TEST_CASE(Merge, cv::gpu::DeviceInfo, cv::Size, MatDepth, Channels, UseRoi)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- int depth;
- int channels;
- bool useRoi;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- depth = GET_PARAM(2);
- channels = GET_PARAM(3);
- useRoi = GET_PARAM(4);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(Merge, Accuracy)
-{
- std::vector<cv::Mat> src;
- src.reserve(channels);
- for (int i = 0; i < channels; ++i)
- src.push_back(cv::Mat(size, depth, cv::Scalar::all(i)));
-
- std::vector<cv::gpu::GpuMat> d_src;
- for (int i = 0; i < channels; ++i)
- d_src.push_back(loadMat(src[i], useRoi));
-
- if (depth == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
- {
- try
- {
- cv::gpu::GpuMat dst;
- cv::gpu::merge(d_src, dst);
- }
- catch (const cv::Exception& e)
- {
- ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
- }
- }
- else
- {
- cv::gpu::GpuMat dst;
- cv::gpu::merge(d_src, dst);
-
- cv::Mat dst_gold;
- cv::merge(src, dst_gold);
-
- EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
- }
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_Core, Merge, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- ALL_DEPTH,
- testing::Values(1, 2, 3, 4),
- WHOLE_SUBMAT));
-
-////////////////////////////////////////////////////////////////////////////////
-// Split
-
-PARAM_TEST_CASE(Split, cv::gpu::DeviceInfo, cv::Size, MatDepth, Channels, UseRoi)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- int depth;
- int channels;
- bool useRoi;
-
- int type;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- depth = GET_PARAM(2);
- channels = GET_PARAM(3);
- useRoi = GET_PARAM(4);
-
- cv::gpu::setDevice(devInfo.deviceID());
-
- type = CV_MAKE_TYPE(depth, channels);
- }
-};
-
-GPU_TEST_P(Split, Accuracy)
-{
- cv::Mat src = randomMat(size, type);
-
- if (depth == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
- {
- try
- {
- std::vector<cv::gpu::GpuMat> dst;
- cv::gpu::split(loadMat(src), dst);
- }
- catch (const cv::Exception& e)
- {
- ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
- }
- }
- else
- {
- std::vector<cv::gpu::GpuMat> dst;
- cv::gpu::split(loadMat(src, useRoi), dst);
-
- std::vector<cv::Mat> dst_gold;
- cv::split(src, dst_gold);
-
- ASSERT_EQ(dst_gold.size(), dst.size());
-
- for (size_t i = 0; i < dst_gold.size(); ++i)
- {
- EXPECT_MAT_NEAR(dst_gold[i], dst[i], 0.0);
- }
- }
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_Core, Split, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- ALL_DEPTH,
- testing::Values(1, 2, 3, 4),
- WHOLE_SUBMAT));
-
-////////////////////////////////////////////////////////////////////////////////
// Add_Array
PARAM_TEST_CASE(Add_Array, cv::gpu::DeviceInfo, cv::Size, std::pair<MatDepth, MatDepth>, Channels, UseRoi)
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Add_Array, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Add_Array, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
DEPTH_PAIRS,
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Add_Array_Mask, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Add_Array_Mask, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
DEPTH_PAIRS,
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Add_Scalar, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Add_Scalar, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
DEPTH_PAIRS,
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Subtract_Array, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Subtract_Array, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
DEPTH_PAIRS,
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Subtract_Array_Mask, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Subtract_Array_Mask, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
DEPTH_PAIRS,
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Subtract_Scalar, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Subtract_Scalar, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
DEPTH_PAIRS,
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Multiply_Array, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Multiply_Array, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
DEPTH_PAIRS,
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Multiply_Array_Special, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Multiply_Array_Special, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
WHOLE_SUBMAT));
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Multiply_Scalar, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Multiply_Scalar, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
DEPTH_PAIRS,
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Divide_Array, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Divide_Array, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
DEPTH_PAIRS,
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Divide_Array_Special, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Divide_Array_Special, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
WHOLE_SUBMAT));
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Divide_Scalar, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Divide_Scalar, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
DEPTH_PAIRS,
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Divide_Scalar_Inv, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Divide_Scalar_Inv, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
DEPTH_PAIRS,
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, AbsDiff, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, AbsDiff, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
ALL_DEPTH,
EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Abs, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Abs, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatDepth(CV_16S), MatDepth(CV_32F)),
EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Sqr, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Sqr, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatDepth(CV_8U),
EXPECT_MAT_NEAR(dst_gold, dst, depth < CV_32F ? 1.0 : 1e-5);
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Sqrt, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Sqrt, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatDepth(CV_8U),
EXPECT_MAT_NEAR(dst_gold, dst, depth < CV_32F ? 1.0 : 1e-6);
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Log, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Log, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatDepth(CV_8U),
EXPECT_MAT_NEAR(dst_gold, dst, depth < CV_32F ? 1.0 : 1e-2);
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Exp, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Exp, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatDepth(CV_8U),
////////////////////////////////////////////////////////////////////////////////
// Compare_Array
-CV_ENUM(CmpCode, CMP_EQ, CMP_NE, CMP_GT, CMP_GE, CMP_LT, CMP_LE)
+CV_ENUM(CmpCode, cv::CMP_EQ, cv::CMP_GT, cv::CMP_GE, cv::CMP_LT, cv::CMP_LE, cv::CMP_NE)
+#define ALL_CMP_CODES testing::Values(CmpCode(cv::CMP_EQ), CmpCode(cv::CMP_NE), CmpCode(cv::CMP_GT), CmpCode(cv::CMP_GE), CmpCode(cv::CMP_LT), CmpCode(cv::CMP_LE))
PARAM_TEST_CASE(Compare_Array, cv::gpu::DeviceInfo, cv::Size, MatDepth, CmpCode, UseRoi)
{
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Compare_Array, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Compare_Array, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
ALL_DEPTH,
- CmpCode::all(),
+ ALL_CMP_CODES,
WHOLE_SUBMAT));
////////////////////////////////////////////////////////////////////////////////
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Compare_Scalar, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Compare_Scalar, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
TYPES(CV_8U, CV_64F, 1, 4),
- CmpCode::all(),
+ ALL_CMP_CODES,
WHOLE_SUBMAT));
//////////////////////////////////////////////////////////////////////////////
EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Bitwise_Array, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Bitwise_Array, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
TYPES(CV_8U, CV_32S, 1, 4)));
EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Bitwise_Scalar, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Bitwise_Scalar, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatDepth(CV_8U), MatDepth(CV_16U), MatDepth(CV_32S)),
EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, RShift, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, RShift, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatDepth(CV_8U),
EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, LShift, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, LShift, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatDepth(CV_8U), MatDepth(CV_16U), MatDepth(CV_32S)),
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Min, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Min, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
ALL_DEPTH,
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Max, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Max, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
ALL_DEPTH,
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Pow, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Pow, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
ALL_DEPTH,
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, AddWeighted, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, AddWeighted, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
ALL_DEPTH,
ALL_DEPTH,
WHOLE_SUBMAT));
-//////////////////////////////////////////////////////////////////////////////
-// GEMM
-
-#ifdef HAVE_CUBLAS
-
-CV_FLAGS(GemmFlags, 0, GEMM_1_T, GEMM_2_T, GEMM_3_T);
-#define ALL_GEMM_FLAGS testing::Values(GemmFlags(0), GemmFlags(cv::GEMM_1_T), GemmFlags(cv::GEMM_2_T), GemmFlags(cv::GEMM_3_T), GemmFlags(cv::GEMM_1_T | cv::GEMM_2_T), GemmFlags(cv::GEMM_1_T | cv::GEMM_3_T), GemmFlags(cv::GEMM_1_T | cv::GEMM_2_T | cv::GEMM_3_T))
+////////////////////////////////////////////////////////////////////////////////
+// Magnitude
-PARAM_TEST_CASE(GEMM, cv::gpu::DeviceInfo, cv::Size, MatType, GemmFlags, UseRoi)
+PARAM_TEST_CASE(Magnitude, cv::gpu::DeviceInfo, cv::Size, UseRoi)
{
cv::gpu::DeviceInfo devInfo;
cv::Size size;
- int type;
- int flags;
bool useRoi;
virtual void SetUp()
{
devInfo = GET_PARAM(0);
size = GET_PARAM(1);
- type = GET_PARAM(2);
- flags = GET_PARAM(3);
- useRoi = GET_PARAM(4);
+ useRoi = GET_PARAM(2);
cv::gpu::setDevice(devInfo.deviceID());
}
};
-GPU_TEST_P(GEMM, Accuracy)
+GPU_TEST_P(Magnitude, NPP)
{
- cv::Mat src1 = randomMat(size, type, -10.0, 10.0);
- cv::Mat src2 = randomMat(size, type, -10.0, 10.0);
- cv::Mat src3 = randomMat(size, type, -10.0, 10.0);
- double alpha = randomDouble(-10.0, 10.0);
- double beta = randomDouble(-10.0, 10.0);
+ cv::Mat src = randomMat(size, CV_32FC2);
- if (CV_MAT_DEPTH(type) == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
- {
- try
- {
- cv::gpu::GpuMat dst;
- cv::gpu::gemm(loadMat(src1), loadMat(src2), alpha, loadMat(src3), beta, dst, flags);
- }
- catch (const cv::Exception& e)
- {
- ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
- }
- }
- else if (type == CV_64FC2 && flags != 0)
- {
- try
- {
- cv::gpu::GpuMat dst;
- cv::gpu::gemm(loadMat(src1), loadMat(src2), alpha, loadMat(src3), beta, dst, flags);
- }
- catch (const cv::Exception& e)
- {
- ASSERT_EQ(cv::Error::StsNotImplemented, e.code);
- }
- }
- else
- {
- cv::gpu::GpuMat dst = createMat(size, type, useRoi);
- cv::gpu::gemm(loadMat(src1, useRoi), loadMat(src2, useRoi), alpha, loadMat(src3, useRoi), beta, dst, flags);
+ cv::gpu::GpuMat dst = createMat(size, CV_32FC1, useRoi);
+ cv::gpu::magnitude(loadMat(src, useRoi), dst);
- cv::Mat dst_gold;
- cv::gemm(src1, src2, alpha, src3, beta, dst_gold, flags);
+ cv::Mat arr[2];
+ cv::split(src, arr);
+ cv::Mat dst_gold;
+ cv::magnitude(arr[0], arr[1], dst_gold);
- EXPECT_MAT_NEAR(dst_gold, dst, CV_MAT_DEPTH(type) == CV_32F ? 1e-1 : 1e-10);
- }
+ EXPECT_MAT_NEAR(dst_gold, dst, 1e-4);
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, GEMM, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- testing::Values(MatType(CV_32FC1), MatType(CV_32FC2), MatType(CV_64FC1), MatType(CV_64FC2)),
- ALL_GEMM_FLAGS,
- WHOLE_SUBMAT));
+GPU_TEST_P(Magnitude, Sqr_NPP)
+{
+ cv::Mat src = randomMat(size, CV_32FC2);
+
+ cv::gpu::GpuMat dst = createMat(size, CV_32FC1, useRoi);
+ cv::gpu::magnitudeSqr(loadMat(src, useRoi), dst);
-#endif // HAVE_CUBLAS
+ cv::Mat arr[2];
+ cv::split(src, arr);
+ cv::Mat dst_gold;
+ cv::magnitude(arr[0], arr[1], dst_gold);
+ cv::multiply(dst_gold, dst_gold, dst_gold);
-////////////////////////////////////////////////////////////////////////////////
-// Transpose
+ EXPECT_MAT_NEAR(dst_gold, dst, 1e-1);
+}
-PARAM_TEST_CASE(Transpose, cv::gpu::DeviceInfo, cv::Size, MatType, UseRoi)
+GPU_TEST_P(Magnitude, Accuracy)
{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- int type;
- bool useRoi;
+ cv::Mat x = randomMat(size, CV_32FC1);
+ cv::Mat y = randomMat(size, CV_32FC1);
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- type = GET_PARAM(2);
- useRoi = GET_PARAM(3);
+ cv::gpu::GpuMat dst = createMat(size, CV_32FC1, useRoi);
+ cv::gpu::magnitude(loadMat(x, useRoi), loadMat(y, useRoi), dst);
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
+ cv::Mat dst_gold;
+ cv::magnitude(x, y, dst_gold);
+
+ EXPECT_MAT_NEAR(dst_gold, dst, 1e-4);
+}
-GPU_TEST_P(Transpose, Accuracy)
+GPU_TEST_P(Magnitude, Sqr_Accuracy)
{
- cv::Mat src = randomMat(size, type);
+ cv::Mat x = randomMat(size, CV_32FC1);
+ cv::Mat y = randomMat(size, CV_32FC1);
- if (CV_MAT_DEPTH(type) == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
- {
- try
- {
- cv::gpu::GpuMat dst;
- cv::gpu::transpose(loadMat(src), dst);
- }
- catch (const cv::Exception& e)
- {
- ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
- }
- }
- else
- {
- cv::gpu::GpuMat dst = createMat(cv::Size(size.height, size.width), type, useRoi);
- cv::gpu::transpose(loadMat(src, useRoi), dst);
+ cv::gpu::GpuMat dst = createMat(size, CV_32FC1, useRoi);
+ cv::gpu::magnitudeSqr(loadMat(x, useRoi), loadMat(y, useRoi), dst);
- cv::Mat dst_gold;
- cv::transpose(src, dst_gold);
+ cv::Mat dst_gold;
+ cv::magnitude(x, y, dst_gold);
+ cv::multiply(dst_gold, dst_gold, dst_gold);
- EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
- }
+ EXPECT_MAT_NEAR(dst_gold, dst, 1e-1);
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Transpose, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Magnitude, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
- testing::Values(MatType(CV_8UC1),
- MatType(CV_8UC4),
- MatType(CV_16UC2),
- MatType(CV_16SC2),
- MatType(CV_32SC1),
- MatType(CV_32SC2),
- MatType(CV_64FC1)),
WHOLE_SUBMAT));
////////////////////////////////////////////////////////////////////////////////
-// Flip
+// Phase
-enum {FLIP_BOTH = 0, FLIP_X = 1, FLIP_Y = -1};
-CV_ENUM(FlipCode, FLIP_BOTH, FLIP_X, FLIP_Y)
-#define ALL_FLIP_CODES testing::Values(FlipCode(FLIP_BOTH), FlipCode(FLIP_X), FlipCode(FLIP_Y))
+namespace
+{
+ IMPLEMENT_PARAM_CLASS(AngleInDegrees, bool)
+}
-PARAM_TEST_CASE(Flip, cv::gpu::DeviceInfo, cv::Size, MatType, FlipCode, UseRoi)
+PARAM_TEST_CASE(Phase, cv::gpu::DeviceInfo, cv::Size, AngleInDegrees, UseRoi)
{
cv::gpu::DeviceInfo devInfo;
cv::Size size;
- int type;
- int flip_code;
+ bool angleInDegrees;
bool useRoi;
virtual void SetUp()
{
devInfo = GET_PARAM(0);
size = GET_PARAM(1);
- type = GET_PARAM(2);
- flip_code = GET_PARAM(3);
- useRoi = GET_PARAM(4);
+ angleInDegrees = GET_PARAM(2);
+ useRoi = GET_PARAM(3);
cv::gpu::setDevice(devInfo.deviceID());
}
};
-GPU_TEST_P(Flip, Accuracy)
+GPU_TEST_P(Phase, Accuracy)
{
- cv::Mat src = randomMat(size, type);
+ cv::Mat x = randomMat(size, CV_32FC1);
+ cv::Mat y = randomMat(size, CV_32FC1);
- cv::gpu::GpuMat dst = createMat(size, type, useRoi);
- cv::gpu::flip(loadMat(src, useRoi), dst, flip_code);
+ cv::gpu::GpuMat dst = createMat(size, CV_32FC1, useRoi);
+ cv::gpu::phase(loadMat(x, useRoi), loadMat(y, useRoi), dst, angleInDegrees);
cv::Mat dst_gold;
- cv::flip(src, dst_gold, flip_code);
+ cv::phase(x, y, dst_gold, angleInDegrees);
- EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
+ EXPECT_MAT_NEAR(dst_gold, dst, angleInDegrees ? 1e-2 : 1e-3);
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, Flip, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Phase, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
- testing::Values(MatType(CV_8UC1),
- MatType(CV_8UC3),
- MatType(CV_8UC4),
- MatType(CV_16UC1),
- MatType(CV_16UC3),
- MatType(CV_16UC4),
- MatType(CV_32SC1),
- MatType(CV_32SC3),
- MatType(CV_32SC4),
- MatType(CV_32FC1),
- MatType(CV_32FC3),
- MatType(CV_32FC4)),
- ALL_FLIP_CODES,
+ testing::Values(AngleInDegrees(false), AngleInDegrees(true)),
WHOLE_SUBMAT));
////////////////////////////////////////////////////////////////////////////////
-// LUT
+// CartToPolar
-PARAM_TEST_CASE(LUT, cv::gpu::DeviceInfo, cv::Size, MatType, UseRoi)
+PARAM_TEST_CASE(CartToPolar, cv::gpu::DeviceInfo, cv::Size, AngleInDegrees, UseRoi)
{
cv::gpu::DeviceInfo devInfo;
cv::Size size;
- int type;
+ bool angleInDegrees;
bool useRoi;
virtual void SetUp()
{
devInfo = GET_PARAM(0);
size = GET_PARAM(1);
- type = GET_PARAM(2);
+ angleInDegrees = GET_PARAM(2);
useRoi = GET_PARAM(3);
cv::gpu::setDevice(devInfo.deviceID());
}
};
-GPU_TEST_P(LUT, OneChannel)
+GPU_TEST_P(CartToPolar, Accuracy)
{
- cv::Mat src = randomMat(size, type);
- cv::Mat lut = randomMat(cv::Size(256, 1), CV_8UC1);
+ cv::Mat x = randomMat(size, CV_32FC1);
+ cv::Mat y = randomMat(size, CV_32FC1);
- cv::gpu::GpuMat dst = createMat(size, CV_MAKE_TYPE(lut.depth(), src.channels()));
- cv::gpu::LUT(loadMat(src, useRoi), lut, dst);
+ cv::gpu::GpuMat mag = createMat(size, CV_32FC1, useRoi);
+ cv::gpu::GpuMat angle = createMat(size, CV_32FC1, useRoi);
+ cv::gpu::cartToPolar(loadMat(x, useRoi), loadMat(y, useRoi), mag, angle, angleInDegrees);
- cv::Mat dst_gold;
- cv::LUT(src, lut, dst_gold);
+ cv::Mat mag_gold;
+ cv::Mat angle_gold;
+ cv::cartToPolar(x, y, mag_gold, angle_gold, angleInDegrees);
- EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
+ EXPECT_MAT_NEAR(mag_gold, mag, 1e-4);
+ EXPECT_MAT_NEAR(angle_gold, angle, angleInDegrees ? 1e-2 : 1e-3);
}
-GPU_TEST_P(LUT, MultiChannel)
-{
- cv::Mat src = randomMat(size, type);
- cv::Mat lut = randomMat(cv::Size(256, 1), CV_MAKE_TYPE(CV_8U, src.channels()));
-
- cv::gpu::GpuMat dst = createMat(size, CV_MAKE_TYPE(lut.depth(), src.channels()), useRoi);
- cv::gpu::LUT(loadMat(src, useRoi), lut, dst);
-
- cv::Mat dst_gold;
- cv::LUT(src, lut, dst_gold);
-
- EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_Core, LUT, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- testing::Values(MatType(CV_8UC1), MatType(CV_8UC3)),
- WHOLE_SUBMAT));
-
-////////////////////////////////////////////////////////////////////////////////
-// Magnitude
-
-PARAM_TEST_CASE(Magnitude, cv::gpu::DeviceInfo, cv::Size, UseRoi)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- bool useRoi;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- useRoi = GET_PARAM(2);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(Magnitude, NPP)
-{
- cv::Mat src = randomMat(size, CV_32FC2);
-
- cv::gpu::GpuMat dst = createMat(size, CV_32FC1, useRoi);
- cv::gpu::magnitude(loadMat(src, useRoi), dst);
-
- cv::Mat arr[2];
- cv::split(src, arr);
- cv::Mat dst_gold;
- cv::magnitude(arr[0], arr[1], dst_gold);
-
- EXPECT_MAT_NEAR(dst_gold, dst, 1e-4);
-}
-
-GPU_TEST_P(Magnitude, Sqr_NPP)
-{
- cv::Mat src = randomMat(size, CV_32FC2);
-
- cv::gpu::GpuMat dst = createMat(size, CV_32FC1, useRoi);
- cv::gpu::magnitudeSqr(loadMat(src, useRoi), dst);
-
- cv::Mat arr[2];
- cv::split(src, arr);
- cv::Mat dst_gold;
- cv::magnitude(arr[0], arr[1], dst_gold);
- cv::multiply(dst_gold, dst_gold, dst_gold);
-
- EXPECT_MAT_NEAR(dst_gold, dst, 1e-1);
-}
-
-GPU_TEST_P(Magnitude, Accuracy)
-{
- cv::Mat x = randomMat(size, CV_32FC1);
- cv::Mat y = randomMat(size, CV_32FC1);
-
- cv::gpu::GpuMat dst = createMat(size, CV_32FC1, useRoi);
- cv::gpu::magnitude(loadMat(x, useRoi), loadMat(y, useRoi), dst);
-
- cv::Mat dst_gold;
- cv::magnitude(x, y, dst_gold);
-
- EXPECT_MAT_NEAR(dst_gold, dst, 1e-4);
-}
-
-GPU_TEST_P(Magnitude, Sqr_Accuracy)
-{
- cv::Mat x = randomMat(size, CV_32FC1);
- cv::Mat y = randomMat(size, CV_32FC1);
-
- cv::gpu::GpuMat dst = createMat(size, CV_32FC1, useRoi);
- cv::gpu::magnitudeSqr(loadMat(x, useRoi), loadMat(y, useRoi), dst);
-
- cv::Mat dst_gold;
- cv::magnitude(x, y, dst_gold);
- cv::multiply(dst_gold, dst_gold, dst_gold);
-
- EXPECT_MAT_NEAR(dst_gold, dst, 1e-1);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_Core, Magnitude, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- WHOLE_SUBMAT));
-
-////////////////////////////////////////////////////////////////////////////////
-// Phase
-
-namespace
-{
- IMPLEMENT_PARAM_CLASS(AngleInDegrees, bool)
-}
-
-PARAM_TEST_CASE(Phase, cv::gpu::DeviceInfo, cv::Size, AngleInDegrees, UseRoi)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- bool angleInDegrees;
- bool useRoi;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- angleInDegrees = GET_PARAM(2);
- useRoi = GET_PARAM(3);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(Phase, Accuracy)
-{
- cv::Mat x = randomMat(size, CV_32FC1);
- cv::Mat y = randomMat(size, CV_32FC1);
-
- cv::gpu::GpuMat dst = createMat(size, CV_32FC1, useRoi);
- cv::gpu::phase(loadMat(x, useRoi), loadMat(y, useRoi), dst, angleInDegrees);
-
- cv::Mat dst_gold;
- cv::phase(x, y, dst_gold, angleInDegrees);
-
- EXPECT_MAT_NEAR(dst_gold, dst, angleInDegrees ? 1e-2 : 1e-3);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_Core, Phase, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- testing::Values(AngleInDegrees(false), AngleInDegrees(true)),
- WHOLE_SUBMAT));
-
-////////////////////////////////////////////////////////////////////////////////
-// CartToPolar
-
-PARAM_TEST_CASE(CartToPolar, cv::gpu::DeviceInfo, cv::Size, AngleInDegrees, UseRoi)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- bool angleInDegrees;
- bool useRoi;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- angleInDegrees = GET_PARAM(2);
- useRoi = GET_PARAM(3);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(CartToPolar, Accuracy)
-{
- cv::Mat x = randomMat(size, CV_32FC1);
- cv::Mat y = randomMat(size, CV_32FC1);
-
- cv::gpu::GpuMat mag = createMat(size, CV_32FC1, useRoi);
- cv::gpu::GpuMat angle = createMat(size, CV_32FC1, useRoi);
- cv::gpu::cartToPolar(loadMat(x, useRoi), loadMat(y, useRoi), mag, angle, angleInDegrees);
-
- cv::Mat mag_gold;
- cv::Mat angle_gold;
- cv::cartToPolar(x, y, mag_gold, angle_gold, angleInDegrees);
-
- EXPECT_MAT_NEAR(mag_gold, mag, 1e-4);
- EXPECT_MAT_NEAR(angle_gold, angle, angleInDegrees ? 1e-2 : 1e-3);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_Core, CartToPolar, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, CartToPolar, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(AngleInDegrees(false), AngleInDegrees(true)),
EXPECT_MAT_NEAR(y_gold, y, 1e-4);
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, PolarToCart, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, PolarToCart, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(AngleInDegrees(false), AngleInDegrees(true)),
WHOLE_SUBMAT));
-////////////////////////////////////////////////////////////////////////////////
-// MeanStdDev
-
-PARAM_TEST_CASE(MeanStdDev, cv::gpu::DeviceInfo, cv::Size, UseRoi)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- bool useRoi;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- useRoi = GET_PARAM(2);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(MeanStdDev, Accuracy)
-{
- cv::Mat src = randomMat(size, CV_8UC1);
-
- if (!supportFeature(devInfo, cv::gpu::FEATURE_SET_COMPUTE_13))
- {
- try
- {
- cv::Scalar mean;
- cv::Scalar stddev;
- cv::gpu::meanStdDev(loadMat(src, useRoi), mean, stddev);
- }
- catch (const cv::Exception& e)
- {
- ASSERT_EQ(cv::Error::StsNotImplemented, e.code);
- }
- }
- else
- {
- cv::Scalar mean;
- cv::Scalar stddev;
- cv::gpu::meanStdDev(loadMat(src, useRoi), mean, stddev);
-
- cv::Scalar mean_gold;
- cv::Scalar stddev_gold;
- cv::meanStdDev(src, mean_gold, stddev_gold);
-
- EXPECT_SCALAR_NEAR(mean_gold, mean, 1e-5);
- EXPECT_SCALAR_NEAR(stddev_gold, stddev, 1e-5);
- }
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_Core, MeanStdDev, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- WHOLE_SUBMAT));
+///////////////////////////////////////////////////////////////////////////////////////////////////////
+// Threshold
-////////////////////////////////////////////////////////////////////////////////
-// Norm
+CV_ENUM(ThreshOp, cv::THRESH_BINARY, cv::THRESH_BINARY_INV, cv::THRESH_TRUNC, cv::THRESH_TOZERO, cv::THRESH_TOZERO_INV)
+#define ALL_THRESH_OPS testing::Values(ThreshOp(cv::THRESH_BINARY), ThreshOp(cv::THRESH_BINARY_INV), ThreshOp(cv::THRESH_TRUNC), ThreshOp(cv::THRESH_TOZERO), ThreshOp(cv::THRESH_TOZERO_INV))
-PARAM_TEST_CASE(Norm, cv::gpu::DeviceInfo, cv::Size, MatDepth, NormCode, UseRoi)
+PARAM_TEST_CASE(Threshold, cv::gpu::DeviceInfo, cv::Size, MatType, ThreshOp, UseRoi)
{
cv::gpu::DeviceInfo devInfo;
cv::Size size;
- int depth;
- int normCode;
+ int type;
+ int threshOp;
bool useRoi;
virtual void SetUp()
{
devInfo = GET_PARAM(0);
size = GET_PARAM(1);
- depth = GET_PARAM(2);
- normCode = GET_PARAM(3);
+ type = GET_PARAM(2);
+ threshOp = GET_PARAM(3);
useRoi = GET_PARAM(4);
cv::gpu::setDevice(devInfo.deviceID());
}
};
-GPU_TEST_P(Norm, Accuracy)
-{
- cv::Mat src = randomMat(size, depth);
- cv::Mat mask = randomMat(size, CV_8UC1, 0, 2);
-
- cv::gpu::GpuMat d_buf;
- double val = cv::gpu::norm(loadMat(src, useRoi), normCode, loadMat(mask, useRoi), d_buf);
-
- double val_gold = cv::norm(src, normCode, mask);
-
- EXPECT_NEAR(val_gold, val, depth < CV_32F ? 0.0 : 1.0);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_Core, Norm, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- testing::Values(MatDepth(CV_8U),
- MatDepth(CV_8S),
- MatDepth(CV_16U),
- MatDepth(CV_16S),
- MatDepth(CV_32S),
- MatDepth(CV_32F)),
- testing::Values(NormCode(cv::NORM_L1), NormCode(cv::NORM_L2), NormCode(cv::NORM_INF)),
- WHOLE_SUBMAT));
-
-////////////////////////////////////////////////////////////////////////////////
-// normDiff
-
-PARAM_TEST_CASE(NormDiff, cv::gpu::DeviceInfo, cv::Size, NormCode, UseRoi)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- int normCode;
- bool useRoi;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- normCode = GET_PARAM(2);
- useRoi = GET_PARAM(3);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(NormDiff, Accuracy)
+GPU_TEST_P(Threshold, Accuracy)
{
- cv::Mat src1 = randomMat(size, CV_8UC1);
- cv::Mat src2 = randomMat(size, CV_8UC1);
+ cv::Mat src = randomMat(size, type);
+ double maxVal = randomDouble(20.0, 127.0);
+ double thresh = randomDouble(0.0, maxVal);
- double val = cv::gpu::norm(loadMat(src1, useRoi), loadMat(src2, useRoi), normCode);
+ cv::gpu::GpuMat dst = createMat(src.size(), src.type(), useRoi);
+ cv::gpu::threshold(loadMat(src, useRoi), dst, thresh, maxVal, threshOp);
- double val_gold = cv::norm(src1, src2, normCode);
+ cv::Mat dst_gold;
+ cv::threshold(src, dst_gold, thresh, maxVal, threshOp);
- EXPECT_NEAR(val_gold, val, 0.0);
+ EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_Core, NormDiff, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Threshold, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
- testing::Values(NormCode(cv::NORM_L1), NormCode(cv::NORM_L2), NormCode(cv::NORM_INF)),
- WHOLE_SUBMAT));
-
-//////////////////////////////////////////////////////////////////////////////
-// Sum
-
-namespace
-{
- template <typename T>
- cv::Scalar absSumImpl(const cv::Mat& src)
- {
- const int cn = src.channels();
-
- cv::Scalar sum = cv::Scalar::all(0);
-
- for (int y = 0; y < src.rows; ++y)
- {
- for (int x = 0; x < src.cols; ++x)
- {
- for (int c = 0; c < cn; ++c)
- sum[c] += std::abs(src.at<T>(y, x * cn + c));
- }
- }
-
- return sum;
- }
-
- cv::Scalar absSumGold(const cv::Mat& src)
- {
- typedef cv::Scalar (*func_t)(const cv::Mat& src);
-
- static const func_t funcs[] =
- {
- absSumImpl<uchar>,
- absSumImpl<schar>,
- absSumImpl<ushort>,
- absSumImpl<short>,
- absSumImpl<int>,
- absSumImpl<float>,
- absSumImpl<double>
- };
-
- return funcs[src.depth()](src);
- }
-
- template <typename T>
- cv::Scalar sqrSumImpl(const cv::Mat& src)
- {
- const int cn = src.channels();
-
- cv::Scalar sum = cv::Scalar::all(0);
-
- for (int y = 0; y < src.rows; ++y)
- {
- for (int x = 0; x < src.cols; ++x)
- {
- for (int c = 0; c < cn; ++c)
- {
- const T val = src.at<T>(y, x * cn + c);
- sum[c] += val * val;
- }
- }
- }
-
- return sum;
- }
-
- cv::Scalar sqrSumGold(const cv::Mat& src)
- {
- typedef cv::Scalar (*func_t)(const cv::Mat& src);
-
- static const func_t funcs[] =
- {
- sqrSumImpl<uchar>,
- sqrSumImpl<schar>,
- sqrSumImpl<ushort>,
- sqrSumImpl<short>,
- sqrSumImpl<int>,
- sqrSumImpl<float>,
- sqrSumImpl<double>
- };
-
- return funcs[src.depth()](src);
- }
-}
-
-PARAM_TEST_CASE(Sum, cv::gpu::DeviceInfo, cv::Size, MatType, UseRoi)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- int type;
- bool useRoi;
-
- cv::Mat src;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- type = GET_PARAM(2);
- useRoi = GET_PARAM(3);
-
- cv::gpu::setDevice(devInfo.deviceID());
-
- src = randomMat(size, type, -128.0, 128.0);
- }
-};
-
-GPU_TEST_P(Sum, Simple)
-{
- cv::Scalar val = cv::gpu::sum(loadMat(src, useRoi));
-
- cv::Scalar val_gold = cv::sum(src);
-
- EXPECT_SCALAR_NEAR(val_gold, val, CV_MAT_DEPTH(type) < CV_32F ? 0.0 : 0.5);
-}
-
-GPU_TEST_P(Sum, Abs)
-{
- cv::Scalar val = cv::gpu::absSum(loadMat(src, useRoi));
-
- cv::Scalar val_gold = absSumGold(src);
-
- EXPECT_SCALAR_NEAR(val_gold, val, CV_MAT_DEPTH(type) < CV_32F ? 0.0 : 0.5);
-}
-
-GPU_TEST_P(Sum, Sqr)
-{
- cv::Scalar val = cv::gpu::sqrSum(loadMat(src, useRoi));
-
- cv::Scalar val_gold = sqrSumGold(src);
-
- EXPECT_SCALAR_NEAR(val_gold, val, CV_MAT_DEPTH(type) < CV_32F ? 0.0 : 0.5);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_Core, Sum, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- TYPES(CV_8U, CV_64F, 1, 4),
- WHOLE_SUBMAT));
-
-////////////////////////////////////////////////////////////////////////////////
-// MinMax
-
-PARAM_TEST_CASE(MinMax, cv::gpu::DeviceInfo, cv::Size, MatDepth, UseRoi)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- int depth;
- bool useRoi;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- depth = GET_PARAM(2);
- useRoi = GET_PARAM(3);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(MinMax, WithoutMask)
-{
- cv::Mat src = randomMat(size, depth);
-
- if (depth == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
- {
- try
- {
- double minVal, maxVal;
- cv::gpu::minMax(loadMat(src), &minVal, &maxVal);
- }
- catch (const cv::Exception& e)
- {
- ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
- }
- }
- else
- {
- double minVal, maxVal;
- cv::gpu::minMax(loadMat(src, useRoi), &minVal, &maxVal);
-
- double minVal_gold, maxVal_gold;
- minMaxLocGold(src, &minVal_gold, &maxVal_gold);
-
- EXPECT_DOUBLE_EQ(minVal_gold, minVal);
- EXPECT_DOUBLE_EQ(maxVal_gold, maxVal);
- }
-}
-
-GPU_TEST_P(MinMax, WithMask)
-{
- cv::Mat src = randomMat(size, depth);
- cv::Mat mask = randomMat(size, CV_8UC1, 0.0, 2.0);
-
- if (depth == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
- {
- try
- {
- double minVal, maxVal;
- cv::gpu::minMax(loadMat(src), &minVal, &maxVal, loadMat(mask));
- }
- catch (const cv::Exception& e)
- {
- ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
- }
- }
- else
- {
- double minVal, maxVal;
- cv::gpu::minMax(loadMat(src, useRoi), &minVal, &maxVal, loadMat(mask, useRoi));
-
- double minVal_gold, maxVal_gold;
- minMaxLocGold(src, &minVal_gold, &maxVal_gold, 0, 0, mask);
-
- EXPECT_DOUBLE_EQ(minVal_gold, minVal);
- EXPECT_DOUBLE_EQ(maxVal_gold, maxVal);
- }
-}
-
-GPU_TEST_P(MinMax, NullPtr)
-{
- cv::Mat src = randomMat(size, depth);
-
- if (depth == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
- {
- try
- {
- double minVal, maxVal;
- cv::gpu::minMax(loadMat(src), &minVal, 0);
- cv::gpu::minMax(loadMat(src), 0, &maxVal);
- }
- catch (const cv::Exception& e)
- {
- ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
- }
- }
- else
- {
- double minVal, maxVal;
- cv::gpu::minMax(loadMat(src, useRoi), &minVal, 0);
- cv::gpu::minMax(loadMat(src, useRoi), 0, &maxVal);
-
- double minVal_gold, maxVal_gold;
- minMaxLocGold(src, &minVal_gold, &maxVal_gold, 0, 0);
-
- EXPECT_DOUBLE_EQ(minVal_gold, minVal);
- EXPECT_DOUBLE_EQ(maxVal_gold, maxVal);
- }
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_Core, MinMax, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- ALL_DEPTH,
- WHOLE_SUBMAT));
-
-////////////////////////////////////////////////////////////////////////////////
-// MinMaxLoc
-
-namespace
-{
- template <typename T>
- void expectEqualImpl(const cv::Mat& src, cv::Point loc_gold, cv::Point loc)
- {
- EXPECT_EQ(src.at<T>(loc_gold.y, loc_gold.x), src.at<T>(loc.y, loc.x));
- }
-
- void expectEqual(const cv::Mat& src, cv::Point loc_gold, cv::Point loc)
- {
- typedef void (*func_t)(const cv::Mat& src, cv::Point loc_gold, cv::Point loc);
-
- static const func_t funcs[] =
- {
- expectEqualImpl<uchar>,
- expectEqualImpl<schar>,
- expectEqualImpl<ushort>,
- expectEqualImpl<short>,
- expectEqualImpl<int>,
- expectEqualImpl<float>,
- expectEqualImpl<double>
- };
-
- funcs[src.depth()](src, loc_gold, loc);
- }
-}
-
-PARAM_TEST_CASE(MinMaxLoc, cv::gpu::DeviceInfo, cv::Size, MatDepth, UseRoi)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- int depth;
- bool useRoi;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- depth = GET_PARAM(2);
- useRoi = GET_PARAM(3);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(MinMaxLoc, WithoutMask)
-{
- cv::Mat src = randomMat(size, depth);
-
- if (depth == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
- {
- try
- {
- double minVal, maxVal;
- cv::Point minLoc, maxLoc;
- cv::gpu::minMaxLoc(loadMat(src), &minVal, &maxVal, &minLoc, &maxLoc);
- }
- catch (const cv::Exception& e)
- {
- ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
- }
- }
- else
- {
- double minVal, maxVal;
- cv::Point minLoc, maxLoc;
- cv::gpu::minMaxLoc(loadMat(src, useRoi), &minVal, &maxVal, &minLoc, &maxLoc);
-
- double minVal_gold, maxVal_gold;
- cv::Point minLoc_gold, maxLoc_gold;
- minMaxLocGold(src, &minVal_gold, &maxVal_gold, &minLoc_gold, &maxLoc_gold);
-
- EXPECT_DOUBLE_EQ(minVal_gold, minVal);
- EXPECT_DOUBLE_EQ(maxVal_gold, maxVal);
-
- expectEqual(src, minLoc_gold, minLoc);
- expectEqual(src, maxLoc_gold, maxLoc);
- }
-}
-
-GPU_TEST_P(MinMaxLoc, WithMask)
-{
- cv::Mat src = randomMat(size, depth);
- cv::Mat mask = randomMat(size, CV_8UC1, 0.0, 2.0);
-
- if (depth == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
- {
- try
- {
- double minVal, maxVal;
- cv::Point minLoc, maxLoc;
- cv::gpu::minMaxLoc(loadMat(src), &minVal, &maxVal, &minLoc, &maxLoc, loadMat(mask));
- }
- catch (const cv::Exception& e)
- {
- ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
- }
- }
- else
- {
- double minVal, maxVal;
- cv::Point minLoc, maxLoc;
- cv::gpu::minMaxLoc(loadMat(src, useRoi), &minVal, &maxVal, &minLoc, &maxLoc, loadMat(mask, useRoi));
-
- double minVal_gold, maxVal_gold;
- cv::Point minLoc_gold, maxLoc_gold;
- minMaxLocGold(src, &minVal_gold, &maxVal_gold, &minLoc_gold, &maxLoc_gold, mask);
-
- EXPECT_DOUBLE_EQ(minVal_gold, minVal);
- EXPECT_DOUBLE_EQ(maxVal_gold, maxVal);
-
- expectEqual(src, minLoc_gold, minLoc);
- expectEqual(src, maxLoc_gold, maxLoc);
- }
-}
-
-GPU_TEST_P(MinMaxLoc, NullPtr)
-{
- cv::Mat src = randomMat(size, depth);
-
- if (depth == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
- {
- try
- {
- double minVal, maxVal;
- cv::Point minLoc, maxLoc;
- cv::gpu::minMaxLoc(loadMat(src, useRoi), &minVal, 0, 0, 0);
- cv::gpu::minMaxLoc(loadMat(src, useRoi), 0, &maxVal, 0, 0);
- cv::gpu::minMaxLoc(loadMat(src, useRoi), 0, 0, &minLoc, 0);
- cv::gpu::minMaxLoc(loadMat(src, useRoi), 0, 0, 0, &maxLoc);
- }
- catch (const cv::Exception& e)
- {
- ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
- }
- }
- else
- {
- double minVal, maxVal;
- cv::Point minLoc, maxLoc;
- cv::gpu::minMaxLoc(loadMat(src, useRoi), &minVal, 0, 0, 0);
- cv::gpu::minMaxLoc(loadMat(src, useRoi), 0, &maxVal, 0, 0);
- cv::gpu::minMaxLoc(loadMat(src, useRoi), 0, 0, &minLoc, 0);
- cv::gpu::minMaxLoc(loadMat(src, useRoi), 0, 0, 0, &maxLoc);
-
- double minVal_gold, maxVal_gold;
- cv::Point minLoc_gold, maxLoc_gold;
- minMaxLocGold(src, &minVal_gold, &maxVal_gold, &minLoc_gold, &maxLoc_gold);
-
- EXPECT_DOUBLE_EQ(minVal_gold, minVal);
- EXPECT_DOUBLE_EQ(maxVal_gold, maxVal);
-
- expectEqual(src, minLoc_gold, minLoc);
- expectEqual(src, maxLoc_gold, maxLoc);
- }
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_Core, MinMaxLoc, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- ALL_DEPTH,
- WHOLE_SUBMAT));
-
-////////////////////////////////////////////////////////////////////////////
-// CountNonZero
-
-PARAM_TEST_CASE(CountNonZero, cv::gpu::DeviceInfo, cv::Size, MatDepth, UseRoi)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- int depth;
- bool useRoi;
-
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- depth = GET_PARAM(2);
- useRoi = GET_PARAM(3);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(CountNonZero, Accuracy)
-{
- cv::Mat srcBase = randomMat(size, CV_8U, 0.0, 1.5);
- cv::Mat src;
- srcBase.convertTo(src, depth);
-
- if (depth == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
- {
- try
- {
- cv::gpu::countNonZero(loadMat(src));
- }
- catch (const cv::Exception& e)
- {
- ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
- }
- }
- else
- {
- int val = cv::gpu::countNonZero(loadMat(src, useRoi));
-
- int val_gold = cv::countNonZero(src);
-
- ASSERT_EQ(val_gold, val);
- }
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_Core, CountNonZero, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- ALL_DEPTH,
- WHOLE_SUBMAT));
-
-//////////////////////////////////////////////////////////////////////////////
-// Reduce
-
-CV_ENUM(ReduceCode, REDUCE_SUM, REDUCE_AVG, REDUCE_MAX, REDUCE_MIN)
-
-PARAM_TEST_CASE(Reduce, cv::gpu::DeviceInfo, cv::Size, MatDepth, Channels, ReduceCode, UseRoi)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- int depth;
- int channels;
- int reduceOp;
- bool useRoi;
-
- int type;
- int dst_depth;
- int dst_type;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- depth = GET_PARAM(2);
- channels = GET_PARAM(3);
- reduceOp = GET_PARAM(4);
- useRoi = GET_PARAM(5);
-
- cv::gpu::setDevice(devInfo.deviceID());
-
- type = CV_MAKE_TYPE(depth, channels);
-
- if (reduceOp == cv::REDUCE_MAX || reduceOp == cv::REDUCE_MIN)
- dst_depth = depth;
- else if (reduceOp == cv::REDUCE_SUM)
- dst_depth = depth == CV_8U ? CV_32S : depth < CV_64F ? CV_32F : depth;
- else
- dst_depth = depth < CV_32F ? CV_32F : depth;
-
- dst_type = CV_MAKE_TYPE(dst_depth, channels);
- }
-
-};
-
-GPU_TEST_P(Reduce, Rows)
-{
- cv::Mat src = randomMat(size, type);
-
- cv::gpu::GpuMat dst = createMat(cv::Size(src.cols, 1), dst_type, useRoi);
- cv::gpu::reduce(loadMat(src, useRoi), dst, 0, reduceOp, dst_depth);
-
- cv::Mat dst_gold;
- cv::reduce(src, dst_gold, 0, reduceOp, dst_depth);
-
- EXPECT_MAT_NEAR(dst_gold, dst, dst_depth < CV_32F ? 0.0 : 0.02);
-}
-
-GPU_TEST_P(Reduce, Cols)
-{
- cv::Mat src = randomMat(size, type);
-
- cv::gpu::GpuMat dst = createMat(cv::Size(src.rows, 1), dst_type, useRoi);
- cv::gpu::reduce(loadMat(src, useRoi), dst, 1, reduceOp, dst_depth);
-
- cv::Mat dst_gold;
- cv::reduce(src, dst_gold, 1, reduceOp, dst_depth);
- dst_gold.cols = dst_gold.rows;
- dst_gold.rows = 1;
- dst_gold.step = dst_gold.cols * dst_gold.elemSize();
-
- EXPECT_MAT_NEAR(dst_gold, dst, dst_depth < CV_32F ? 0.0 : 0.02);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_Core, Reduce, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- testing::Values(MatDepth(CV_8U),
- MatDepth(CV_16U),
- MatDepth(CV_16S),
- MatDepth(CV_32F),
- MatDepth(CV_64F)),
- ALL_CHANNELS,
- ReduceCode::all(),
- WHOLE_SUBMAT));
-
-//////////////////////////////////////////////////////////////////////////////
-// Normalize
-
-PARAM_TEST_CASE(Normalize, cv::gpu::DeviceInfo, cv::Size, MatDepth, NormCode, UseRoi)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- int type;
- int norm_type;
- bool useRoi;
-
- double alpha;
- double beta;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- type = GET_PARAM(2);
- norm_type = GET_PARAM(3);
- useRoi = GET_PARAM(4);
-
- cv::gpu::setDevice(devInfo.deviceID());
-
- alpha = 1;
- beta = 0;
- }
-
-};
-
-GPU_TEST_P(Normalize, WithOutMask)
-{
- cv::Mat src = randomMat(size, type);
-
- cv::gpu::GpuMat dst = createMat(size, type, useRoi);
- cv::gpu::normalize(loadMat(src, useRoi), dst, alpha, beta, norm_type, type);
-
- cv::Mat dst_gold;
- cv::normalize(src, dst_gold, alpha, beta, norm_type, type);
-
- EXPECT_MAT_NEAR(dst_gold, dst, 1e-6);
-}
-
-GPU_TEST_P(Normalize, WithMask)
-{
- cv::Mat src = randomMat(size, type);
- cv::Mat mask = randomMat(size, CV_8UC1, 0, 2);
-
- cv::gpu::GpuMat dst = createMat(size, type, useRoi);
- dst.setTo(cv::Scalar::all(0));
- cv::gpu::normalize(loadMat(src, useRoi), dst, alpha, beta, norm_type, type, loadMat(mask, useRoi));
-
- cv::Mat dst_gold(size, type);
- dst_gold.setTo(cv::Scalar::all(0));
- cv::normalize(src, dst_gold, alpha, beta, norm_type, type, mask);
-
- EXPECT_MAT_NEAR(dst_gold, dst, 1e-6);
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_Core, Normalize, testing::Combine(
- ALL_DEVICES,
- DIFFERENT_SIZES,
- ALL_DEPTH,
- testing::Values(NormCode(cv::NORM_L1), NormCode(cv::NORM_L2), NormCode(cv::NORM_INF), NormCode(cv::NORM_MINMAX)),
+ testing::Values(MatType(CV_8UC1), MatType(CV_16SC1), MatType(CV_32FC1)),
+ ALL_THRESH_OPS,
WHOLE_SUBMAT));
#endif // HAVE_CUDA
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+CV_GPU_TEST_MAIN("gpu")
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifdef __GNUC__
+# pragma GCC diagnostic ignored "-Wmissing-declarations"
+# if defined __clang__ || defined __APPLE__
+# pragma GCC diagnostic ignored "-Wmissing-prototypes"
+# pragma GCC diagnostic ignored "-Wextra"
+# endif
+#endif
+
+#ifndef __OPENCV_TEST_PRECOMP_HPP__
+#define __OPENCV_TEST_PRECOMP_HPP__
+
+#include "opencv2/ts.hpp"
+#include "opencv2/ts/gpu_test.hpp"
+
+#include "opencv2/gpuarithm.hpp"
+#include "opencv2/core.hpp"
+#include "opencv2/imgproc.hpp"
+
+#endif
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+#ifdef HAVE_CUDA
+
+using namespace cvtest;
+
+////////////////////////////////////////////////////////////////////////////////
+// Norm
+
+PARAM_TEST_CASE(Norm, cv::gpu::DeviceInfo, cv::Size, MatDepth, NormCode, UseRoi)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ int depth;
+ int normCode;
+ bool useRoi;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ depth = GET_PARAM(2);
+ normCode = GET_PARAM(3);
+ useRoi = GET_PARAM(4);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(Norm, Accuracy)
+{
+ cv::Mat src = randomMat(size, depth);
+ cv::Mat mask = randomMat(size, CV_8UC1, 0, 2);
+
+ cv::gpu::GpuMat d_buf;
+ double val = cv::gpu::norm(loadMat(src, useRoi), normCode, loadMat(mask, useRoi), d_buf);
+
+ double val_gold = cv::norm(src, normCode, mask);
+
+ EXPECT_NEAR(val_gold, val, depth < CV_32F ? 0.0 : 1.0);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Norm, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ testing::Values(MatDepth(CV_8U),
+ MatDepth(CV_8S),
+ MatDepth(CV_16U),
+ MatDepth(CV_16S),
+ MatDepth(CV_32S),
+ MatDepth(CV_32F)),
+ testing::Values(NormCode(cv::NORM_L1), NormCode(cv::NORM_L2), NormCode(cv::NORM_INF)),
+ WHOLE_SUBMAT));
+
+////////////////////////////////////////////////////////////////////////////////
+// normDiff
+
+PARAM_TEST_CASE(NormDiff, cv::gpu::DeviceInfo, cv::Size, NormCode, UseRoi)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ int normCode;
+ bool useRoi;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ normCode = GET_PARAM(2);
+ useRoi = GET_PARAM(3);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(NormDiff, Accuracy)
+{
+ cv::Mat src1 = randomMat(size, CV_8UC1);
+ cv::Mat src2 = randomMat(size, CV_8UC1);
+
+ double val = cv::gpu::norm(loadMat(src1, useRoi), loadMat(src2, useRoi), normCode);
+
+ double val_gold = cv::norm(src1, src2, normCode);
+
+ EXPECT_NEAR(val_gold, val, 0.0);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, NormDiff, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ testing::Values(NormCode(cv::NORM_L1), NormCode(cv::NORM_L2), NormCode(cv::NORM_INF)),
+ WHOLE_SUBMAT));
+
+//////////////////////////////////////////////////////////////////////////////
+// Sum
+
+namespace
+{
+ template <typename T>
+ cv::Scalar absSumImpl(const cv::Mat& src)
+ {
+ const int cn = src.channels();
+
+ cv::Scalar sum = cv::Scalar::all(0);
+
+ for (int y = 0; y < src.rows; ++y)
+ {
+ for (int x = 0; x < src.cols; ++x)
+ {
+ for (int c = 0; c < cn; ++c)
+ sum[c] += std::abs(src.at<T>(y, x * cn + c));
+ }
+ }
+
+ return sum;
+ }
+
+ cv::Scalar absSumGold(const cv::Mat& src)
+ {
+ typedef cv::Scalar (*func_t)(const cv::Mat& src);
+
+ static const func_t funcs[] =
+ {
+ absSumImpl<uchar>,
+ absSumImpl<schar>,
+ absSumImpl<ushort>,
+ absSumImpl<short>,
+ absSumImpl<int>,
+ absSumImpl<float>,
+ absSumImpl<double>
+ };
+
+ return funcs[src.depth()](src);
+ }
+
+ template <typename T>
+ cv::Scalar sqrSumImpl(const cv::Mat& src)
+ {
+ const int cn = src.channels();
+
+ cv::Scalar sum = cv::Scalar::all(0);
+
+ for (int y = 0; y < src.rows; ++y)
+ {
+ for (int x = 0; x < src.cols; ++x)
+ {
+ for (int c = 0; c < cn; ++c)
+ {
+ const T val = src.at<T>(y, x * cn + c);
+ sum[c] += val * val;
+ }
+ }
+ }
+
+ return sum;
+ }
+
+ cv::Scalar sqrSumGold(const cv::Mat& src)
+ {
+ typedef cv::Scalar (*func_t)(const cv::Mat& src);
+
+ static const func_t funcs[] =
+ {
+ sqrSumImpl<uchar>,
+ sqrSumImpl<schar>,
+ sqrSumImpl<ushort>,
+ sqrSumImpl<short>,
+ sqrSumImpl<int>,
+ sqrSumImpl<float>,
+ sqrSumImpl<double>
+ };
+
+ return funcs[src.depth()](src);
+ }
+}
+
+PARAM_TEST_CASE(Sum, cv::gpu::DeviceInfo, cv::Size, MatType, UseRoi)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ int type;
+ bool useRoi;
+
+ cv::Mat src;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ type = GET_PARAM(2);
+ useRoi = GET_PARAM(3);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+
+ src = randomMat(size, type, -128.0, 128.0);
+ }
+};
+
+GPU_TEST_P(Sum, Simple)
+{
+ cv::Scalar val = cv::gpu::sum(loadMat(src, useRoi));
+
+ cv::Scalar val_gold = cv::sum(src);
+
+ EXPECT_SCALAR_NEAR(val_gold, val, CV_MAT_DEPTH(type) < CV_32F ? 0.0 : 0.5);
+}
+
+GPU_TEST_P(Sum, Abs)
+{
+ cv::Scalar val = cv::gpu::absSum(loadMat(src, useRoi));
+
+ cv::Scalar val_gold = absSumGold(src);
+
+ EXPECT_SCALAR_NEAR(val_gold, val, CV_MAT_DEPTH(type) < CV_32F ? 0.0 : 0.5);
+}
+
+GPU_TEST_P(Sum, Sqr)
+{
+ cv::Scalar val = cv::gpu::sqrSum(loadMat(src, useRoi));
+
+ cv::Scalar val_gold = sqrSumGold(src);
+
+ EXPECT_SCALAR_NEAR(val_gold, val, CV_MAT_DEPTH(type) < CV_32F ? 0.0 : 0.5);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Sum, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ TYPES(CV_8U, CV_64F, 1, 4),
+ WHOLE_SUBMAT));
+
+////////////////////////////////////////////////////////////////////////////////
+// MinMax
+
+PARAM_TEST_CASE(MinMax, cv::gpu::DeviceInfo, cv::Size, MatDepth, UseRoi)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ int depth;
+ bool useRoi;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ depth = GET_PARAM(2);
+ useRoi = GET_PARAM(3);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(MinMax, WithoutMask)
+{
+ cv::Mat src = randomMat(size, depth);
+
+ if (depth == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
+ {
+ try
+ {
+ double minVal, maxVal;
+ cv::gpu::minMax(loadMat(src), &minVal, &maxVal);
+ }
+ catch (const cv::Exception& e)
+ {
+ ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
+ }
+ }
+ else
+ {
+ double minVal, maxVal;
+ cv::gpu::minMax(loadMat(src, useRoi), &minVal, &maxVal);
+
+ double minVal_gold, maxVal_gold;
+ minMaxLocGold(src, &minVal_gold, &maxVal_gold);
+
+ EXPECT_DOUBLE_EQ(minVal_gold, minVal);
+ EXPECT_DOUBLE_EQ(maxVal_gold, maxVal);
+ }
+}
+
+GPU_TEST_P(MinMax, WithMask)
+{
+ cv::Mat src = randomMat(size, depth);
+ cv::Mat mask = randomMat(size, CV_8UC1, 0.0, 2.0);
+
+ if (depth == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
+ {
+ try
+ {
+ double minVal, maxVal;
+ cv::gpu::minMax(loadMat(src), &minVal, &maxVal, loadMat(mask));
+ }
+ catch (const cv::Exception& e)
+ {
+ ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
+ }
+ }
+ else
+ {
+ double minVal, maxVal;
+ cv::gpu::minMax(loadMat(src, useRoi), &minVal, &maxVal, loadMat(mask, useRoi));
+
+ double minVal_gold, maxVal_gold;
+ minMaxLocGold(src, &minVal_gold, &maxVal_gold, 0, 0, mask);
+
+ EXPECT_DOUBLE_EQ(minVal_gold, minVal);
+ EXPECT_DOUBLE_EQ(maxVal_gold, maxVal);
+ }
+}
+
+GPU_TEST_P(MinMax, NullPtr)
+{
+ cv::Mat src = randomMat(size, depth);
+
+ if (depth == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
+ {
+ try
+ {
+ double minVal, maxVal;
+ cv::gpu::minMax(loadMat(src), &minVal, 0);
+ cv::gpu::minMax(loadMat(src), 0, &maxVal);
+ }
+ catch (const cv::Exception& e)
+ {
+ ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
+ }
+ }
+ else
+ {
+ double minVal, maxVal;
+ cv::gpu::minMax(loadMat(src, useRoi), &minVal, 0);
+ cv::gpu::minMax(loadMat(src, useRoi), 0, &maxVal);
+
+ double minVal_gold, maxVal_gold;
+ minMaxLocGold(src, &minVal_gold, &maxVal_gold, 0, 0);
+
+ EXPECT_DOUBLE_EQ(minVal_gold, minVal);
+ EXPECT_DOUBLE_EQ(maxVal_gold, maxVal);
+ }
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, MinMax, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ ALL_DEPTH,
+ WHOLE_SUBMAT));
+
+////////////////////////////////////////////////////////////////////////////////
+// MinMaxLoc
+
+namespace
+{
+ template <typename T>
+ void expectEqualImpl(const cv::Mat& src, cv::Point loc_gold, cv::Point loc)
+ {
+ EXPECT_EQ(src.at<T>(loc_gold.y, loc_gold.x), src.at<T>(loc.y, loc.x));
+ }
+
+ void expectEqual(const cv::Mat& src, cv::Point loc_gold, cv::Point loc)
+ {
+ typedef void (*func_t)(const cv::Mat& src, cv::Point loc_gold, cv::Point loc);
+
+ static const func_t funcs[] =
+ {
+ expectEqualImpl<uchar>,
+ expectEqualImpl<schar>,
+ expectEqualImpl<ushort>,
+ expectEqualImpl<short>,
+ expectEqualImpl<int>,
+ expectEqualImpl<float>,
+ expectEqualImpl<double>
+ };
+
+ funcs[src.depth()](src, loc_gold, loc);
+ }
+}
+
+PARAM_TEST_CASE(MinMaxLoc, cv::gpu::DeviceInfo, cv::Size, MatDepth, UseRoi)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ int depth;
+ bool useRoi;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ depth = GET_PARAM(2);
+ useRoi = GET_PARAM(3);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(MinMaxLoc, WithoutMask)
+{
+ cv::Mat src = randomMat(size, depth);
+
+ if (depth == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
+ {
+ try
+ {
+ double minVal, maxVal;
+ cv::Point minLoc, maxLoc;
+ cv::gpu::minMaxLoc(loadMat(src), &minVal, &maxVal, &minLoc, &maxLoc);
+ }
+ catch (const cv::Exception& e)
+ {
+ ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
+ }
+ }
+ else
+ {
+ double minVal, maxVal;
+ cv::Point minLoc, maxLoc;
+ cv::gpu::minMaxLoc(loadMat(src, useRoi), &minVal, &maxVal, &minLoc, &maxLoc);
+
+ double minVal_gold, maxVal_gold;
+ cv::Point minLoc_gold, maxLoc_gold;
+ minMaxLocGold(src, &minVal_gold, &maxVal_gold, &minLoc_gold, &maxLoc_gold);
+
+ EXPECT_DOUBLE_EQ(minVal_gold, minVal);
+ EXPECT_DOUBLE_EQ(maxVal_gold, maxVal);
+
+ expectEqual(src, minLoc_gold, minLoc);
+ expectEqual(src, maxLoc_gold, maxLoc);
+ }
+}
+
+GPU_TEST_P(MinMaxLoc, WithMask)
+{
+ cv::Mat src = randomMat(size, depth);
+ cv::Mat mask = randomMat(size, CV_8UC1, 0.0, 2.0);
+
+ if (depth == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
+ {
+ try
+ {
+ double minVal, maxVal;
+ cv::Point minLoc, maxLoc;
+ cv::gpu::minMaxLoc(loadMat(src), &minVal, &maxVal, &minLoc, &maxLoc, loadMat(mask));
+ }
+ catch (const cv::Exception& e)
+ {
+ ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
+ }
+ }
+ else
+ {
+ double minVal, maxVal;
+ cv::Point minLoc, maxLoc;
+ cv::gpu::minMaxLoc(loadMat(src, useRoi), &minVal, &maxVal, &minLoc, &maxLoc, loadMat(mask, useRoi));
+
+ double minVal_gold, maxVal_gold;
+ cv::Point minLoc_gold, maxLoc_gold;
+ minMaxLocGold(src, &minVal_gold, &maxVal_gold, &minLoc_gold, &maxLoc_gold, mask);
+
+ EXPECT_DOUBLE_EQ(minVal_gold, minVal);
+ EXPECT_DOUBLE_EQ(maxVal_gold, maxVal);
+
+ expectEqual(src, minLoc_gold, minLoc);
+ expectEqual(src, maxLoc_gold, maxLoc);
+ }
+}
+
+GPU_TEST_P(MinMaxLoc, NullPtr)
+{
+ cv::Mat src = randomMat(size, depth);
+
+ if (depth == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
+ {
+ try
+ {
+ double minVal, maxVal;
+ cv::Point minLoc, maxLoc;
+ cv::gpu::minMaxLoc(loadMat(src, useRoi), &minVal, 0, 0, 0);
+ cv::gpu::minMaxLoc(loadMat(src, useRoi), 0, &maxVal, 0, 0);
+ cv::gpu::minMaxLoc(loadMat(src, useRoi), 0, 0, &minLoc, 0);
+ cv::gpu::minMaxLoc(loadMat(src, useRoi), 0, 0, 0, &maxLoc);
+ }
+ catch (const cv::Exception& e)
+ {
+ ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
+ }
+ }
+ else
+ {
+ double minVal, maxVal;
+ cv::Point minLoc, maxLoc;
+ cv::gpu::minMaxLoc(loadMat(src, useRoi), &minVal, 0, 0, 0);
+ cv::gpu::minMaxLoc(loadMat(src, useRoi), 0, &maxVal, 0, 0);
+ cv::gpu::minMaxLoc(loadMat(src, useRoi), 0, 0, &minLoc, 0);
+ cv::gpu::minMaxLoc(loadMat(src, useRoi), 0, 0, 0, &maxLoc);
+
+ double minVal_gold, maxVal_gold;
+ cv::Point minLoc_gold, maxLoc_gold;
+ minMaxLocGold(src, &minVal_gold, &maxVal_gold, &minLoc_gold, &maxLoc_gold);
+
+ EXPECT_DOUBLE_EQ(minVal_gold, minVal);
+ EXPECT_DOUBLE_EQ(maxVal_gold, maxVal);
+
+ expectEqual(src, minLoc_gold, minLoc);
+ expectEqual(src, maxLoc_gold, maxLoc);
+ }
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, MinMaxLoc, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ ALL_DEPTH,
+ WHOLE_SUBMAT));
+
+////////////////////////////////////////////////////////////////////////////
+// CountNonZero
+
+PARAM_TEST_CASE(CountNonZero, cv::gpu::DeviceInfo, cv::Size, MatDepth, UseRoi)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ int depth;
+ bool useRoi;
+
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ depth = GET_PARAM(2);
+ useRoi = GET_PARAM(3);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(CountNonZero, Accuracy)
+{
+ cv::Mat srcBase = randomMat(size, CV_8U, 0.0, 1.5);
+ cv::Mat src;
+ srcBase.convertTo(src, depth);
+
+ if (depth == CV_64F && !supportFeature(devInfo, cv::gpu::NATIVE_DOUBLE))
+ {
+ try
+ {
+ cv::gpu::countNonZero(loadMat(src));
+ }
+ catch (const cv::Exception& e)
+ {
+ ASSERT_EQ(cv::Error::StsUnsupportedFormat, e.code);
+ }
+ }
+ else
+ {
+ int val = cv::gpu::countNonZero(loadMat(src, useRoi));
+
+ int val_gold = cv::countNonZero(src);
+
+ ASSERT_EQ(val_gold, val);
+ }
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, CountNonZero, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ ALL_DEPTH,
+ WHOLE_SUBMAT));
+
+//////////////////////////////////////////////////////////////////////////////
+// Reduce
+
+CV_ENUM(ReduceCode, cv::REDUCE_SUM, cv::REDUCE_AVG, cv::REDUCE_MAX, cv::REDUCE_MIN)
+#define ALL_REDUCE_CODES testing::Values(ReduceCode(cv::REDUCE_SUM), ReduceCode(cv::REDUCE_AVG), ReduceCode(cv::REDUCE_MAX), ReduceCode(cv::REDUCE_MIN))
+
+PARAM_TEST_CASE(Reduce, cv::gpu::DeviceInfo, cv::Size, MatDepth, Channels, ReduceCode, UseRoi)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ int depth;
+ int channels;
+ int reduceOp;
+ bool useRoi;
+
+ int type;
+ int dst_depth;
+ int dst_type;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ depth = GET_PARAM(2);
+ channels = GET_PARAM(3);
+ reduceOp = GET_PARAM(4);
+ useRoi = GET_PARAM(5);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+
+ type = CV_MAKE_TYPE(depth, channels);
+
+ if (reduceOp == cv::REDUCE_MAX || reduceOp == cv::REDUCE_MIN)
+ dst_depth = depth;
+ else if (reduceOp == cv::REDUCE_SUM)
+ dst_depth = depth == CV_8U ? CV_32S : depth < CV_64F ? CV_32F : depth;
+ else
+ dst_depth = depth < CV_32F ? CV_32F : depth;
+
+ dst_type = CV_MAKE_TYPE(dst_depth, channels);
+ }
+
+};
+
+GPU_TEST_P(Reduce, Rows)
+{
+ cv::Mat src = randomMat(size, type);
+
+ cv::gpu::GpuMat dst = createMat(cv::Size(src.cols, 1), dst_type, useRoi);
+ cv::gpu::reduce(loadMat(src, useRoi), dst, 0, reduceOp, dst_depth);
+
+ cv::Mat dst_gold;
+ cv::reduce(src, dst_gold, 0, reduceOp, dst_depth);
+
+ EXPECT_MAT_NEAR(dst_gold, dst, dst_depth < CV_32F ? 0.0 : 0.02);
+}
+
+GPU_TEST_P(Reduce, Cols)
+{
+ cv::Mat src = randomMat(size, type);
+
+ cv::gpu::GpuMat dst = createMat(cv::Size(src.rows, 1), dst_type, useRoi);
+ cv::gpu::reduce(loadMat(src, useRoi), dst, 1, reduceOp, dst_depth);
+
+ cv::Mat dst_gold;
+ cv::reduce(src, dst_gold, 1, reduceOp, dst_depth);
+ dst_gold.cols = dst_gold.rows;
+ dst_gold.rows = 1;
+ dst_gold.step = dst_gold.cols * dst_gold.elemSize();
+
+ EXPECT_MAT_NEAR(dst_gold, dst, dst_depth < CV_32F ? 0.0 : 0.02);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Reduce, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ testing::Values(MatDepth(CV_8U),
+ MatDepth(CV_16U),
+ MatDepth(CV_16S),
+ MatDepth(CV_32F),
+ MatDepth(CV_64F)),
+ ALL_CHANNELS,
+ ALL_REDUCE_CODES,
+ WHOLE_SUBMAT));
+
+//////////////////////////////////////////////////////////////////////////////
+// Normalize
+
+PARAM_TEST_CASE(Normalize, cv::gpu::DeviceInfo, cv::Size, MatDepth, NormCode, UseRoi)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ int type;
+ int norm_type;
+ bool useRoi;
+
+ double alpha;
+ double beta;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ type = GET_PARAM(2);
+ norm_type = GET_PARAM(3);
+ useRoi = GET_PARAM(4);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+
+ alpha = 1;
+ beta = 0;
+ }
+
+};
+
+GPU_TEST_P(Normalize, WithOutMask)
+{
+ cv::Mat src = randomMat(size, type);
+
+ cv::gpu::GpuMat dst = createMat(size, type, useRoi);
+ cv::gpu::normalize(loadMat(src, useRoi), dst, alpha, beta, norm_type, type);
+
+ cv::Mat dst_gold;
+ cv::normalize(src, dst_gold, alpha, beta, norm_type, type);
+
+ EXPECT_MAT_NEAR(dst_gold, dst, 1e-6);
+}
+
+GPU_TEST_P(Normalize, WithMask)
+{
+ cv::Mat src = randomMat(size, type);
+ cv::Mat mask = randomMat(size, CV_8UC1, 0, 2);
+
+ cv::gpu::GpuMat dst = createMat(size, type, useRoi);
+ dst.setTo(cv::Scalar::all(0));
+ cv::gpu::normalize(loadMat(src, useRoi), dst, alpha, beta, norm_type, type, loadMat(mask, useRoi));
+
+ cv::Mat dst_gold(size, type);
+ dst_gold.setTo(cv::Scalar::all(0));
+ cv::normalize(src, dst_gold, alpha, beta, norm_type, type, mask);
+
+ EXPECT_MAT_NEAR(dst_gold, dst, 1e-6);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, Normalize, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ ALL_DEPTH,
+ testing::Values(NormCode(cv::NORM_L1), NormCode(cv::NORM_L2), NormCode(cv::NORM_INF), NormCode(cv::NORM_MINMAX)),
+ WHOLE_SUBMAT));
+
+////////////////////////////////////////////////////////////////////////////////
+// MeanStdDev
+
+PARAM_TEST_CASE(MeanStdDev, cv::gpu::DeviceInfo, cv::Size, UseRoi)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ bool useRoi;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ useRoi = GET_PARAM(2);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(MeanStdDev, Accuracy)
+{
+ cv::Mat src = randomMat(size, CV_8UC1);
+
+ if (!supportFeature(devInfo, cv::gpu::FEATURE_SET_COMPUTE_13))
+ {
+ try
+ {
+ cv::Scalar mean;
+ cv::Scalar stddev;
+ cv::gpu::meanStdDev(loadMat(src, useRoi), mean, stddev);
+ }
+ catch (const cv::Exception& e)
+ {
+ ASSERT_EQ(cv::Error::StsNotImplemented, e.code);
+ }
+ }
+ else
+ {
+ cv::Scalar mean;
+ cv::Scalar stddev;
+ cv::gpu::meanStdDev(loadMat(src, useRoi), mean, stddev);
+
+ cv::Scalar mean_gold;
+ cv::Scalar stddev_gold;
+ cv::meanStdDev(src, mean_gold, stddev_gold);
+
+ EXPECT_SCALAR_NEAR(mean_gold, mean, 1e-5);
+ EXPECT_SCALAR_NEAR(stddev_gold, stddev, 1e-5);
+ }
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Arithm, MeanStdDev, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ WHOLE_SUBMAT));
+
+#endif // HAVE_CUDA
--- /dev/null
+if(ANDROID OR IOS)
+ ocv_module_disable(gpubgsegm)
+endif()
+
+set(the_description "GPU-accelerated Background Segmentation")
+
+ocv_warnings_disable(CMAKE_CXX_FLAGS /wd4127 /wd4324 /wd4512 -Wundef -Wmissing-declarations)
+
+ocv_define_module(gpubgsegm opencv_video opencv_imgproc opencv_legacy opencv_gpuarithm opencv_gpufilters opencv_gpuimgproc)
--- /dev/null
+Background Segmentation
+=======================
+
+.. highlight:: cpp
+
+
+
+gpu::FGDStatModel
+-----------------
+.. ocv:class:: gpu::FGDStatModel
+
+Class used for background/foreground segmentation. ::
+
+ class FGDStatModel
+ {
+ public:
+ struct Params
+ {
+ ...
+ };
+
+ explicit FGDStatModel(int out_cn = 3);
+ explicit FGDStatModel(const cv::gpu::GpuMat& firstFrame, const Params& params = Params(), int out_cn = 3);
+
+ ~FGDStatModel();
+
+ void create(const cv::gpu::GpuMat& firstFrame, const Params& params = Params());
+ void release();
+
+ int update(const cv::gpu::GpuMat& curFrame);
+
+ //8UC3 or 8UC4 reference background image
+ cv::gpu::GpuMat background;
+
+ //8UC1 foreground image
+ cv::gpu::GpuMat foreground;
+
+ std::vector< std::vector<cv::Point> > foreground_regions;
+ };
+
+ The class discriminates between foreground and background pixels by building and maintaining a model of the background. Any pixel which does not fit this model is then deemed to be foreground. The class implements algorithm described in [FGD2003]_.
+
+ The results are available through the class fields:
+
+ .. ocv:member:: cv::gpu::GpuMat background
+
+ The output background image.
+
+ .. ocv:member:: cv::gpu::GpuMat foreground
+
+ The output foreground mask as an 8-bit binary image.
+
+ .. ocv:member:: cv::gpu::GpuMat foreground_regions
+
+ The output foreground regions calculated by :ocv:func:`findContours`.
+
+
+
+gpu::FGDStatModel::FGDStatModel
+-------------------------------
+Constructors.
+
+.. ocv:function:: gpu::FGDStatModel::FGDStatModel(int out_cn = 3)
+.. ocv:function:: gpu::FGDStatModel::FGDStatModel(const cv::gpu::GpuMat& firstFrame, const Params& params = Params(), int out_cn = 3)
+
+ :param firstFrame: First frame from video stream. Supports 3- and 4-channels input ( ``CV_8UC3`` and ``CV_8UC4`` ).
+
+ :param params: Algorithm's parameters. See [FGD2003]_ for explanation.
+
+ :param out_cn: Channels count in output result and inner buffers. Can be 3 or 4. 4-channels version requires more memory, but works a bit faster.
+
+.. seealso:: :ocv:func:`gpu::FGDStatModel::create`
+
+
+
+gpu::FGDStatModel::create
+-------------------------
+Initializes background model.
+
+.. ocv:function:: void gpu::FGDStatModel::create(const cv::gpu::GpuMat& firstFrame, const Params& params = Params())
+
+ :param firstFrame: First frame from video stream. Supports 3- and 4-channels input ( ``CV_8UC3`` and ``CV_8UC4`` ).
+
+ :param params: Algorithm's parameters. See [FGD2003]_ for explanation.
+
+
+
+gpu::FGDStatModel::release
+--------------------------
+Releases all inner buffer's memory.
+
+.. ocv:function:: void gpu::FGDStatModel::release()
+
+
+
+gpu::FGDStatModel::update
+--------------------------
+Updates the background model and returns foreground regions count.
+
+.. ocv:function:: int gpu::FGDStatModel::update(const cv::gpu::GpuMat& curFrame)
+
+ :param curFrame: Next video frame.
+
+
+
+gpu::MOG_GPU
+------------
+.. ocv:class:: gpu::MOG_GPU
+
+Gaussian Mixture-based Backbround/Foreground Segmentation Algorithm. ::
+
+ class MOG_GPU
+ {
+ public:
+ MOG_GPU(int nmixtures = -1);
+
+ void initialize(Size frameSize, int frameType);
+
+ void operator()(const GpuMat& frame, GpuMat& fgmask, float learningRate = 0.0f, Stream& stream = Stream::Null());
+
+ void getBackgroundImage(GpuMat& backgroundImage, Stream& stream = Stream::Null()) const;
+
+ void release();
+
+ int history;
+ float varThreshold;
+ float backgroundRatio;
+ float noiseSigma;
+ };
+
+The class discriminates between foreground and background pixels by building and maintaining a model of the background. Any pixel which does not fit this model is then deemed to be foreground. The class implements algorithm described in [MOG2001]_.
+
+.. seealso:: :ocv:class:`BackgroundSubtractorMOG`
+
+
+
+gpu::MOG_GPU::MOG_GPU
+---------------------
+The constructor.
+
+.. ocv:function:: gpu::MOG_GPU::MOG_GPU(int nmixtures = -1)
+
+ :param nmixtures: Number of Gaussian mixtures.
+
+Default constructor sets all parameters to default values.
+
+
+
+gpu::MOG_GPU::operator()
+------------------------
+Updates the background model and returns the foreground mask.
+
+.. ocv:function:: void gpu::MOG_GPU::operator()(const GpuMat& frame, GpuMat& fgmask, float learningRate = 0.0f, Stream& stream = Stream::Null())
+
+ :param frame: Next video frame.
+
+ :param fgmask: The output foreground mask as an 8-bit binary image.
+
+ :param stream: Stream for the asynchronous version.
+
+
+
+gpu::MOG_GPU::getBackgroundImage
+--------------------------------
+Computes a background image.
+
+.. ocv:function:: void gpu::MOG_GPU::getBackgroundImage(GpuMat& backgroundImage, Stream& stream = Stream::Null()) const
+
+ :param backgroundImage: The output background image.
+
+ :param stream: Stream for the asynchronous version.
+
+
+
+gpu::MOG_GPU::release
+---------------------
+Releases all inner buffer's memory.
+
+.. ocv:function:: void gpu::MOG_GPU::release()
+
+
+
+gpu::MOG2_GPU
+-------------
+.. ocv:class:: gpu::MOG2_GPU
+
+Gaussian Mixture-based Background/Foreground Segmentation Algorithm. ::
+
+ class MOG2_GPU
+ {
+ public:
+ MOG2_GPU(int nmixtures = -1);
+
+ void initialize(Size frameSize, int frameType);
+
+ void operator()(const GpuMat& frame, GpuMat& fgmask, float learningRate = 0.0f, Stream& stream = Stream::Null());
+
+ void getBackgroundImage(GpuMat& backgroundImage, Stream& stream = Stream::Null()) const;
+
+ void release();
+
+ // parameters
+ ...
+ };
+
+ The class discriminates between foreground and background pixels by building and maintaining a model of the background. Any pixel which does not fit this model is then deemed to be foreground. The class implements algorithm described in [MOG2004]_.
+
+ Here are important members of the class that control the algorithm, which you can set after constructing the class instance:
+
+ .. ocv:member:: float backgroundRatio
+
+ Threshold defining whether the component is significant enough to be included into the background model ( corresponds to ``TB=1-cf`` from the paper??which paper??). ``cf=0.1 => TB=0.9`` is default. For ``alpha=0.001``, it means that the mode should exist for approximately 105 frames before it is considered foreground.
+
+ .. ocv:member:: float varThreshold
+
+ Threshold for the squared Mahalanobis distance that helps decide when a sample is close to the existing components (corresponds to ``Tg``). If it is not close to any component, a new component is generated. ``3 sigma => Tg=3*3=9`` is default. A smaller ``Tg`` value generates more components. A higher ``Tg`` value may result in a small number of components but they can grow too large.
+
+ .. ocv:member:: float fVarInit
+
+ Initial variance for the newly generated components. It affects the speed of adaptation. The parameter value is based on your estimate of the typical standard deviation from the images. OpenCV uses 15 as a reasonable value.
+
+ .. ocv:member:: float fVarMin
+
+ Parameter used to further control the variance.
+
+ .. ocv:member:: float fVarMax
+
+ Parameter used to further control the variance.
+
+ .. ocv:member:: float fCT
+
+ Complexity reduction parameter. This parameter defines the number of samples needed to accept to prove the component exists. ``CT=0.05`` is a default value for all the samples. By setting ``CT=0`` you get an algorithm very similar to the standard Stauffer&Grimson algorithm.
+
+ .. ocv:member:: uchar nShadowDetection
+
+ The value for marking shadow pixels in the output foreground mask. Default value is 127.
+
+ .. ocv:member:: float fTau
+
+ Shadow threshold. The shadow is detected if the pixel is a darker version of the background. ``Tau`` is a threshold defining how much darker the shadow can be. ``Tau= 0.5`` means that if a pixel is more than twice darker then it is not shadow. See [ShadowDetect2003]_.
+
+ .. ocv:member:: bool bShadowDetection
+
+ Parameter defining whether shadow detection should be enabled.
+
+.. seealso:: :ocv:class:`BackgroundSubtractorMOG2`
+
+
+
+gpu::MOG2_GPU::MOG2_GPU
+-----------------------
+The constructor.
+
+.. ocv:function:: gpu::MOG2_GPU::MOG2_GPU(int nmixtures = -1)
+
+ :param nmixtures: Number of Gaussian mixtures.
+
+Default constructor sets all parameters to default values.
+
+
+
+gpu::MOG2_GPU::operator()
+-------------------------
+Updates the background model and returns the foreground mask.
+
+.. ocv:function:: void gpu::MOG2_GPU::operator()( const GpuMat& frame, GpuMat& fgmask, float learningRate=-1.0f, Stream& stream=Stream::Null() )
+
+ :param frame: Next video frame.
+
+ :param fgmask: The output foreground mask as an 8-bit binary image.
+
+ :param stream: Stream for the asynchronous version.
+
+
+
+gpu::MOG2_GPU::getBackgroundImage
+---------------------------------
+Computes a background image.
+
+.. ocv:function:: void gpu::MOG2_GPU::getBackgroundImage(GpuMat& backgroundImage, Stream& stream = Stream::Null()) const
+
+ :param backgroundImage: The output background image.
+
+ :param stream: Stream for the asynchronous version.
+
+
+
+gpu::MOG2_GPU::release
+----------------------
+Releases all inner buffer's memory.
+
+.. ocv:function:: void gpu::MOG2_GPU::release()
+
+
+
+gpu::GMG_GPU
+------------
+.. ocv:class:: gpu::GMG_GPU
+
+ Class used for background/foreground segmentation. ::
+
+ class GMG_GPU_GPU
+ {
+ public:
+ GMG_GPU();
+
+ void initialize(Size frameSize, float min = 0.0f, float max = 255.0f);
+
+ void operator ()(const GpuMat& frame, GpuMat& fgmask, float learningRate = -1.0f, Stream& stream = Stream::Null());
+
+ void release();
+
+ int maxFeatures;
+ float learningRate;
+ int numInitializationFrames;
+ int quantizationLevels;
+ float backgroundPrior;
+ float decisionThreshold;
+ int smoothingRadius;
+
+ ...
+ };
+
+ The class discriminates between foreground and background pixels by building and maintaining a model of the background. Any pixel which does not fit this model is then deemed to be foreground. The class implements algorithm described in [GMG2012]_.
+
+ Here are important members of the class that control the algorithm, which you can set after constructing the class instance:
+
+ .. ocv:member:: int maxFeatures
+
+ Total number of distinct colors to maintain in histogram.
+
+ .. ocv:member:: float learningRate
+
+ Set between 0.0 and 1.0, determines how quickly features are "forgotten" from histograms.
+
+ .. ocv:member:: int numInitializationFrames
+
+ Number of frames of video to use to initialize histograms.
+
+ .. ocv:member:: int quantizationLevels
+
+ Number of discrete levels in each channel to be used in histograms.
+
+ .. ocv:member:: float backgroundPrior
+
+ Prior probability that any given pixel is a background pixel. A sensitivity parameter.
+
+ .. ocv:member:: float decisionThreshold
+
+ Value above which pixel is determined to be FG.
+
+ .. ocv:member:: float smoothingRadius
+
+ Smoothing radius, in pixels, for cleaning up FG image.
+
+
+
+gpu::GMG_GPU::GMG_GPU
+---------------------
+The default constructor.
+
+.. ocv:function:: gpu::GMG_GPU::GMG_GPU()
+
+Default constructor sets all parameters to default values.
+
+
+
+gpu::GMG_GPU::initialize
+------------------------
+Initialize background model and allocates all inner buffers.
+
+.. ocv:function:: void gpu::GMG_GPU::initialize(Size frameSize, float min = 0.0f, float max = 255.0f)
+
+ :param frameSize: Input frame size.
+
+ :param min: Minimum value taken on by pixels in image sequence. Usually 0.
+
+ :param max: Maximum value taken on by pixels in image sequence, e.g. 1.0 or 255.
+
+
+
+gpu::GMG_GPU::operator()
+------------------------
+Updates the background model and returns the foreground mask
+
+.. ocv:function:: void gpu::GMG_GPU::operator ()( const GpuMat& frame, GpuMat& fgmask, float learningRate=-1.0f, Stream& stream=Stream::Null() )
+
+ :param frame: Next video frame.
+
+ :param fgmask: The output foreground mask as an 8-bit binary image.
+
+ :param stream: Stream for the asynchronous version.
+
+
+
+gpu::GMG_GPU::release
+---------------------
+Releases all inner buffer's memory.
+
+.. ocv:function:: void gpu::GMG_GPU::release()
+
+
+
+.. [FGD2003] Liyuan Li, Weimin Huang, Irene Y.H. Gu, and Qi Tian. *Foreground Object Detection from Videos Containing Complex Background*. ACM MM2003 9p, 2003.
+.. [MOG2001] P. KadewTraKuPong and R. Bowden. *An improved adaptive background mixture model for real-time tracking with shadow detection*. Proc. 2nd European Workshop on Advanced Video-Based Surveillance Systems, 2001
+.. [MOG2004] Z. Zivkovic. *Improved adaptive Gausian mixture model for background subtraction*. International Conference Pattern Recognition, UK, August, 2004
+.. [ShadowDetect2003] Prati, Mikic, Trivedi and Cucchiarra. *Detecting Moving Shadows...*. IEEE PAMI, 2003
+.. [GMG2012] A. Godbehere, A. Matsukawa and K. Goldberg. *Visual Tracking of Human Visitors under Variable-Lighting Conditions for a Responsive Audio Art Installation*. American Control Conference, Montreal, June 2012
--- /dev/null
+**************************************************
+gpubgsegm. GPU-accelerated Background Segmentation
+**************************************************
+
+.. toctree::
+ :maxdepth: 1
+
+ background_segmentation
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_GPUBGSEGM_HPP__
+#define __OPENCV_GPUBGSEGM_HPP__
+
+#ifndef __cplusplus
+# error gpubgsegm.hpp header must be compiled as C++
+#endif
+
+#include <memory>
+
+#include "opencv2/core/gpumat.hpp"
+#include "opencv2/gpufilters.hpp"
+
+namespace cv { namespace gpu {
+
+// Foreground Object Detection from Videos Containing Complex Background.
+// Liyuan Li, Weimin Huang, Irene Y.H. Gu, and Qi Tian.
+// ACM MM2003 9p
+class CV_EXPORTS FGDStatModel
+{
+public:
+ struct CV_EXPORTS Params
+ {
+ int Lc; // Quantized levels per 'color' component. Power of two, typically 32, 64 or 128.
+ int N1c; // Number of color vectors used to model normal background color variation at a given pixel.
+ int N2c; // Number of color vectors retained at given pixel. Must be > N1c, typically ~ 5/3 of N1c.
+ // Used to allow the first N1c vectors to adapt over time to changing background.
+
+ int Lcc; // Quantized levels per 'color co-occurrence' component. Power of two, typically 16, 32 or 64.
+ int N1cc; // Number of color co-occurrence vectors used to model normal background color variation at a given pixel.
+ int N2cc; // Number of color co-occurrence vectors retained at given pixel. Must be > N1cc, typically ~ 5/3 of N1cc.
+ // Used to allow the first N1cc vectors to adapt over time to changing background.
+
+ bool is_obj_without_holes; // If TRUE we ignore holes within foreground blobs. Defaults to TRUE.
+ int perform_morphing; // Number of erode-dilate-erode foreground-blob cleanup iterations.
+ // These erase one-pixel junk blobs and merge almost-touching blobs. Default value is 1.
+
+ float alpha1; // How quickly we forget old background pixel values seen. Typically set to 0.1.
+ float alpha2; // "Controls speed of feature learning". Depends on T. Typical value circa 0.005.
+ float alpha3; // Alternate to alpha2, used (e.g.) for quicker initial convergence. Typical value 0.1.
+
+ float delta; // Affects color and color co-occurrence quantization, typically set to 2.
+ float T; // A percentage value which determines when new features can be recognized as new background. (Typically 0.9).
+ float minArea; // Discard foreground blobs whose bounding box is smaller than this threshold.
+
+ // default Params
+ Params();
+ };
+
+ // out_cn - channels count in output result (can be 3 or 4)
+ // 4-channels require more memory, but a bit faster
+ explicit FGDStatModel(int out_cn = 3);
+ explicit FGDStatModel(const cv::gpu::GpuMat& firstFrame, const Params& params = Params(), int out_cn = 3);
+
+ ~FGDStatModel();
+
+ void create(const cv::gpu::GpuMat& firstFrame, const Params& params = Params());
+ void release();
+
+ int update(const cv::gpu::GpuMat& curFrame);
+
+ //8UC3 or 8UC4 reference background image
+ cv::gpu::GpuMat background;
+
+ //8UC1 foreground image
+ cv::gpu::GpuMat foreground;
+
+ std::vector< std::vector<cv::Point> > foreground_regions;
+
+private:
+ FGDStatModel(const FGDStatModel&);
+ FGDStatModel& operator=(const FGDStatModel&);
+
+ class Impl;
+ std::auto_ptr<Impl> impl_;
+};
+
+/*!
+ Gaussian Mixture-based Backbround/Foreground Segmentation Algorithm
+
+ The class implements the following algorithm:
+ "An improved adaptive background mixture model for real-time tracking with shadow detection"
+ P. KadewTraKuPong and R. Bowden,
+ Proc. 2nd European Workshp on Advanced Video-Based Surveillance Systems, 2001."
+ http://personal.ee.surrey.ac.uk/Personal/R.Bowden/publications/avbs01/avbs01.pdf
+*/
+class CV_EXPORTS MOG_GPU
+{
+public:
+ //! the default constructor
+ MOG_GPU(int nmixtures = -1);
+
+ //! re-initiaization method
+ void initialize(Size frameSize, int frameType);
+
+ //! the update operator
+ void operator()(const GpuMat& frame, GpuMat& fgmask, float learningRate = 0.0f, Stream& stream = Stream::Null());
+
+ //! computes a background image which are the mean of all background gaussians
+ void getBackgroundImage(GpuMat& backgroundImage, Stream& stream = Stream::Null()) const;
+
+ //! releases all inner buffers
+ void release();
+
+ int history;
+ float varThreshold;
+ float backgroundRatio;
+ float noiseSigma;
+
+private:
+ int nmixtures_;
+
+ Size frameSize_;
+ int frameType_;
+ int nframes_;
+
+ GpuMat weight_;
+ GpuMat sortKey_;
+ GpuMat mean_;
+ GpuMat var_;
+};
+
+/*!
+ The class implements the following algorithm:
+ "Improved adaptive Gausian mixture model for background subtraction"
+ Z.Zivkovic
+ International Conference Pattern Recognition, UK, August, 2004.
+ http://www.zoranz.net/Publications/zivkovic2004ICPR.pdf
+*/
+class CV_EXPORTS MOG2_GPU
+{
+public:
+ //! the default constructor
+ MOG2_GPU(int nmixtures = -1);
+
+ //! re-initiaization method
+ void initialize(Size frameSize, int frameType);
+
+ //! the update operator
+ void operator()(const GpuMat& frame, GpuMat& fgmask, float learningRate = -1.0f, Stream& stream = Stream::Null());
+
+ //! computes a background image which are the mean of all background gaussians
+ void getBackgroundImage(GpuMat& backgroundImage, Stream& stream = Stream::Null()) const;
+
+ //! releases all inner buffers
+ void release();
+
+ // parameters
+ // you should call initialize after parameters changes
+
+ int history;
+
+ //! here it is the maximum allowed number of mixture components.
+ //! Actual number is determined dynamically per pixel
+ float varThreshold;
+ // threshold on the squared Mahalanobis distance to decide if it is well described
+ // by the background model or not. Related to Cthr from the paper.
+ // This does not influence the update of the background. A typical value could be 4 sigma
+ // and that is varThreshold=4*4=16; Corresponds to Tb in the paper.
+
+ /////////////////////////
+ // less important parameters - things you might change but be carefull
+ ////////////////////////
+
+ float backgroundRatio;
+ // corresponds to fTB=1-cf from the paper
+ // TB - threshold when the component becomes significant enough to be included into
+ // the background model. It is the TB=1-cf from the paper. So I use cf=0.1 => TB=0.
+ // For alpha=0.001 it means that the mode should exist for approximately 105 frames before
+ // it is considered foreground
+ // float noiseSigma;
+ float varThresholdGen;
+
+ //correspondts to Tg - threshold on the squared Mahalan. dist. to decide
+ //when a sample is close to the existing components. If it is not close
+ //to any a new component will be generated. I use 3 sigma => Tg=3*3=9.
+ //Smaller Tg leads to more generated components and higher Tg might make
+ //lead to small number of components but they can grow too large
+ float fVarInit;
+ float fVarMin;
+ float fVarMax;
+
+ //initial variance for the newly generated components.
+ //It will will influence the speed of adaptation. A good guess should be made.
+ //A simple way is to estimate the typical standard deviation from the images.
+ //I used here 10 as a reasonable value
+ // min and max can be used to further control the variance
+ float fCT; //CT - complexity reduction prior
+ //this is related to the number of samples needed to accept that a component
+ //actually exists. We use CT=0.05 of all the samples. By setting CT=0 you get
+ //the standard Stauffer&Grimson algorithm (maybe not exact but very similar)
+
+ //shadow detection parameters
+ bool bShadowDetection; //default 1 - do shadow detection
+ unsigned char nShadowDetection; //do shadow detection - insert this value as the detection result - 127 default value
+ float fTau;
+ // Tau - shadow threshold. The shadow is detected if the pixel is darker
+ //version of the background. Tau is a threshold on how much darker the shadow can be.
+ //Tau= 0.5 means that if pixel is more than 2 times darker then it is not shadow
+ //See: Prati,Mikic,Trivedi,Cucchiarra,"Detecting Moving Shadows...",IEEE PAMI,2003.
+
+private:
+ int nmixtures_;
+
+ Size frameSize_;
+ int frameType_;
+ int nframes_;
+
+ GpuMat weight_;
+ GpuMat variance_;
+ GpuMat mean_;
+
+ GpuMat bgmodelUsedModes_; //keep track of number of modes per pixel
+};
+
+/**
+ * Background Subtractor module. Takes a series of images and returns a sequence of mask (8UC1)
+ * images of the same size, where 255 indicates Foreground and 0 represents Background.
+ * This class implements an algorithm described in "Visual Tracking of Human Visitors under
+ * Variable-Lighting Conditions for a Responsive Audio Art Installation," A. Godbehere,
+ * A. Matsukawa, K. Goldberg, American Control Conference, Montreal, June 2012.
+ */
+class CV_EXPORTS GMG_GPU
+{
+public:
+ GMG_GPU();
+
+ /**
+ * Validate parameters and set up data structures for appropriate frame size.
+ * @param frameSize Input frame size
+ * @param min Minimum value taken on by pixels in image sequence. Usually 0
+ * @param max Maximum value taken on by pixels in image sequence. e.g. 1.0 or 255
+ */
+ void initialize(Size frameSize, float min = 0.0f, float max = 255.0f);
+
+ /**
+ * Performs single-frame background subtraction and builds up a statistical background image
+ * model.
+ * @param frame Input frame
+ * @param fgmask Output mask image representing foreground and background pixels
+ * @param stream Stream for the asynchronous version
+ */
+ void operator ()(const GpuMat& frame, GpuMat& fgmask, float learningRate = -1.0f, Stream& stream = Stream::Null());
+
+ //! Releases all inner buffers
+ void release();
+
+ //! Total number of distinct colors to maintain in histogram.
+ int maxFeatures;
+
+ //! Set between 0.0 and 1.0, determines how quickly features are "forgotten" from histograms.
+ float learningRate;
+
+ //! Number of frames of video to use to initialize histograms.
+ int numInitializationFrames;
+
+ //! Number of discrete levels in each channel to be used in histograms.
+ int quantizationLevels;
+
+ //! Prior probability that any given pixel is a background pixel. A sensitivity parameter.
+ float backgroundPrior;
+
+ //! Value above which pixel is determined to be FG.
+ float decisionThreshold;
+
+ //! Smoothing radius, in pixels, for cleaning up FG image.
+ int smoothingRadius;
+
+ //! Perform background model update.
+ bool updateBackgroundModel;
+
+private:
+ float maxVal_, minVal_;
+
+ Size frameSize_;
+
+ int frameNum_;
+
+ GpuMat nfeatures_;
+ GpuMat colors_;
+ GpuMat weights_;
+
+ Ptr<FilterEngine_GPU> boxFilter_;
+ GpuMat buf_;
+};
+
+}} // namespace cv { namespace gpu {
+
+#endif /* __OPENCV_GPUBGSEGM_HPP__ */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+#include "opencv2/legacy.hpp"
+
+using namespace std;
+using namespace testing;
+using namespace perf;
+
+#if defined(HAVE_XINE) || \
+ defined(HAVE_GSTREAMER) || \
+ defined(HAVE_QUICKTIME) || \
+ defined(HAVE_AVFOUNDATION) || \
+ defined(HAVE_FFMPEG) || \
+ defined(WIN32) /* assume that we have ffmpeg */
+
+# define BUILD_WITH_VIDEO_INPUT_SUPPORT 1
+#else
+# define BUILD_WITH_VIDEO_INPUT_SUPPORT 0
+#endif
+
+namespace cv
+{
+ template<> void Ptr<CvBGStatModel>::delete_obj()
+ {
+ cvReleaseBGStatModel(&obj);
+ }
+}
+
+//////////////////////////////////////////////////////
+// FGDStatModel
+
+#if BUILD_WITH_VIDEO_INPUT_SUPPORT
+
+DEF_PARAM_TEST_1(Video, string);
+
+PERF_TEST_P(Video, FGDStatModel,
+ Values(string("gpu/video/768x576.avi")))
+{
+ declare.time(60);
+
+ const string inputFile = perf::TestBase::getDataPath(GetParam());
+
+ cv::VideoCapture cap(inputFile);
+ ASSERT_TRUE(cap.isOpened());
+
+ cv::Mat frame;
+ cap >> frame;
+ ASSERT_FALSE(frame.empty());
+
+ if (PERF_RUN_GPU())
+ {
+ cv::gpu::GpuMat d_frame(frame);
+
+ cv::gpu::FGDStatModel d_model(4);
+ d_model.create(d_frame);
+
+ for (int i = 0; i < 10; ++i)
+ {
+ cap >> frame;
+ ASSERT_FALSE(frame.empty());
+
+ d_frame.upload(frame);
+
+ startTimer(); next();
+ d_model.update(d_frame);
+ stopTimer();
+ }
+
+ const cv::gpu::GpuMat background = d_model.background;
+ const cv::gpu::GpuMat foreground = d_model.foreground;
+
+ GPU_SANITY_CHECK(background, 1e-2, ERROR_RELATIVE);
+ GPU_SANITY_CHECK(foreground, 1e-2, ERROR_RELATIVE);
+ }
+ else
+ {
+ IplImage ipl_frame = frame;
+ cv::Ptr<CvBGStatModel> model(cvCreateFGDStatModel(&ipl_frame));
+
+ for (int i = 0; i < 10; ++i)
+ {
+ cap >> frame;
+ ASSERT_FALSE(frame.empty());
+
+ ipl_frame = frame;
+
+ startTimer(); next();
+ cvUpdateBGStatModel(&ipl_frame, model);
+ stopTimer();
+ }
+
+ const cv::Mat background = cv::cvarrToMat(model->background);
+ const cv::Mat foreground = cv::cvarrToMat(model->foreground);
+
+ CPU_SANITY_CHECK(background);
+ CPU_SANITY_CHECK(foreground);
+ }
+}
+
+#endif
+
+//////////////////////////////////////////////////////
+// MOG
+
+#if BUILD_WITH_VIDEO_INPUT_SUPPORT
+
+DEF_PARAM_TEST(Video_Cn_LearningRate, string, MatCn, double);
+
+PERF_TEST_P(Video_Cn_LearningRate, MOG,
+ Combine(Values("gpu/video/768x576.avi", "gpu/video/1920x1080.avi"),
+ GPU_CHANNELS_1_3_4,
+ Values(0.0, 0.01)))
+{
+ const string inputFile = perf::TestBase::getDataPath(GET_PARAM(0));
+ const int cn = GET_PARAM(1);
+ const float learningRate = static_cast<float>(GET_PARAM(2));
+
+ cv::VideoCapture cap(inputFile);
+ ASSERT_TRUE(cap.isOpened());
+
+ cv::Mat frame;
+
+ cap >> frame;
+ ASSERT_FALSE(frame.empty());
+
+ if (cn != 3)
+ {
+ cv::Mat temp;
+ if (cn == 1)
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
+ else
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
+ cv::swap(temp, frame);
+ }
+
+ if (PERF_RUN_GPU())
+ {
+ cv::gpu::GpuMat d_frame(frame);
+ cv::gpu::MOG_GPU d_mog;
+ cv::gpu::GpuMat foreground;
+
+ d_mog(d_frame, foreground, learningRate);
+
+ for (int i = 0; i < 10; ++i)
+ {
+ cap >> frame;
+ ASSERT_FALSE(frame.empty());
+
+ if (cn != 3)
+ {
+ cv::Mat temp;
+ if (cn == 1)
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
+ else
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
+ cv::swap(temp, frame);
+ }
+
+ d_frame.upload(frame);
+
+ startTimer(); next();
+ d_mog(d_frame, foreground, learningRate);
+ stopTimer();
+ }
+
+ GPU_SANITY_CHECK(foreground);
+ }
+ else
+ {
+ cv::Ptr<cv::BackgroundSubtractor> mog = cv::createBackgroundSubtractorMOG();
+ cv::Mat foreground;
+
+ mog->apply(frame, foreground, learningRate);
+
+ for (int i = 0; i < 10; ++i)
+ {
+ cap >> frame;
+ ASSERT_FALSE(frame.empty());
+
+ if (cn != 3)
+ {
+ cv::Mat temp;
+ if (cn == 1)
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
+ else
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
+ cv::swap(temp, frame);
+ }
+
+ startTimer(); next();
+ mog->apply(frame, foreground, learningRate);
+ stopTimer();
+ }
+
+ CPU_SANITY_CHECK(foreground);
+ }
+}
+
+#endif
+
+//////////////////////////////////////////////////////
+// MOG2
+
+#if BUILD_WITH_VIDEO_INPUT_SUPPORT
+
+DEF_PARAM_TEST(Video_Cn, string, int);
+
+PERF_TEST_P(Video_Cn, MOG2,
+ Combine(Values("gpu/video/768x576.avi", "gpu/video/1920x1080.avi"),
+ GPU_CHANNELS_1_3_4))
+{
+ const string inputFile = perf::TestBase::getDataPath(GET_PARAM(0));
+ const int cn = GET_PARAM(1);
+
+ cv::VideoCapture cap(inputFile);
+ ASSERT_TRUE(cap.isOpened());
+
+ cv::Mat frame;
+
+ cap >> frame;
+ ASSERT_FALSE(frame.empty());
+
+ if (cn != 3)
+ {
+ cv::Mat temp;
+ if (cn == 1)
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
+ else
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
+ cv::swap(temp, frame);
+ }
+
+ if (PERF_RUN_GPU())
+ {
+ cv::gpu::MOG2_GPU d_mog2;
+ d_mog2.bShadowDetection = false;
+
+ cv::gpu::GpuMat d_frame(frame);
+ cv::gpu::GpuMat foreground;
+
+ d_mog2(d_frame, foreground);
+
+ for (int i = 0; i < 10; ++i)
+ {
+ cap >> frame;
+ ASSERT_FALSE(frame.empty());
+
+ if (cn != 3)
+ {
+ cv::Mat temp;
+ if (cn == 1)
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
+ else
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
+ cv::swap(temp, frame);
+ }
+
+ d_frame.upload(frame);
+
+ startTimer(); next();
+ d_mog2(d_frame, foreground);
+ stopTimer();
+ }
+
+ GPU_SANITY_CHECK(foreground);
+ }
+ else
+ {
+ cv::Ptr<cv::BackgroundSubtractor> mog2 = cv::createBackgroundSubtractorMOG2();
+ mog2->set("detectShadows", false);
+
+ cv::Mat foreground;
+
+ mog2->apply(frame, foreground);
+
+ for (int i = 0; i < 10; ++i)
+ {
+ cap >> frame;
+ ASSERT_FALSE(frame.empty());
+
+ if (cn != 3)
+ {
+ cv::Mat temp;
+ if (cn == 1)
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
+ else
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
+ cv::swap(temp, frame);
+ }
+
+ startTimer(); next();
+ mog2->apply(frame, foreground);
+ stopTimer();
+ }
+
+ CPU_SANITY_CHECK(foreground);
+ }
+}
+
+#endif
+
+//////////////////////////////////////////////////////
+// MOG2GetBackgroundImage
+
+#if BUILD_WITH_VIDEO_INPUT_SUPPORT
+
+PERF_TEST_P(Video_Cn, MOG2GetBackgroundImage,
+ Combine(Values("gpu/video/768x576.avi", "gpu/video/1920x1080.avi"),
+ GPU_CHANNELS_1_3_4))
+{
+ const string inputFile = perf::TestBase::getDataPath(GET_PARAM(0));
+ const int cn = GET_PARAM(1);
+
+ cv::VideoCapture cap(inputFile);
+ ASSERT_TRUE(cap.isOpened());
+
+ cv::Mat frame;
+
+ if (PERF_RUN_GPU())
+ {
+ cv::gpu::GpuMat d_frame;
+ cv::gpu::MOG2_GPU d_mog2;
+ cv::gpu::GpuMat d_foreground;
+
+ for (int i = 0; i < 10; ++i)
+ {
+ cap >> frame;
+ ASSERT_FALSE(frame.empty());
+
+ if (cn != 3)
+ {
+ cv::Mat temp;
+ if (cn == 1)
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
+ else
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
+ cv::swap(temp, frame);
+ }
+
+ d_frame.upload(frame);
+
+ d_mog2(d_frame, d_foreground);
+ }
+
+ cv::gpu::GpuMat background;
+
+ TEST_CYCLE() d_mog2.getBackgroundImage(background);
+
+ GPU_SANITY_CHECK(background, 1);
+ }
+ else
+ {
+ cv::Ptr<cv::BackgroundSubtractor> mog2 = cv::createBackgroundSubtractorMOG2();
+ cv::Mat foreground;
+
+ for (int i = 0; i < 10; ++i)
+ {
+ cap >> frame;
+ ASSERT_FALSE(frame.empty());
+
+ if (cn != 3)
+ {
+ cv::Mat temp;
+ if (cn == 1)
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
+ else
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
+ cv::swap(temp, frame);
+ }
+
+ mog2->apply(frame, foreground);
+ }
+
+ cv::Mat background;
+
+ TEST_CYCLE() mog2->getBackgroundImage(background);
+
+ CPU_SANITY_CHECK(background);
+ }
+}
+
+#endif
+
+//////////////////////////////////////////////////////
+// GMG
+
+#if BUILD_WITH_VIDEO_INPUT_SUPPORT
+
+DEF_PARAM_TEST(Video_Cn_MaxFeatures, string, MatCn, int);
+
+PERF_TEST_P(Video_Cn_MaxFeatures, GMG,
+ Combine(Values(string("gpu/video/768x576.avi")),
+ GPU_CHANNELS_1_3_4,
+ Values(20, 40, 60)))
+{
+ const std::string inputFile = perf::TestBase::getDataPath(GET_PARAM(0));
+ const int cn = GET_PARAM(1);
+ const int maxFeatures = GET_PARAM(2);
+
+ cv::VideoCapture cap(inputFile);
+ ASSERT_TRUE(cap.isOpened());
+
+ cv::Mat frame;
+ cap >> frame;
+ ASSERT_FALSE(frame.empty());
+
+ if (cn != 3)
+ {
+ cv::Mat temp;
+ if (cn == 1)
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
+ else
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
+ cv::swap(temp, frame);
+ }
+
+ if (PERF_RUN_GPU())
+ {
+ cv::gpu::GpuMat d_frame(frame);
+ cv::gpu::GpuMat foreground;
+
+ cv::gpu::GMG_GPU d_gmg;
+ d_gmg.maxFeatures = maxFeatures;
+
+ d_gmg(d_frame, foreground);
+
+ for (int i = 0; i < 150; ++i)
+ {
+ cap >> frame;
+ if (frame.empty())
+ {
+ cap.release();
+ cap.open(inputFile);
+ cap >> frame;
+ }
+
+ if (cn != 3)
+ {
+ cv::Mat temp;
+ if (cn == 1)
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
+ else
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
+ cv::swap(temp, frame);
+ }
+
+ d_frame.upload(frame);
+
+ startTimer(); next();
+ d_gmg(d_frame, foreground);
+ stopTimer();
+ }
+
+ GPU_SANITY_CHECK(foreground);
+ }
+ else
+ {
+ cv::Mat foreground;
+ cv::Mat zeros(frame.size(), CV_8UC1, cv::Scalar::all(0));
+
+ cv::Ptr<cv::BackgroundSubtractor> gmg = cv::createBackgroundSubtractorGMG();
+ gmg->set("maxFeatures", maxFeatures);
+ //gmg.initialize(frame.size(), 0.0, 255.0);
+
+ gmg->apply(frame, foreground);
+
+ for (int i = 0; i < 150; ++i)
+ {
+ cap >> frame;
+ if (frame.empty())
+ {
+ cap.release();
+ cap.open(inputFile);
+ cap >> frame;
+ }
+
+ if (cn != 3)
+ {
+ cv::Mat temp;
+ if (cn == 1)
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2GRAY);
+ else
+ cv::cvtColor(frame, temp, cv::COLOR_BGR2BGRA);
+ cv::swap(temp, frame);
+ }
+
+ startTimer(); next();
+ gmg->apply(frame, foreground);
+ stopTimer();
+ }
+
+ CPU_SANITY_CHECK(foreground);
+ }
+}
+
+#endif
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace perf;
+
+CV_PERF_TEST_MAIN(gpubgsegm, printCudaInfo())
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifdef __GNUC__
+# pragma GCC diagnostic ignored "-Wmissing-declarations"
+# if defined __clang__ || defined __APPLE__
+# pragma GCC diagnostic ignored "-Wmissing-prototypes"
+# pragma GCC diagnostic ignored "-Wextra"
+# endif
+#endif
+
+#ifndef __OPENCV_PERF_PRECOMP_HPP__
+#define __OPENCV_PERF_PRECOMP_HPP__
+
+#include "opencv2/ts.hpp"
+#include "opencv2/ts/gpu_perf.hpp"
+
+#include "opencv2/gpubgsegm.hpp"
+#include "opencv2/video.hpp"
+
+#ifdef GTEST_CREATE_SHARED_LIBRARY
+#error no modules except ts should have GTEST_CREATE_SHARED_LIBRARY defined
+#endif
+
+#endif
#include "opencv2/core/cuda/utility.hpp"
#include "opencv2/core/cuda/reduce.hpp"
#include "opencv2/core/cuda/functional.hpp"
-#include "fgd_bgfg_common.hpp"
+#include "fgd.hpp"
using namespace cv::gpu;
using namespace cv::gpu::cudev;
#else
-#include "fgd_bgfg_common.hpp"
+#include "cuda/fgd.hpp"
#include "opencv2/imgproc/imgproc_c.h"
namespace
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_PRECOMP_H__
+#define __OPENCV_PRECOMP_H__
+
+#include <limits>
+
+#include "opencv2/gpubgsegm.hpp"
+#include "opencv2/gpuarithm.hpp"
+#include "opencv2/gpufilters.hpp"
+#include "opencv2/gpuimgproc.hpp"
+
+#include "opencv2/core/gpu_private.hpp"
+
+#endif /* __OPENCV_PRECOMP_H__ */
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Video, FGDStatModel, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_BgSegm, FGDStatModel, testing::Combine(
ALL_DEVICES,
testing::Values(std::string("768x576.avi")),
testing::Values(Channels(3), Channels(4))));
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Video, MOG, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_BgSegm, MOG, testing::Combine(
ALL_DEVICES,
testing::Values(std::string("768x576.avi")),
testing::Values(UseGray(true), UseGray(false)),
ASSERT_MAT_NEAR(background_gold, background, 0);
}
-INSTANTIATE_TEST_CASE_P(GPU_Video, MOG2, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_BgSegm, MOG2, testing::Combine(
ALL_DEVICES,
testing::Values(std::string("768x576.avi")),
testing::Values(UseGray(true), UseGray(false)),
ASSERT_MAT_NEAR(fullfg, d_fgmask, 0);
}
-INSTANTIATE_TEST_CASE_P(GPU_Video, GMG, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_BgSegm, GMG, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatType(CV_8U), MatType(CV_16U), MatType(CV_32F)),
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+CV_GPU_TEST_MAIN("gpu")
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifdef __GNUC__
+# pragma GCC diagnostic ignored "-Wmissing-declarations"
+# if defined __clang__ || defined __APPLE__
+# pragma GCC diagnostic ignored "-Wmissing-prototypes"
+# pragma GCC diagnostic ignored "-Wextra"
+# endif
+#endif
+
+#ifndef __OPENCV_TEST_PRECOMP_HPP__
+#define __OPENCV_TEST_PRECOMP_HPP__
+
+#include <fstream>
+
+#include "opencv2/ts.hpp"
+#include "opencv2/ts/gpu_test.hpp"
+
+#include "opencv2/gpubgsegm.hpp"
+#include "opencv2/video.hpp"
+
+#endif
--- /dev/null
+if(ANDROID OR IOS OR APPLE)
+ ocv_module_disable(gpucodec)
+endif()
+
+set(the_description "GPU-accelerated Video Encoding/Decoding")
+
+ocv_warnings_disable(CMAKE_CXX_FLAGS /wd4127 /wd4324 /wd4512 -Wundef -Wmissing-declarations)
+
+ocv_add_module(gpucodec opencv_highgui)
+
+ocv_module_include_directories()
+ocv_glob_module_sources()
+
+set(extra_libs ${HIGHGUI_LIBRARIES})
+
+if(HAVE_NVCUVID)
+ list(APPEND extra_libs ${CUDA_CUDA_LIBRARY} ${CUDA_nvcuvid_LIBRARY})
+
+ if(WIN32)
+ list(APPEND extra_libs ${CUDA_nvcuvenc_LIBRARY})
+ endif()
+endif()
+
+ocv_create_module(${extra_libs})
+
+ocv_add_precompiled_headers(${the_module})
+
+ocv_add_accuracy_tests()
+ocv_add_perf_tests()
--- /dev/null
+*************************************************
+gpucodec. GPU-accelerated Video Encoding/Decoding
+*************************************************
+
+.. toctree::
+ :maxdepth: 1
+
+ videodec
+ videoenc
--- /dev/null
+Video Decoding
+==============
+
+.. highlight:: cpp
+
+
+
+gpu::VideoReader_GPU
+--------------------
+Video reader class.
+
+.. ocv:class:: gpu::VideoReader_GPU
+
+
+
+gpu::VideoReader_GPU::Codec
+---------------------------
+
+Video codecs supported by :ocv:class:`gpu::VideoReader_GPU` .
+
+.. ocv:enum:: gpu::VideoReader_GPU::Codec
+
+ .. ocv:emember:: MPEG1 = 0
+ .. ocv:emember:: MPEG2
+ .. ocv:emember:: MPEG4
+ .. ocv:emember:: VC1
+ .. ocv:emember:: H264
+ .. ocv:emember:: JPEG
+ .. ocv:emember:: H264_SVC
+ .. ocv:emember:: H264_MVC
+
+ .. ocv:emember:: Uncompressed_YUV420 = (('I'<<24)|('Y'<<16)|('U'<<8)|('V'))
+
+ Y,U,V (4:2:0)
+
+ .. ocv:emember:: Uncompressed_YV12 = (('Y'<<24)|('V'<<16)|('1'<<8)|('2'))
+
+ Y,V,U (4:2:0)
+
+ .. ocv:emember:: Uncompressed_NV12 = (('N'<<24)|('V'<<16)|('1'<<8)|('2'))
+
+ Y,UV (4:2:0)
+
+ .. ocv:emember:: Uncompressed_YUYV = (('Y'<<24)|('U'<<16)|('Y'<<8)|('V'))
+
+ YUYV/YUY2 (4:2:2)
+
+ .. ocv:emember:: Uncompressed_UYVY = (('U'<<24)|('Y'<<16)|('V'<<8)|('Y'))
+
+ UYVY (4:2:2)
+
+
+gpu::VideoReader_GPU::ChromaFormat
+----------------------------------
+
+Chroma formats supported by :ocv:class:`gpu::VideoReader_GPU` .
+
+.. ocv:enum:: gpu::VideoReader_GPU::ChromaFormat
+
+ .. ocv:emember:: Monochrome = 0
+ .. ocv:emember:: YUV420
+ .. ocv:emember:: YUV422
+ .. ocv:emember:: YUV444
+
+
+gpu::VideoReader_GPU::FormatInfo
+--------------------------------
+.. ocv:struct:: gpu::VideoReader_GPU::FormatInfo
+
+Struct providing information about video file format. ::
+
+ struct FormatInfo
+ {
+ Codec codec;
+ ChromaFormat chromaFormat;
+ int width;
+ int height;
+ };
+
+
+gpu::VideoReader_GPU::VideoReader_GPU
+-------------------------------------
+Constructors.
+
+.. ocv:function:: gpu::VideoReader_GPU::VideoReader_GPU()
+.. ocv:function:: gpu::VideoReader_GPU::VideoReader_GPU(const String& filename)
+.. ocv:function:: gpu::VideoReader_GPU::VideoReader_GPU(const cv::Ptr<VideoSource>& source)
+
+ :param filename: Name of the input video file.
+
+ :param source: Video file parser implemented by user.
+
+The constructors initialize video reader. FFMPEG is used to read videos. User can implement own demultiplexing with :ocv:class:`gpu::VideoReader_GPU::VideoSource` .
+
+
+
+gpu::VideoReader_GPU::open
+--------------------------
+Initializes or reinitializes video reader.
+
+.. ocv:function:: void gpu::VideoReader_GPU::open(const String& filename)
+.. ocv:function:: void gpu::VideoReader_GPU::open(const cv::Ptr<VideoSource>& source)
+
+The method opens video reader. Parameters are the same as in the constructor :ocv:func:`gpu::VideoReader_GPU::VideoReader_GPU` . The method throws :ocv:class:`Exception` if error occurs.
+
+
+
+gpu::VideoReader_GPU::isOpened
+------------------------------
+Returns true if video reader has been successfully initialized.
+
+.. ocv:function:: bool gpu::VideoReader_GPU::isOpened() const
+
+
+
+gpu::VideoReader_GPU::close
+---------------------------
+Releases the video reader.
+
+.. ocv:function:: void gpu::VideoReader_GPU::close()
+
+
+
+gpu::VideoReader_GPU::read
+--------------------------
+Grabs, decodes and returns the next video frame.
+
+.. ocv:function:: bool gpu::VideoReader_GPU::read(GpuMat& image)
+
+If no frames has been grabbed (there are no more frames in video file), the methods return ``false`` . The method throws :ocv:class:`Exception` if error occurs.
+
+
+
+gpu::VideoReader_GPU::format
+----------------------------
+Returns information about video file format.
+
+.. ocv:function:: FormatInfo gpu::VideoReader_GPU::format() const
+
+The method throws :ocv:class:`Exception` if video reader wasn't initialized.
+
+
+
+gpu::VideoReader_GPU::dumpFormat
+--------------------------------
+Dump information about video file format to specified stream.
+
+.. ocv:function:: void gpu::VideoReader_GPU::dumpFormat(std::ostream& st)
+
+ :param st: Output stream.
+
+The method throws :ocv:class:`Exception` if video reader wasn't initialized.
+
+
+
+gpu::VideoReader_GPU::VideoSource
+-----------------------------------
+.. ocv:class:: gpu::VideoReader_GPU::VideoSource
+
+Interface for video demultiplexing. ::
+
+ class VideoSource
+ {
+ public:
+ VideoSource();
+ virtual ~VideoSource() {}
+
+ virtual FormatInfo format() const = 0;
+ virtual void start() = 0;
+ virtual void stop() = 0;
+ virtual bool isStarted() const = 0;
+ virtual bool hasError() const = 0;
+
+ protected:
+ bool parseVideoData(const unsigned char* data, size_t size, bool endOfStream = false);
+ };
+
+User can implement own demultiplexing by implementing this interface.
+
+
+
+gpu::VideoReader_GPU::VideoSource::format
+-----------------------------------------
+Returns information about video file format.
+
+.. ocv:function:: virtual FormatInfo gpu::VideoReader_GPU::VideoSource::format() const = 0
+
+
+
+gpu::VideoReader_GPU::VideoSource::start
+----------------------------------------
+Starts processing.
+
+.. ocv:function:: virtual void gpu::VideoReader_GPU::VideoSource::start() = 0
+
+Implementation must create own thread with video processing and call periodic :ocv:func:`gpu::VideoReader_GPU::VideoSource::parseVideoData` .
+
+
+
+gpu::VideoReader_GPU::VideoSource::stop
+---------------------------------------
+Stops processing.
+
+.. ocv:function:: virtual void gpu::VideoReader_GPU::VideoSource::stop() = 0
+
+
+
+gpu::VideoReader_GPU::VideoSource::isStarted
+--------------------------------------------
+Returns ``true`` if processing was successfully started.
+
+.. ocv:function:: virtual bool gpu::VideoReader_GPU::VideoSource::isStarted() const = 0
+
+
+
+gpu::VideoReader_GPU::VideoSource::hasError
+-------------------------------------------
+Returns ``true`` if error occured during processing.
+
+.. ocv:function:: virtual bool gpu::VideoReader_GPU::VideoSource::hasError() const = 0
+
+
+
+gpu::VideoReader_GPU::VideoSource::parseVideoData
+-------------------------------------------------
+Parse next video frame. Implementation must call this method after new frame was grabbed.
+
+.. ocv:function:: bool gpu::VideoReader_GPU::VideoSource::parseVideoData(const uchar* data, size_t size, bool endOfStream = false)
+
+ :param data: Pointer to frame data. Can be ``NULL`` if ``endOfStream`` if ``true`` .
+
+ :param size: Size in bytes of current frame.
+
+ :param endOfStream: Indicates that it is end of stream.
--- /dev/null
+Video Encoding
+==============
+
+.. highlight:: cpp
+
+
+
+gpu::VideoWriter_GPU
+---------------------
+Video writer class.
+
+.. ocv:class:: gpu::VideoWriter_GPU
+
+The class uses H264 video codec.
+
+.. note:: Currently only Windows platform is supported.
+
+
+
+gpu::VideoWriter_GPU::VideoWriter_GPU
+-------------------------------------
+Constructors.
+
+.. ocv:function:: gpu::VideoWriter_GPU::VideoWriter_GPU()
+.. ocv:function:: gpu::VideoWriter_GPU::VideoWriter_GPU(const String& fileName, cv::Size frameSize, double fps, SurfaceFormat format = SF_BGR)
+.. ocv:function:: gpu::VideoWriter_GPU::VideoWriter_GPU(const String& fileName, cv::Size frameSize, double fps, const EncoderParams& params, SurfaceFormat format = SF_BGR)
+.. ocv:function:: gpu::VideoWriter_GPU::VideoWriter_GPU(const cv::Ptr<EncoderCallBack>& encoderCallback, cv::Size frameSize, double fps, SurfaceFormat format = SF_BGR)
+.. ocv:function:: gpu::VideoWriter_GPU::VideoWriter_GPU(const cv::Ptr<EncoderCallBack>& encoderCallback, cv::Size frameSize, double fps, const EncoderParams& params, SurfaceFormat format = SF_BGR)
+
+ :param fileName: Name of the output video file. Only AVI file format is supported.
+
+ :param frameSize: Size of the input video frames.
+
+ :param fps: Framerate of the created video stream.
+
+ :param params: Encoder parameters. See :ocv:struct:`gpu::VideoWriter_GPU::EncoderParams` .
+
+ :param format: Surface format of input frames ( ``SF_UYVY`` , ``SF_YUY2`` , ``SF_YV12`` , ``SF_NV12`` , ``SF_IYUV`` , ``SF_BGR`` or ``SF_GRAY``). BGR or gray frames will be converted to YV12 format before encoding, frames with other formats will be used as is.
+
+ :param encoderCallback: Callbacks for video encoder. See :ocv:class:`gpu::VideoWriter_GPU::EncoderCallBack` . Use it if you want to work with raw video stream.
+
+The constructors initialize video writer. FFMPEG is used to write videos. User can implement own multiplexing with :ocv:class:`gpu::VideoWriter_GPU::EncoderCallBack` .
+
+
+
+gpu::VideoWriter_GPU::open
+--------------------------
+Initializes or reinitializes video writer.
+
+.. ocv:function:: void gpu::VideoWriter_GPU::open(const String& fileName, cv::Size frameSize, double fps, SurfaceFormat format = SF_BGR)
+.. ocv:function:: void gpu::VideoWriter_GPU::open(const String& fileName, cv::Size frameSize, double fps, const EncoderParams& params, SurfaceFormat format = SF_BGR)
+.. ocv:function:: void gpu::VideoWriter_GPU::open(const cv::Ptr<EncoderCallBack>& encoderCallback, cv::Size frameSize, double fps, SurfaceFormat format = SF_BGR)
+.. ocv:function:: void gpu::VideoWriter_GPU::open(const cv::Ptr<EncoderCallBack>& encoderCallback, cv::Size frameSize, double fps, const EncoderParams& params, SurfaceFormat format = SF_BGR)
+
+The method opens video writer. Parameters are the same as in the constructor :ocv:func:`gpu::VideoWriter_GPU::VideoWriter_GPU` . The method throws :ocv:class:`Exception` if error occurs.
+
+
+
+gpu::VideoWriter_GPU::isOpened
+------------------------------
+Returns true if video writer has been successfully initialized.
+
+.. ocv:function:: bool gpu::VideoWriter_GPU::isOpened() const
+
+
+
+gpu::VideoWriter_GPU::close
+---------------------------
+Releases the video writer.
+
+.. ocv:function:: void gpu::VideoWriter_GPU::close()
+
+
+
+gpu::VideoWriter_GPU::write
+---------------------------
+Writes the next video frame.
+
+.. ocv:function:: void gpu::VideoWriter_GPU::write(const cv::gpu::GpuMat& image, bool lastFrame = false)
+
+ :param image: The written frame.
+
+ :param lastFrame: Indicates that it is end of stream. The parameter can be ignored.
+
+The method write the specified image to video file. The image must have the same size and the same surface format as has been specified when opening the video writer.
+
+
+
+gpu::VideoWriter_GPU::EncoderParams
+-----------------------------------
+.. ocv:struct:: gpu::VideoWriter_GPU::EncoderParams
+
+Different parameters for CUDA video encoder. ::
+
+ struct EncoderParams
+ {
+ int P_Interval; // NVVE_P_INTERVAL,
+ int IDR_Period; // NVVE_IDR_PERIOD,
+ int DynamicGOP; // NVVE_DYNAMIC_GOP,
+ int RCType; // NVVE_RC_TYPE,
+ int AvgBitrate; // NVVE_AVG_BITRATE,
+ int PeakBitrate; // NVVE_PEAK_BITRATE,
+ int QP_Level_Intra; // NVVE_QP_LEVEL_INTRA,
+ int QP_Level_InterP; // NVVE_QP_LEVEL_INTER_P,
+ int QP_Level_InterB; // NVVE_QP_LEVEL_INTER_B,
+ int DeblockMode; // NVVE_DEBLOCK_MODE,
+ int ProfileLevel; // NVVE_PROFILE_LEVEL,
+ int ForceIntra; // NVVE_FORCE_INTRA,
+ int ForceIDR; // NVVE_FORCE_IDR,
+ int ClearStat; // NVVE_CLEAR_STAT,
+ int DIMode; // NVVE_SET_DEINTERLACE,
+ int Presets; // NVVE_PRESETS,
+ int DisableCabac; // NVVE_DISABLE_CABAC,
+ int NaluFramingType; // NVVE_CONFIGURE_NALU_FRAMING_TYPE
+ int DisableSPSPPS; // NVVE_DISABLE_SPS_PPS
+
+ EncoderParams();
+ explicit EncoderParams(const String& configFile);
+
+ void load(const String& configFile);
+ void save(const String& configFile) const;
+ };
+
+
+
+gpu::VideoWriter_GPU::EncoderParams::EncoderParams
+--------------------------------------------------
+Constructors.
+
+.. ocv:function:: gpu::VideoWriter_GPU::EncoderParams::EncoderParams()
+.. ocv:function:: gpu::VideoWriter_GPU::EncoderParams::EncoderParams(const String& configFile)
+
+ :param configFile: Config file name.
+
+Creates default parameters or reads parameters from config file.
+
+
+
+gpu::VideoWriter_GPU::EncoderParams::load
+-----------------------------------------
+Reads parameters from config file.
+
+.. ocv:function:: void gpu::VideoWriter_GPU::EncoderParams::load(const String& configFile)
+
+ :param configFile: Config file name.
+
+
+
+gpu::VideoWriter_GPU::EncoderParams::save
+-----------------------------------------
+Saves parameters to config file.
+
+.. ocv:function:: void gpu::VideoWriter_GPU::EncoderParams::save(const String& configFile) const
+
+ :param configFile: Config file name.
+
+
+
+gpu::VideoWriter_GPU::EncoderCallBack
+-------------------------------------
+.. ocv:class:: gpu::VideoWriter_GPU::EncoderCallBack
+
+Callbacks for CUDA video encoder. ::
+
+ class EncoderCallBack
+ {
+ public:
+ enum PicType
+ {
+ IFRAME = 1,
+ PFRAME = 2,
+ BFRAME = 3
+ };
+
+ virtual ~EncoderCallBack() {}
+
+ virtual unsigned char* acquireBitStream(int* bufferSize) = 0;
+ virtual void releaseBitStream(unsigned char* data, int size) = 0;
+ virtual void onBeginFrame(int frameNumber, PicType picType) = 0;
+ virtual void onEndFrame(int frameNumber, PicType picType) = 0;
+ };
+
+
+
+gpu::VideoWriter_GPU::EncoderCallBack::acquireBitStream
+-------------------------------------------------------
+Callback function to signal the start of bitstream that is to be encoded.
+
+.. ocv:function:: virtual uchar* gpu::VideoWriter_GPU::EncoderCallBack::acquireBitStream(int* bufferSize) = 0
+
+Callback must allocate buffer for CUDA encoder and return pointer to it and it's size.
+
+
+
+gpu::VideoWriter_GPU::EncoderCallBack::releaseBitStream
+-------------------------------------------------------
+Callback function to signal that the encoded bitstream is ready to be written to file.
+
+.. ocv:function:: virtual void gpu::VideoWriter_GPU::EncoderCallBack::releaseBitStream(unsigned char* data, int size) = 0
+
+
+
+gpu::VideoWriter_GPU::EncoderCallBack::onBeginFrame
+---------------------------------------------------
+Callback function to signal that the encoding operation on the frame has started.
+
+.. ocv:function:: virtual void gpu::VideoWriter_GPU::EncoderCallBack::onBeginFrame(int frameNumber, PicType picType) = 0
+
+ :param picType: Specify frame type (I-Frame, P-Frame or B-Frame).
+
+
+
+gpu::VideoWriter_GPU::EncoderCallBack::onEndFrame
+-------------------------------------------------
+Callback function signals that the encoding operation on the frame has finished.
+
+.. ocv:function:: virtual void gpu::VideoWriter_GPU::EncoderCallBack::onEndFrame(int frameNumber, PicType picType) = 0
+
+ :param picType: Specify frame type (I-Frame, P-Frame or B-Frame).
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_GPUCODEC_HPP__
+#define __OPENCV_GPUCODEC_HPP__
+
+#ifndef __cplusplus
+# error gpucodec.hpp header must be compiled as C++
+#endif
+
+#include <iosfwd>
+
+#include "opencv2/core/gpumat.hpp"
+
+namespace cv { namespace gpu {
+
+////////////////////////////////// Video Encoding //////////////////////////////////
+
+// Works only under Windows
+// Supports olny H264 video codec and AVI files
+class CV_EXPORTS VideoWriter_GPU
+{
+public:
+ struct EncoderParams;
+
+ // Callbacks for video encoder, use it if you want to work with raw video stream
+ class EncoderCallBack;
+
+ enum SurfaceFormat
+ {
+ SF_UYVY = 0,
+ SF_YUY2,
+ SF_YV12,
+ SF_NV12,
+ SF_IYUV,
+ SF_BGR,
+ SF_GRAY = SF_BGR
+ };
+
+ VideoWriter_GPU();
+ VideoWriter_GPU(const String& fileName, cv::Size frameSize, double fps, SurfaceFormat format = SF_BGR);
+ VideoWriter_GPU(const String& fileName, cv::Size frameSize, double fps, const EncoderParams& params, SurfaceFormat format = SF_BGR);
+ VideoWriter_GPU(const cv::Ptr<EncoderCallBack>& encoderCallback, cv::Size frameSize, double fps, SurfaceFormat format = SF_BGR);
+ VideoWriter_GPU(const cv::Ptr<EncoderCallBack>& encoderCallback, cv::Size frameSize, double fps, const EncoderParams& params, SurfaceFormat format = SF_BGR);
+ ~VideoWriter_GPU();
+
+ // all methods throws cv::Exception if error occurs
+ void open(const String& fileName, cv::Size frameSize, double fps, SurfaceFormat format = SF_BGR);
+ void open(const String& fileName, cv::Size frameSize, double fps, const EncoderParams& params, SurfaceFormat format = SF_BGR);
+ void open(const cv::Ptr<EncoderCallBack>& encoderCallback, cv::Size frameSize, double fps, SurfaceFormat format = SF_BGR);
+ void open(const cv::Ptr<EncoderCallBack>& encoderCallback, cv::Size frameSize, double fps, const EncoderParams& params, SurfaceFormat format = SF_BGR);
+
+ bool isOpened() const;
+ void close();
+
+ void write(const cv::gpu::GpuMat& image, bool lastFrame = false);
+
+ struct CV_EXPORTS EncoderParams
+ {
+ int P_Interval; // NVVE_P_INTERVAL,
+ int IDR_Period; // NVVE_IDR_PERIOD,
+ int DynamicGOP; // NVVE_DYNAMIC_GOP,
+ int RCType; // NVVE_RC_TYPE,
+ int AvgBitrate; // NVVE_AVG_BITRATE,
+ int PeakBitrate; // NVVE_PEAK_BITRATE,
+ int QP_Level_Intra; // NVVE_QP_LEVEL_INTRA,
+ int QP_Level_InterP; // NVVE_QP_LEVEL_INTER_P,
+ int QP_Level_InterB; // NVVE_QP_LEVEL_INTER_B,
+ int DeblockMode; // NVVE_DEBLOCK_MODE,
+ int ProfileLevel; // NVVE_PROFILE_LEVEL,
+ int ForceIntra; // NVVE_FORCE_INTRA,
+ int ForceIDR; // NVVE_FORCE_IDR,
+ int ClearStat; // NVVE_CLEAR_STAT,
+ int DIMode; // NVVE_SET_DEINTERLACE,
+ int Presets; // NVVE_PRESETS,
+ int DisableCabac; // NVVE_DISABLE_CABAC,
+ int NaluFramingType; // NVVE_CONFIGURE_NALU_FRAMING_TYPE
+ int DisableSPSPPS; // NVVE_DISABLE_SPS_PPS
+
+ EncoderParams();
+ explicit EncoderParams(const String& configFile);
+
+ void load(const String& configFile);
+ void save(const String& configFile) const;
+ };
+
+ EncoderParams getParams() const;
+
+ class CV_EXPORTS EncoderCallBack
+ {
+ public:
+ enum PicType
+ {
+ IFRAME = 1,
+ PFRAME = 2,
+ BFRAME = 3
+ };
+
+ virtual ~EncoderCallBack() {}
+
+ // callback function to signal the start of bitstream that is to be encoded
+ // must return pointer to buffer
+ virtual uchar* acquireBitStream(int* bufferSize) = 0;
+
+ // callback function to signal that the encoded bitstream is ready to be written to file
+ virtual void releaseBitStream(unsigned char* data, int size) = 0;
+
+ // callback function to signal that the encoding operation on the frame has started
+ virtual void onBeginFrame(int frameNumber, PicType picType) = 0;
+
+ // callback function signals that the encoding operation on the frame has finished
+ virtual void onEndFrame(int frameNumber, PicType picType) = 0;
+ };
+
+ class Impl;
+
+private:
+ cv::Ptr<Impl> impl_;
+};
+
+////////////////////////////////// Video Decoding //////////////////////////////////////////
+
+namespace detail
+{
+ class FrameQueue;
+ class VideoParser;
+}
+
+class CV_EXPORTS VideoReader_GPU
+{
+public:
+ enum Codec
+ {
+ MPEG1 = 0,
+ MPEG2,
+ MPEG4,
+ VC1,
+ H264,
+ JPEG,
+ H264_SVC,
+ H264_MVC,
+
+ Uncompressed_YUV420 = (('I'<<24)|('Y'<<16)|('U'<<8)|('V')), // Y,U,V (4:2:0)
+ Uncompressed_YV12 = (('Y'<<24)|('V'<<16)|('1'<<8)|('2')), // Y,V,U (4:2:0)
+ Uncompressed_NV12 = (('N'<<24)|('V'<<16)|('1'<<8)|('2')), // Y,UV (4:2:0)
+ Uncompressed_YUYV = (('Y'<<24)|('U'<<16)|('Y'<<8)|('V')), // YUYV/YUY2 (4:2:2)
+ Uncompressed_UYVY = (('U'<<24)|('Y'<<16)|('V'<<8)|('Y')), // UYVY (4:2:2)
+ };
+
+ enum ChromaFormat
+ {
+ Monochrome=0,
+ YUV420,
+ YUV422,
+ YUV444,
+ };
+
+ struct FormatInfo
+ {
+ Codec codec;
+ ChromaFormat chromaFormat;
+ int width;
+ int height;
+ };
+
+ class VideoSource;
+
+ VideoReader_GPU();
+ explicit VideoReader_GPU(const String& filename);
+ explicit VideoReader_GPU(const cv::Ptr<VideoSource>& source);
+
+ ~VideoReader_GPU();
+
+ void open(const String& filename);
+ void open(const cv::Ptr<VideoSource>& source);
+ bool isOpened() const;
+
+ void close();
+
+ bool read(GpuMat& image);
+
+ FormatInfo format() const;
+ void dumpFormat(std::ostream& st);
+
+ class CV_EXPORTS VideoSource
+ {
+ public:
+ VideoSource() : frameQueue_(0), videoParser_(0) {}
+ virtual ~VideoSource() {}
+
+ virtual FormatInfo format() const = 0;
+ virtual void start() = 0;
+ virtual void stop() = 0;
+ virtual bool isStarted() const = 0;
+ virtual bool hasError() const = 0;
+
+ void setFrameQueue(detail::FrameQueue* frameQueue) { frameQueue_ = frameQueue; }
+ void setVideoParser(detail::VideoParser* videoParser) { videoParser_ = videoParser; }
+
+ protected:
+ bool parseVideoData(const uchar* data, size_t size, bool endOfStream = false);
+
+ private:
+ VideoSource(const VideoSource&);
+ VideoSource& operator =(const VideoSource&);
+
+ detail::FrameQueue* frameQueue_;
+ detail::VideoParser* videoParser_;
+ };
+
+ class Impl;
+
+private:
+ cv::Ptr<Impl> impl_;
+};
+
+}} // namespace cv { namespace gpu {
+
+namespace cv {
+
+template <> CV_EXPORTS void Ptr<cv::gpu::VideoWriter_GPU::Impl>::delete_obj();
+template <> CV_EXPORTS void Ptr<cv::gpu::VideoReader_GPU::Impl>::delete_obj();
+
+}
+
+#endif /* __OPENCV_GPUCODEC_HPP__ */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace perf;
+
+CV_PERF_TEST_MAIN(gpucodec, printCudaInfo())
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifdef __GNUC__
+# pragma GCC diagnostic ignored "-Wmissing-declarations"
+# if defined __clang__ || defined __APPLE__
+# pragma GCC diagnostic ignored "-Wmissing-prototypes"
+# pragma GCC diagnostic ignored "-Wextra"
+# endif
+#endif
+
+#ifndef __OPENCV_PERF_PRECOMP_HPP__
+#define __OPENCV_PERF_PRECOMP_HPP__
+
+#include "opencv2/ts.hpp"
+#include "opencv2/ts/gpu_perf.hpp"
+
+#include "opencv2/gpucodec.hpp"
+#include "opencv2/highgui.hpp"
+
+#ifdef GTEST_CREATE_SHARED_LIBRARY
+#error no modules except ts should have GTEST_CREATE_SHARED_LIBRARY defined
+#endif
+
+#endif
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+#include "opencv2/highgui/highgui_c.h"
+
+using namespace std;
+using namespace testing;
+using namespace perf;
+
+#if defined(HAVE_XINE) || \
+ defined(HAVE_GSTREAMER) || \
+ defined(HAVE_QUICKTIME) || \
+ defined(HAVE_AVFOUNDATION) || \
+ defined(HAVE_FFMPEG) || \
+ defined(WIN32) /* assume that we have ffmpeg */
+
+# define BUILD_WITH_VIDEO_INPUT_SUPPORT 1
+#else
+# define BUILD_WITH_VIDEO_INPUT_SUPPORT 0
+#endif
+
+DEF_PARAM_TEST_1(FileName, string);
+
+//////////////////////////////////////////////////////
+// VideoReader
+
+#if defined(HAVE_NVCUVID) && BUILD_WITH_VIDEO_INPUT_SUPPORT
+
+PERF_TEST_P(FileName, VideoReader, Values("gpu/video/768x576.avi", "gpu/video/1920x1080.avi"))
+{
+ declare.time(20);
+
+ const string inputFile = perf::TestBase::getDataPath(GetParam());
+
+ if (PERF_RUN_GPU())
+ {
+ cv::gpu::VideoReader_GPU d_reader(inputFile);
+ ASSERT_TRUE( d_reader.isOpened() );
+
+ cv::gpu::GpuMat frame;
+
+ TEST_CYCLE_N(10) d_reader.read(frame);
+
+ GPU_SANITY_CHECK(frame);
+ }
+ else
+ {
+ cv::VideoCapture reader(inputFile);
+ ASSERT_TRUE( reader.isOpened() );
+
+ cv::Mat frame;
+
+ TEST_CYCLE_N(10) reader >> frame;
+
+ CPU_SANITY_CHECK(frame);
+ }
+}
+
+#endif
+
+//////////////////////////////////////////////////////
+// VideoWriter
+
+#if defined(HAVE_NVCUVID) && defined(WIN32)
+
+PERF_TEST_P(FileName, VideoWriter, Values("gpu/video/768x576.avi", "gpu/video/1920x1080.avi"))
+{
+ declare.time(30);
+
+ const string inputFile = perf::TestBase::getDataPath(GetParam());
+ const string outputFile = cv::tempfile(".avi");
+
+ const double FPS = 25.0;
+
+ cv::VideoCapture reader(inputFile);
+ ASSERT_TRUE( reader.isOpened() );
+
+ cv::Mat frame;
+
+ if (PERF_RUN_GPU())
+ {
+ cv::gpu::VideoWriter_GPU d_writer;
+
+ cv::gpu::GpuMat d_frame;
+
+ for (int i = 0; i < 10; ++i)
+ {
+ reader >> frame;
+ ASSERT_FALSE(frame.empty());
+
+ d_frame.upload(frame);
+
+ if (!d_writer.isOpened())
+ d_writer.open(outputFile, frame.size(), FPS);
+
+ startTimer(); next();
+ d_writer.write(d_frame);
+ stopTimer();
+ }
+ }
+ else
+ {
+ cv::VideoWriter writer;
+
+ for (int i = 0; i < 10; ++i)
+ {
+ reader >> frame;
+ ASSERT_FALSE(frame.empty());
+
+ if (!writer.isOpened())
+ writer.open(outputFile, CV_FOURCC('X', 'V', 'I', 'D'), FPS, frame.size());
+
+ startTimer(); next();
+ writer.write(frame);
+ stopTimer();
+ }
+ }
+
+ SANITY_CHECK(frame);
+}
+
+#endif
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+/*
+ * NV12ToARGB color space conversion CUDA kernel
+ *
+ * This sample uses CUDA to perform a simple NV12 (YUV 4:2:0 planar)
+ * source and converts to output in ARGB format
+ */
+
+#include "opencv2/core/cuda/common.hpp"
+
+namespace cv { namespace gpu { namespace cudev
+{
+ __constant__ float constHueColorSpaceMat[9];
+
+ void loadHueCSC(float hueCSC[9])
+ {
+ cudaSafeCall( cudaMemcpyToSymbol(constHueColorSpaceMat, hueCSC, 9 * sizeof(float)) );
+ }
+
+ __device__ void YUV2RGB(const uint* yuvi, float* red, float* green, float* blue)
+ {
+ float luma, chromaCb, chromaCr;
+
+ // Prepare for hue adjustment
+ luma = (float)yuvi[0];
+ chromaCb = (float)((int)yuvi[1] - 512.0f);
+ chromaCr = (float)((int)yuvi[2] - 512.0f);
+
+ // Convert YUV To RGB with hue adjustment
+ *red = (luma * constHueColorSpaceMat[0]) +
+ (chromaCb * constHueColorSpaceMat[1]) +
+ (chromaCr * constHueColorSpaceMat[2]);
+
+ *green = (luma * constHueColorSpaceMat[3]) +
+ (chromaCb * constHueColorSpaceMat[4]) +
+ (chromaCr * constHueColorSpaceMat[5]);
+
+ *blue = (luma * constHueColorSpaceMat[6]) +
+ (chromaCb * constHueColorSpaceMat[7]) +
+ (chromaCr * constHueColorSpaceMat[8]);
+ }
+
+ __device__ uint RGBA_pack_10bit(float red, float green, float blue, uint alpha)
+ {
+ uint ARGBpixel = 0;
+
+ // Clamp final 10 bit results
+ red = ::fmin(::fmax(red, 0.0f), 1023.f);
+ green = ::fmin(::fmax(green, 0.0f), 1023.f);
+ blue = ::fmin(::fmax(blue, 0.0f), 1023.f);
+
+ // Convert to 8 bit unsigned integers per color component
+ ARGBpixel = (((uint)blue >> 2) |
+ (((uint)green >> 2) << 8) |
+ (((uint)red >> 2) << 16) |
+ (uint)alpha);
+
+ return ARGBpixel;
+ }
+
+ // CUDA kernel for outputing the final ARGB output from NV12
+
+ #define COLOR_COMPONENT_BIT_SIZE 10
+ #define COLOR_COMPONENT_MASK 0x3FF
+
+ __global__ void NV12_to_RGB(uchar* srcImage, size_t nSourcePitch,
+ uint* dstImage, size_t nDestPitch,
+ uint width, uint height)
+ {
+ // Pad borders with duplicate pixels, and we multiply by 2 because we process 2 pixels per thread
+ const int x = blockIdx.x * (blockDim.x << 1) + (threadIdx.x << 1);
+ const int y = blockIdx.y * blockDim.y + threadIdx.y;
+
+ if (x >= width || y >= height)
+ return;
+
+ // Read 2 Luma components at a time, so we don't waste processing since CbCr are decimated this way.
+ // if we move to texture we could read 4 luminance values
+
+ uint yuv101010Pel[2];
+
+ yuv101010Pel[0] = (srcImage[y * nSourcePitch + x ]) << 2;
+ yuv101010Pel[1] = (srcImage[y * nSourcePitch + x + 1]) << 2;
+
+ const size_t chromaOffset = nSourcePitch * height;
+
+ const int y_chroma = y >> 1;
+
+ if (y & 1) // odd scanline ?
+ {
+ uint chromaCb = srcImage[chromaOffset + y_chroma * nSourcePitch + x ];
+ uint chromaCr = srcImage[chromaOffset + y_chroma * nSourcePitch + x + 1];
+
+ if (y_chroma < ((height >> 1) - 1)) // interpolate chroma vertically
+ {
+ chromaCb = (chromaCb + srcImage[chromaOffset + (y_chroma + 1) * nSourcePitch + x ] + 1) >> 1;
+ chromaCr = (chromaCr + srcImage[chromaOffset + (y_chroma + 1) * nSourcePitch + x + 1] + 1) >> 1;
+ }
+
+ yuv101010Pel[0] |= (chromaCb << ( COLOR_COMPONENT_BIT_SIZE + 2));
+ yuv101010Pel[0] |= (chromaCr << ((COLOR_COMPONENT_BIT_SIZE << 1) + 2));
+
+ yuv101010Pel[1] |= (chromaCb << ( COLOR_COMPONENT_BIT_SIZE + 2));
+ yuv101010Pel[1] |= (chromaCr << ((COLOR_COMPONENT_BIT_SIZE << 1) + 2));
+ }
+ else
+ {
+ yuv101010Pel[0] |= ((uint)srcImage[chromaOffset + y_chroma * nSourcePitch + x ] << ( COLOR_COMPONENT_BIT_SIZE + 2));
+ yuv101010Pel[0] |= ((uint)srcImage[chromaOffset + y_chroma * nSourcePitch + x + 1] << ((COLOR_COMPONENT_BIT_SIZE << 1) + 2));
+
+ yuv101010Pel[1] |= ((uint)srcImage[chromaOffset + y_chroma * nSourcePitch + x ] << ( COLOR_COMPONENT_BIT_SIZE + 2));
+ yuv101010Pel[1] |= ((uint)srcImage[chromaOffset + y_chroma * nSourcePitch + x + 1] << ((COLOR_COMPONENT_BIT_SIZE << 1) + 2));
+ }
+
+ // this steps performs the color conversion
+ uint yuvi[6];
+ float red[2], green[2], blue[2];
+
+ yuvi[0] = (yuv101010Pel[0] & COLOR_COMPONENT_MASK );
+ yuvi[1] = ((yuv101010Pel[0] >> COLOR_COMPONENT_BIT_SIZE) & COLOR_COMPONENT_MASK);
+ yuvi[2] = ((yuv101010Pel[0] >> (COLOR_COMPONENT_BIT_SIZE << 1)) & COLOR_COMPONENT_MASK);
+
+ yuvi[3] = (yuv101010Pel[1] & COLOR_COMPONENT_MASK );
+ yuvi[4] = ((yuv101010Pel[1] >> COLOR_COMPONENT_BIT_SIZE) & COLOR_COMPONENT_MASK);
+ yuvi[5] = ((yuv101010Pel[1] >> (COLOR_COMPONENT_BIT_SIZE << 1)) & COLOR_COMPONENT_MASK);
+
+ // YUV to RGB Transformation conversion
+ YUV2RGB(&yuvi[0], &red[0], &green[0], &blue[0]);
+ YUV2RGB(&yuvi[3], &red[1], &green[1], &blue[1]);
+
+ // Clamp the results to RGBA
+
+ const size_t dstImagePitch = nDestPitch >> 2;
+
+ dstImage[y * dstImagePitch + x ] = RGBA_pack_10bit(red[0], green[0], blue[0], ((uint)0xff << 24));
+ dstImage[y * dstImagePitch + x + 1 ] = RGBA_pack_10bit(red[1], green[1], blue[1], ((uint)0xff << 24));
+ }
+
+ void NV12_to_RGB(const PtrStepb decodedFrame, PtrStepSz<uint> interopFrame, cudaStream_t stream)
+ {
+ dim3 block(32, 8);
+ dim3 grid(divUp(interopFrame.cols, 2 * block.x), divUp(interopFrame.rows, block.y));
+
+ NV12_to_RGB<<<grid, block, 0, stream>>>(decodedFrame.data, decodedFrame.step, interopFrame.data, interopFrame.step,
+ interopFrame.cols, interopFrame.rows);
+
+ cudaSafeCall( cudaGetLastError() );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+}}}
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/vec_traits.hpp"
+
+namespace cv { namespace gpu { namespace cudev
+{
+ __device__ __forceinline__ void rgb_to_y(const uchar b, const uchar g, const uchar r, uchar& y)
+ {
+ y = static_cast<uchar>(((int)(30 * r) + (int)(59 * g) + (int)(11 * b)) / 100);
+ }
+
+ __device__ __forceinline__ void rgb_to_yuv(const uchar b, const uchar g, const uchar r, uchar& y, uchar& u, uchar& v)
+ {
+ rgb_to_y(b, g, r, y);
+ u = static_cast<uchar>(((int)(-17 * r) - (int)(33 * g) + (int)(50 * b) + 12800) / 100);
+ v = static_cast<uchar>(((int)(50 * r) - (int)(42 * g) - (int)(8 * b) + 12800) / 100);
+ }
+
+ __global__ void Gray_to_YV12(const PtrStepSzb src, PtrStepb dst)
+ {
+ const int x = (blockIdx.x * blockDim.x + threadIdx.x) * 2;
+ const int y = (blockIdx.y * blockDim.y + threadIdx.y) * 2;
+
+ if (x + 1 >= src.cols || y + 1 >= src.rows)
+ return;
+
+ // get pointers to the data
+ const size_t planeSize = src.rows * dst.step;
+ PtrStepb y_plane(dst.data, dst.step);
+ PtrStepb u_plane(y_plane.data + planeSize, dst.step / 2);
+ PtrStepb v_plane(u_plane.data + (planeSize / 4), dst.step / 2);
+
+ uchar pix;
+ uchar y_val, u_val, v_val;
+
+ pix = src(y, x);
+ rgb_to_y(pix, pix, pix, y_val);
+ y_plane(y, x) = y_val;
+
+ pix = src(y, x + 1);
+ rgb_to_y(pix, pix, pix, y_val);
+ y_plane(y, x + 1) = y_val;
+
+ pix = src(y + 1, x);
+ rgb_to_y(pix, pix, pix, y_val);
+ y_plane(y + 1, x) = y_val;
+
+ pix = src(y + 1, x + 1);
+ rgb_to_yuv(pix, pix, pix, y_val, u_val, v_val);
+ y_plane(y + 1, x + 1) = y_val;
+ u_plane(y / 2, x / 2) = u_val;
+ v_plane(y / 2, x / 2) = v_val;
+ }
+
+ template <typename T>
+ __global__ void RGB_to_YV12(const PtrStepSz<T> src, PtrStepb dst)
+ {
+ const int x = (blockIdx.x * blockDim.x + threadIdx.x) * 2;
+ const int y = (blockIdx.y * blockDim.y + threadIdx.y) * 2;
+
+ if (x + 1 >= src.cols || y + 1 >= src.rows)
+ return;
+
+ // get pointers to the data
+ const size_t planeSize = src.rows * dst.step;
+ PtrStepb y_plane(dst.data, dst.step);
+ PtrStepb u_plane(y_plane.data + planeSize, dst.step / 2);
+ PtrStepb v_plane(u_plane.data + (planeSize / 4), dst.step / 2);
+
+ T pix;
+ uchar y_val, u_val, v_val;
+
+ pix = src(y, x);
+ rgb_to_y(pix.z, pix.y, pix.x, y_val);
+ y_plane(y, x) = y_val;
+
+ pix = src(y, x + 1);
+ rgb_to_y(pix.z, pix.y, pix.x, y_val);
+ y_plane(y, x + 1) = y_val;
+
+ pix = src(y + 1, x);
+ rgb_to_y(pix.z, pix.y, pix.x, y_val);
+ y_plane(y + 1, x) = y_val;
+
+ pix = src(y + 1, x + 1);
+ rgb_to_yuv(pix.z, pix.y, pix.x, y_val, u_val, v_val);
+ y_plane(y + 1, x + 1) = y_val;
+ u_plane(y / 2, x / 2) = u_val;
+ v_plane(y / 2, x / 2) = v_val;
+ }
+
+ void Gray_to_YV12_caller(const PtrStepSzb src, PtrStepb dst, cudaStream_t stream)
+ {
+ dim3 block(32, 8);
+ dim3 grid(divUp(src.cols, block.x * 2), divUp(src.rows, block.y * 2));
+
+ Gray_to_YV12<<<grid, block, 0, stream>>>(src, dst);
+ cudaSafeCall( cudaGetLastError() );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+ template <int cn>
+ void RGB_to_YV12_caller(const PtrStepSzb src, PtrStepb dst, cudaStream_t stream)
+ {
+ typedef typename TypeVec<uchar, cn>::vec_type src_t;
+
+ dim3 block(32, 8);
+ dim3 grid(divUp(src.cols, block.x * 2), divUp(src.rows, block.y * 2));
+
+ RGB_to_YV12<<<grid, block, 0, stream>>>(static_cast< PtrStepSz<src_t> >(src), dst);
+ cudaSafeCall( cudaGetLastError() );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+
+ void RGB_to_YV12(const PtrStepSzb src, int cn, PtrStepSzb dst, cudaStream_t stream)
+ {
+ typedef void (*func_t)(const PtrStepSzb src, PtrStepb dst, cudaStream_t stream);
+
+ static const func_t funcs[] =
+ {
+ 0, Gray_to_YV12_caller, 0, RGB_to_YV12_caller<3>, RGB_to_YV12_caller<4>
+ };
+
+ funcs[cn](src, dst, stream);
+ }
+}}}
//
//M*/
-#include "cuvid_video_source.h"
+#include "precomp.hpp"
-#if defined(HAVE_CUDA) && defined(HAVE_NVCUVID)
+#ifdef HAVE_NVCUVID
cv::gpu::detail::CuvidVideoSource::CuvidVideoSource(const String& fname)
{
format_.height = vidfmt.coded_height;
}
+cv::gpu::detail::CuvidVideoSource::~CuvidVideoSource()
+{
+ cuvidDestroyVideoSource(videoSource_);
+}
+
cv::gpu::VideoReader_GPU::FormatInfo cv::gpu::detail::CuvidVideoSource::format() const
{
return format_;
return thiz->parseVideoData(packet->payload, packet->payload_size, (packet->flags & CUVID_PKT_ENDOFSTREAM) != 0);
}
-#endif // defined(HAVE_CUDA) && !defined(__APPLE__)
+#endif // HAVE_NVCUVID
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __CUVUD_VIDEO_SOURCE_H__
+#define __CUVUD_VIDEO_SOURCE_H__
+
+#include "opencv2/core/gpu_private.hpp"
+#include "opencv2/gpucodec.hpp"
+#include "thread.h"
+
+#include <nvcuvid.h>
+
+namespace cv { namespace gpu { namespace detail
+{
+
+class CuvidVideoSource : public VideoReader_GPU::VideoSource
+{
+public:
+ explicit CuvidVideoSource(const String& fname);
+ ~CuvidVideoSource();
+
+ VideoReader_GPU::FormatInfo format() const;
+ void start();
+ void stop();
+ bool isStarted() const;
+ bool hasError() const;
+
+private:
+ // Callback for handling packages of demuxed video data.
+ //
+ // Parameters:
+ // pUserData - Pointer to user data. We must pass a pointer to a
+ // VideoSourceData struct here, that contains a valid CUvideoparser
+ // and FrameQueue.
+ // pPacket - video-source data packet.
+ //
+ // NOTE: called from a different thread that doesn't not have a cuda context
+ //
+ static int CUDAAPI HandleVideoData(void* pUserData, CUVIDSOURCEDATAPACKET* pPacket);
+
+ CUvideosource videoSource_;
+ VideoReader_GPU::FormatInfo format_;
+};
+
+}}}
+
+#endif // __CUVUD_VIDEO_SOURCE_H__
//
//M*/
-#include "ffmpeg_video_source.h"
+#include "precomp.hpp"
-#if defined(HAVE_CUDA) && defined(HAVE_NVCUVID)
+#ifdef HAVE_NVCUVID
-#if defined(HAVE_FFMPEG) && defined(BUILD_SHARED_LIBS)
+#if defined(HAVE_FFMPEG) && defined(BUILD_SHARED_LIBS) && !defined(WIN32)
#include "../src/cap_ffmpeg_impl.hpp"
-#else
- #include "../src/cap_ffmpeg_api.hpp"
#endif
namespace
format_.height = height;
}
-cv::gpu::detail::FFmpegVideoSource::~FFmpegVideoSource()
-{
- release_InputMediaStream_FFMPEG_p(stream_);
-}
-
cv::gpu::VideoReader_GPU::FormatInfo cv::gpu::detail::FFmpegVideoSource::format() const
{
return format_;
{
stop_ = false;
hasError_ = false;
- thread_.reset(new Thread(readLoop, this));
+ thread_ = new Thread(readLoop, this);
}
void cv::gpu::detail::FFmpegVideoSource::stop()
{
stop_ = true;
thread_->wait();
- thread_.reset();
+ thread_.release();
}
bool cv::gpu::detail::FFmpegVideoSource::isStarted() const
thiz->parseVideoData(0, 0, true);
}
+template <> void cv::Ptr<InputMediaStream_FFMPEG>::delete_obj()
+{
+ if (obj) release_InputMediaStream_FFMPEG_p(obj);
+}
+
#endif // HAVE_CUDA
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __FFMPEG_VIDEO_SOURCE_H__
+#define __FFMPEG_VIDEO_SOURCE_H__
+
+#include "opencv2/gpucodec.hpp"
+#include "thread.h"
+
+struct InputMediaStream_FFMPEG;
+
+namespace cv { namespace gpu { namespace detail {
+
+class FFmpegVideoSource : public VideoReader_GPU::VideoSource
+{
+public:
+ FFmpegVideoSource(const String& fname);
+
+ VideoReader_GPU::FormatInfo format() const;
+ void start();
+ void stop();
+ bool isStarted() const;
+ bool hasError() const;
+
+private:
+ VideoReader_GPU::FormatInfo format_;
+
+ cv::Ptr<InputMediaStream_FFMPEG> stream_;
+
+ cv::Ptr<Thread> thread_;
+ volatile bool stop_;
+ volatile bool hasError_;
+
+ static void readLoop(void* userData);
+};
+
+}}}
+
+namespace cv {
+ template <> void Ptr<InputMediaStream_FFMPEG>::delete_obj();
+}
+
+#endif // __FFMPEG_VIDEO_SOURCE_H__
//
//M*/
-#include "frame_queue.h"
+#include "precomp.hpp"
-#if defined(HAVE_CUDA) && defined(HAVE_NVCUVID)
+#ifdef HAVE_NVCUVID
cv::gpu::detail::FrameQueue::FrameQueue() :
endOfDecode_(0),
bool isFramePlaced = false;
{
- CriticalSection::AutoLock autoLock(criticalSection_);
+ AutoLock autoLock(mtx_);
if (framesInQueue_ < MaximumSize)
{
bool cv::gpu::detail::FrameQueue::dequeue(CUVIDPARSERDISPINFO& displayInfo)
{
- CriticalSection::AutoLock autoLock(criticalSection_);
+ AutoLock autoLock(mtx_);
if (framesInQueue_ > 0)
{
return false;
}
-#endif // HAVE_CUDA
+#endif // HAVE_NVCUVID
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __FRAME_QUEUE_H__
+#define __FRAME_QUEUE_H__
+
+#include "opencv2/core/utility.hpp"
+#include "opencv2/core/gpu_private.hpp"
+
+#include <nvcuvid.h>
+
+namespace cv { namespace gpu { namespace detail
+{
+
+class FrameQueue
+{
+public:
+ static const int MaximumSize = 20; // MAX_FRM_CNT;
+
+ FrameQueue();
+
+ void endDecode() { endOfDecode_ = true; }
+ bool isEndOfDecode() const { return endOfDecode_ != 0;}
+
+ // Spins until frame becomes available or decoding gets canceled.
+ // If the requested frame is available the method returns true.
+ // If decoding was interupted before the requested frame becomes
+ // available, the method returns false.
+ bool waitUntilFrameAvailable(int pictureIndex);
+
+ void enqueue(const CUVIDPARSERDISPINFO* picParams);
+
+ // Deque the next frame.
+ // Parameters:
+ // displayInfo - New frame info gets placed into this object.
+ // Returns:
+ // true, if a new frame was returned,
+ // false, if the queue was empty and no new frame could be returned.
+ bool dequeue(CUVIDPARSERDISPINFO& displayInfo);
+
+ void releaseFrame(const CUVIDPARSERDISPINFO& picParams) { isFrameInUse_[picParams.picture_index] = false; }
+
+private:
+ bool isInUse(int pictureIndex) const { return isFrameInUse_[pictureIndex] != 0; }
+
+ Mutex mtx_;
+
+ volatile int isFrameInUse_[MaximumSize];
+ volatile int endOfDecode_;
+
+ int framesInQueue_;
+ int readPosition_;
+ CUVIDPARSERDISPINFO displayQueue_[MaximumSize];
+};
+
+}}}
+
+#endif // __FRAME_QUEUE_H__
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_PRECOMP_H__
+#define __OPENCV_PRECOMP_H__
+
+#include <cstdlib>
+#include <cstring>
+#include <deque>
+#include <utility>
+#include <stdexcept>
+#include <iostream>
+
+#include "opencv2/gpucodec.hpp"
+
+#include "opencv2/core/gpu_private.hpp"
+
+#ifdef HAVE_NVCUVID
+ #include <nvcuvid.h>
+
+ #ifdef WIN32
+ #define NOMINMAX
+ #include <windows.h>
+ #include <NVEncoderAPI.h>
+ #else
+ #include <pthread.h>
+ #include <unistd.h>
+ #endif
+
+ #include "thread.h"
+ #include "ffmpeg_video_source.h"
+ #include "cuvid_video_source.h"
+ #include "frame_queue.h"
+ #include "video_decoder.h"
+ #include "video_parser.h"
+
+ #include "../src/cap_ffmpeg_api.hpp"
+#endif
+
+#endif /* __OPENCV_PRECOMP_H__ */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "precomp.hpp"
+
+#ifdef HAVE_NVCUVID
+
+using namespace cv::gpu::detail;
+
+#ifdef WIN32
+
+namespace
+{
+ struct UserData
+ {
+ Thread::Func func;
+ void* param;
+ };
+
+ DWORD WINAPI WinThreadFunction(LPVOID lpParam)
+ {
+ UserData* userData = static_cast<UserData*>(lpParam);
+
+ userData->func(userData->param);
+
+ return 0;
+ }
+}
+
+class cv::gpu::detail::Thread::Impl
+{
+public:
+ Impl(Thread::Func func, void* userData)
+ {
+ userData_.func = func;
+ userData_.param = userData;
+
+ thread_ = CreateThread(
+ NULL, // default security attributes
+ 0, // use default stack size
+ WinThreadFunction, // thread function name
+ &userData_, // argument to thread function
+ 0, // use default creation flags
+ &threadId_); // returns the thread identifier
+ }
+
+ ~Impl()
+ {
+ CloseHandle(thread_);
+ }
+
+ void wait()
+ {
+ WaitForSingleObject(thread_, INFINITE);
+ }
+
+private:
+ UserData userData_;
+ HANDLE thread_;
+ DWORD threadId_;
+};
+
+#else
+
+namespace
+{
+ struct UserData
+ {
+ Thread::Func func;
+ void* param;
+ };
+
+ void* PThreadFunction(void* lpParam)
+ {
+ UserData* userData = static_cast<UserData*>(lpParam);
+
+ userData->func(userData->param);
+
+ return 0;
+ }
+}
+
+class cv::gpu::detail::Thread::Impl
+{
+public:
+ Impl(Thread::Func func, void* userData)
+ {
+ userData_.func = func;
+ userData_.param = userData;
+
+ pthread_create(&thread_, NULL, PThreadFunction, &userData_);
+ }
+
+ ~Impl()
+ {
+ pthread_detach(thread_);
+ }
+
+ void wait()
+ {
+ pthread_join(thread_, NULL);
+ }
+
+private:
+ pthread_t thread_;
+ UserData userData_;
+};
+
+#endif
+
+cv::gpu::detail::Thread::Thread(Func func, void* userData) :
+ impl_(new Impl(func, userData))
+{
+}
+
+void cv::gpu::detail::Thread::wait()
+{
+ impl_->wait();
+}
+
+void cv::gpu::detail::Thread::sleep(int ms)
+{
+#ifdef WIN32
+ ::Sleep(ms);
+#else
+ ::usleep(ms * 1000);
+#endif
+}
+
+template <> void cv::Ptr<cv::gpu::detail::Thread::Impl>::delete_obj()
+{
+ if (obj) delete obj;
+}
+
+#endif // HAVE_NVCUVID
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __THREAD_WRAPPERS_H__
+#define __THREAD_WRAPPERS_H__
+
+#include "opencv2/core.hpp"
+
+namespace cv { namespace gpu { namespace detail {
+
+class Thread
+{
+public:
+ typedef void (*Func)(void* userData);
+
+ explicit Thread(Func func, void* userData = 0);
+
+ void wait();
+
+ static void sleep(int ms);
+
+ class Impl;
+
+private:
+ cv::Ptr<Impl> impl_;
+};
+
+}}}
+
+namespace cv {
+ template <> void Ptr<cv::gpu::detail::Thread::Impl>::delete_obj();
+}
+
+#endif // __THREAD_WRAPPERS_H__
//
//M*/
-#include "video_decoder.h"
-#include "frame_queue.h"
+#include "precomp.hpp"
-#if defined(HAVE_CUDA) && defined(HAVE_NVCUVID)
+#ifdef HAVE_NVCUVID
void cv::gpu::detail::VideoDecoder::create(const VideoReader_GPU::FormatInfo& videoFormat)
{
}
}
-#endif // HAVE_CUDA
+#endif // HAVE_NVCUVID
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __VIDEO_DECODER_H__
+#define __VIDEO_DECODER_H__
+
+#include "opencv2/core/gpu_private.hpp"
+#include "opencv2/gpucodec.hpp"
+
+#include <nvcuvid.h>
+
+namespace cv { namespace gpu { namespace detail
+{
+
+class VideoDecoder
+{
+public:
+ VideoDecoder(const VideoReader_GPU::FormatInfo& videoFormat, CUvideoctxlock lock) : lock_(lock), decoder_(0)
+ {
+ create(videoFormat);
+ }
+
+ ~VideoDecoder()
+ {
+ release();
+ }
+
+ void create(const VideoReader_GPU::FormatInfo& videoFormat);
+ void release();
+
+ // Get the code-type currently used.
+ cudaVideoCodec codec() const { return createInfo_.CodecType; }
+ unsigned long maxDecodeSurfaces() const { return createInfo_.ulNumDecodeSurfaces; }
+
+ unsigned long frameWidth() const { return createInfo_.ulWidth; }
+ unsigned long frameHeight() const { return createInfo_.ulHeight; }
+
+ unsigned long targetWidth() const { return createInfo_.ulTargetWidth; }
+ unsigned long targetHeight() const { return createInfo_.ulTargetHeight; }
+
+ cudaVideoChromaFormat chromaFormat() const { return createInfo_.ChromaFormat; }
+
+ bool decodePicture(CUVIDPICPARAMS* picParams)
+ {
+ return cuvidDecodePicture(decoder_, picParams) == CUDA_SUCCESS;
+ }
+
+ cv::gpu::GpuMat mapFrame(int picIdx, CUVIDPROCPARAMS& videoProcParams)
+ {
+ CUdeviceptr ptr;
+ unsigned int pitch;
+
+ cuSafeCall( cuvidMapVideoFrame(decoder_, picIdx, &ptr, &pitch, &videoProcParams) );
+
+ return GpuMat(targetHeight() * 3 / 2, targetWidth(), CV_8UC1, (void*) ptr, pitch);
+ }
+
+ void unmapFrame(cv::gpu::GpuMat& frame)
+ {
+ cuSafeCall( cuvidUnmapVideoFrame(decoder_, (CUdeviceptr) frame.data) );
+ frame.release();
+ }
+
+private:
+ CUvideoctxlock lock_;
+ CUVIDDECODECREATEINFO createInfo_;
+ CUvideodecoder decoder_;
+};
+
+}}}
+
+#endif // __VIDEO_DECODER_H__
//
//M*/
-#include "video_parser.h"
+#include "precomp.hpp"
-#if defined(HAVE_CUDA) && defined(HAVE_NVCUVID)
+#ifdef HAVE_NVCUVID
cv::gpu::detail::VideoParser::VideoParser(VideoDecoder* videoDecoder, FrameQueue* frameQueue) :
videoDecoder_(videoDecoder), frameQueue_(frameQueue), unparsedPackets_(0), hasError_(false)
return true;
}
-#endif // HAVE_CUDA
+#endif // HAVE_NVCUVID
//
//M*/
-#ifndef __OPENCV_CUDA_SAFE_CALL_HPP__
-#define __OPENCV_CUDA_SAFE_CALL_HPP__
+#ifndef __VIDEO_PARSER_H__
+#define __VIDEO_PARSER_H__
-#include <cuda_runtime_api.h>
-#include <cufft.h>
-#include <cublas.h>
-#include "NCV.hpp"
+#include "opencv2/core/gpu_private.hpp"
+#include "opencv2/gpucodec.hpp"
+#include "frame_queue.h"
+#include "video_decoder.h"
-#if defined(__GNUC__)
- #define ncvSafeCall(expr) ___ncvSafeCall(expr, __FILE__, __LINE__, __func__)
- #define cufftSafeCall(expr) ___cufftSafeCall(expr, __FILE__, __LINE__, __func__)
- #define cublasSafeCall(expr) ___cublasSafeCall(expr, __FILE__, __LINE__, __func__)
-#else /* defined(__CUDACC__) || defined(__MSVC__) */
- #define ncvSafeCall(expr) ___ncvSafeCall(expr, __FILE__, __LINE__)
- #define cufftSafeCall(expr) ___cufftSafeCall(expr, __FILE__, __LINE__)
- #define cublasSafeCall(expr) ___cublasSafeCall(expr, __FILE__, __LINE__)
-#endif
+#include <nvcuvid.h>
-namespace cv { namespace gpu
+namespace cv { namespace gpu { namespace detail
{
- void ncvError(int err, const char *file, const int line, const char *func = "");
- void cufftError(int err, const char *file, const int line, const char *func = "");
- void cublasError(int err, const char *file, const int line, const char *func = "");
-}}
-static inline void ___ncvSafeCall(int err, const char *file, const int line, const char *func = "")
+class VideoParser
{
- if (NCV_SUCCESS != err)
- cv::gpu::ncvError(err, file, line, func);
-}
+public:
+ VideoParser(VideoDecoder* videoDecoder, FrameQueue* frameQueue);
-static inline void ___cufftSafeCall(cufftResult_t err, const char *file, const int line, const char *func = "")
-{
- if (CUFFT_SUCCESS != err)
- cv::gpu::cufftError(err, file, line, func);
-}
+ ~VideoParser()
+ {
+ cuvidDestroyVideoParser(parser_);
+ }
-static inline void ___cublasSafeCall(cublasStatus_t err, const char *file, const int line, const char *func = "")
-{
- if (CUBLAS_STATUS_SUCCESS != err)
- cv::gpu::cublasError(err, file, line, func);
-}
+ bool parseVideoData(const unsigned char* data, size_t size, bool endOfStream);
+
+ bool hasError() const { return hasError_; }
+
+private:
+ VideoDecoder* videoDecoder_;
+ FrameQueue* frameQueue_;
+ CUvideoparser parser_;
+ int unparsedPackets_;
+ volatile bool hasError_;
+
+ // Called when the decoder encounters a video format change (or initial sequence header)
+ // This particular implementation of the callback returns 0 in case the video format changes
+ // to something different than the original format. Returning 0 causes a stop of the app.
+ static int CUDAAPI HandleVideoSequence(void* pUserData, CUVIDEOFORMAT* pFormat);
+
+ // Called by the video parser to decode a single picture
+ // Since the parser will deliver data as fast as it can, we need to make sure that the picture
+ // index we're attempting to use for decode is no longer used for display
+ static int CUDAAPI HandlePictureDecode(void* pUserData, CUVIDPICPARAMS* pPicParams);
+
+ // Called by the video parser to display a video frame (in the case of field pictures, there may be
+ // 2 decode calls per 1 display call, since two fields make up one frame)
+ static int CUDAAPI HandlePictureDisplay(void* pUserData, CUVIDPARSERDISPINFO* pPicParams);
+};
+
+}}}
-#endif /* __OPENCV_CUDA_SAFE_CALL_HPP__ */
+#endif // __VIDEO_PARSER_H__
#include "precomp.hpp"
-#if !defined(HAVE_CUDA) || defined(CUDA_DISABLER) || !defined(HAVE_NVCUVID)
+#ifndef HAVE_NVCUVID
class cv::gpu::VideoReader_GPU::Impl
{
bool cv::gpu::VideoReader_GPU::VideoSource::parseVideoData(const unsigned char*, size_t, bool) { throw_no_cuda(); return false; }
void cv::gpu::VideoReader_GPU::dumpFormat(std::ostream&) { throw_no_cuda(); }
-#else // HAVE_CUDA
-
-#include "frame_queue.h"
-#include "video_decoder.h"
-#include "video_parser.h"
-
-#include "cuvid_video_source.h"
-#include "ffmpeg_video_source.h"
+#else // HAVE_NVCUVID
class cv::gpu::VideoReader_GPU::Impl
{
cv::gpu::VideoReader_GPU::FormatInfo format() const { return videoSource_->format(); }
private:
- Impl(const Impl&);
- Impl& operator =(const Impl&);
-
cv::Ptr<cv::gpu::VideoReader_GPU::VideoSource> videoSource_;
- std::auto_ptr<cv::gpu::detail::FrameQueue> frameQueue_;
- std::auto_ptr<cv::gpu::detail::VideoDecoder> videoDecoder_;
- std::auto_ptr<cv::gpu::detail::VideoParser> videoParser_;
+ cv::Ptr<cv::gpu::detail::FrameQueue> frameQueue_;
+ cv::Ptr<cv::gpu::detail::VideoDecoder> videoDecoder_;
+ cv::Ptr<cv::gpu::detail::VideoParser> videoParser_;
CUvideoctxlock lock_;
cuSafeCall( cuCtxGetCurrent(&ctx) );
cuSafeCall( cuvidCtxLockCreate(&lock_, ctx) );
- frameQueue_.reset(new detail::FrameQueue);
- videoDecoder_.reset(new detail::VideoDecoder(videoSource_->format(), lock_));
- videoParser_.reset(new detail::VideoParser(videoDecoder_.get(), frameQueue_.get()));
+ frameQueue_ = new detail::FrameQueue;
+ videoDecoder_ = new detail::VideoDecoder(videoSource_->format(), lock_);
+ videoParser_ = new detail::VideoParser(videoDecoder_, frameQueue_);
- videoSource_->setFrameQueue(frameQueue_.get());
- videoSource_->setVideoParser(videoParser_.get());
+ videoSource_->setFrameQueue(frameQueue_);
+ videoSource_->setVideoParser(videoParser_);
videoSource_->start();
}
videoSource_->stop();
}
-namespace cv { namespace gpu { namespace cudev {
- namespace video_decoding
- {
- void loadHueCSC(float hueCSC[9]);
- void NV12ToARGB_gpu(const PtrStepb decodedFrame, PtrStepSz<unsigned int> interopFrame, cudaStream_t stream = 0);
- }
+namespace cv { namespace gpu { namespace cudev
+{
+ void loadHueCSC(float hueCSC[9]);
+ void NV12_to_RGB(const PtrStepb decodedFrame, PtrStepSz<uint> interopFrame, cudaStream_t stream = 0);
}}}
namespace
void cudaPostProcessFrame(const cv::gpu::GpuMat& decodedFrame, cv::gpu::GpuMat& interopFrame, int width, int height)
{
- using namespace cv::gpu::cudev::video_decoding;
+ using namespace cv::gpu::cudev;
static bool updateCSC = true;
static float hueColorSpaceMat[9];
loadHueCSC(hueColorSpaceMat);
- NV12ToARGB_gpu(decodedFrame, interopFrame);
+ NV12_to_RGB(decodedFrame, interopFrame);
}
}
{
CV_Assert( !source.empty() );
close();
- impl_.reset(new Impl(source));
+ impl_ = new Impl(source);
}
bool cv::gpu::VideoReader_GPU::isOpened() const
{
- return impl_.get() != 0;
+ return !impl_.empty();
}
void cv::gpu::VideoReader_GPU::close()
{
- impl_.reset();
+ impl_.release();
}
bool cv::gpu::VideoReader_GPU::read(GpuMat& image)
st << "Chroma Format : " << chromas[_format.chromaFormat] << std::endl;
}
-#endif // HAVE_CUDA
+#endif // HAVE_NVCUVID
+
+template <> void cv::Ptr<cv::gpu::VideoReader_GPU::Impl>::delete_obj()
+{
+ if (obj) delete obj;
+}
#include "precomp.hpp"
-#if !defined(HAVE_CUDA) || defined(CUDA_DISABLER) || !defined(HAVE_NVCUVID) || !defined(WIN32)
+#if !defined(HAVE_NVCUVID) || !defined(WIN32)
class cv::gpu::VideoWriter_GPU::Impl
{
#else // !defined HAVE_CUDA || !defined WIN32
-#ifdef HAVE_FFMPEG
- #include "../src/cap_ffmpeg_impl.hpp"
-#else
- #include "../src/cap_ffmpeg_api.hpp"
-#endif
-
-
///////////////////////////////////////////////////////////////////////////
// VideoWriter_GPU::Impl
err = NVGetHWEncodeCaps();
if (err)
- CV_Error(CV_GpuNotSupported, "No CUDA capability present");
+ CV_Error(cv::Error::GpuNotSupported, "No CUDA capability present");
// Create the Encoder API Interface
err = NVCreateEncoder(&encoder_);
};
err = NVSetCodec(encoder_, codecs_id[codec_]);
if (err)
- CV_Error(CV_StsNotImplemented, "Codec format is not supported");
+ CV_Error(cv::Error::StsNotImplemented, "Codec format is not supported");
// Set default params
CV_Assert( err == 0 );
}
-namespace cv { namespace gpu { namespace cudev
-{
- namespace video_encoding
- {
- void YV12_gpu(const PtrStepSzb src, int cn, PtrStepSzb dst);
- }
-}}}
-
namespace
{
// UYVY/YUY2 are both 4:2:2 formats (16bpc)
}
}
+namespace cv { namespace gpu { namespace cudev
+{
+ void RGB_to_YV12(const PtrStepSzb src, int cn, PtrStepSzb dst, cudaStream_t stream = 0);
+}}}
+
void cv::gpu::VideoWriter_GPU::Impl::write(const cv::gpu::GpuMat& frame, bool lastFrame)
{
if (inputFormat_ == SF_BGR)
CV_Assert( res == CUDA_SUCCESS );
if (inputFormat_ == SF_BGR)
- cv::gpu::cudev::video_encoding::YV12_gpu(frame, frame.channels(), videoFrame_);
+ cv::gpu::cudev::RGB_to_YV12(frame, frame.channels(), videoFrame_);
else
{
switch (surfaceFormat_)
void EncoderCallBackFFMPEG::onBeginFrame(int frameNumber, PicType picType)
{
+ (void) frameNumber;
isKeyFrame_ = picType == IFRAME;
}
void EncoderCallBackFFMPEG::onEndFrame(int frameNumber, PicType picType)
{
+ (void) frameNumber;
+ (void) picType;
}
///////////////////////////////////////////////////////////////////////////
void cv::gpu::VideoWriter_GPU::open(const cv::Ptr<EncoderCallBack>& encoderCallback, cv::Size frameSize, double fps, SurfaceFormat format)
{
close();
- impl_.reset(new Impl(encoderCallback, frameSize, fps, format));
+ impl_ = new Impl(encoderCallback, frameSize, fps, format);
}
void cv::gpu::VideoWriter_GPU::open(const cv::Ptr<EncoderCallBack>& encoderCallback, cv::Size frameSize, double fps, const EncoderParams& params, SurfaceFormat format)
{
close();
- impl_.reset(new Impl(encoderCallback, frameSize, fps, params, format));
+ impl_ = new Impl(encoderCallback, frameSize, fps, params, format);
}
bool cv::gpu::VideoWriter_GPU::isOpened() const
{
- return impl_.get() != 0;
+ return !impl_.empty();
}
void cv::gpu::VideoWriter_GPU::close()
{
- impl_.reset();
+ impl_.release();
}
void cv::gpu::VideoWriter_GPU::write(const cv::gpu::GpuMat& image, bool lastFrame)
}
#endif // !defined HAVE_CUDA || !defined WIN32
+
+template <> void cv::Ptr<cv::gpu::VideoWriter_GPU::Impl>::delete_obj()
+{
+ if (obj) delete obj;
+}
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+CV_GPU_TEST_MAIN("gpu")
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifdef __GNUC__
+# pragma GCC diagnostic ignored "-Wmissing-declarations"
+# if defined __clang__ || defined __APPLE__
+# pragma GCC diagnostic ignored "-Wmissing-prototypes"
+# pragma GCC diagnostic ignored "-Wextra"
+# endif
+#endif
+
+#ifndef __OPENCV_TEST_PRECOMP_HPP__
+#define __OPENCV_TEST_PRECOMP_HPP__
+
+#include "opencv2/ts.hpp"
+#include "opencv2/ts/gpu_test.hpp"
+
+#include "opencv2/gpucodec.hpp"
+#include "opencv2/highgui.hpp"
+
+#endif
#include "test_precomp.hpp"
-#if defined(HAVE_CUDA) && defined(HAVE_NVCUVID)
+#ifdef HAVE_NVCUVID
+
+PARAM_TEST_CASE(Video, cv::gpu::DeviceInfo, std::string)
+{
+};
//////////////////////////////////////////////////////
// VideoReader
-PARAM_TEST_CASE(VideoReader, cv::gpu::DeviceInfo, std::string)
+GPU_TEST_P(Video, Reader)
{
- cv::gpu::DeviceInfo devInfo;
- std::string inputFile;
+ cv::gpu::setDevice(GET_PARAM(0).deviceID());
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- inputFile = GET_PARAM(1);
-
- cv::gpu::setDevice(devInfo.deviceID());
+ const std::string inputFile = std::string(cvtest::TS::ptr()->get_data_path()) + "video/" + GET_PARAM(1);
- inputFile = std::string(cvtest::TS::ptr()->get_data_path()) + "video/" + inputFile;
- }
-};
-
-GPU_TEST_P(VideoReader, Regression)
-{
cv::gpu::VideoReader_GPU reader(inputFile);
ASSERT_TRUE(reader.isOpened());
ASSERT_FALSE(reader.isOpened());
}
-INSTANTIATE_TEST_CASE_P(GPU_Video, VideoReader, testing::Combine(
- ALL_DEVICES,
- testing::Values(std::string("768x576.avi"), std::string("1920x1080.avi"))));
-
//////////////////////////////////////////////////////
// VideoWriter
#ifdef WIN32
-PARAM_TEST_CASE(VideoWriter, cv::gpu::DeviceInfo, std::string)
+GPU_TEST_P(Video, Writer)
{
- cv::gpu::DeviceInfo devInfo;
- std::string inputFile;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- inputFile = GET_PARAM(1);
+ cv::gpu::setDevice(GET_PARAM(0).deviceID());
- cv::gpu::setDevice(devInfo.deviceID());
-
- inputFile = std::string(cvtest::TS::ptr()->get_data_path()) + std::string("video/") + inputFile;
- }
-};
+ const std::string inputFile = std::string(cvtest::TS::ptr()->get_data_path()) + "video/" + GET_PARAM(1);
-GPU_TEST_P(VideoWriter, Regression)
-{
std::string outputFile = cv::tempfile(".avi");
const double FPS = 25.0;
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Video, VideoWriter, testing::Combine(
+#endif // WIN32
+
+INSTANTIATE_TEST_CASE_P(GPU_Codec, Video, testing::Combine(
ALL_DEVICES,
testing::Values(std::string("768x576.avi"), std::string("1920x1080.avi"))));
-#endif // WIN32
-
-#endif // defined(HAVE_CUDA) && defined(HAVE_NVCUVID)
+#endif // HAVE_NVCUVID
--- /dev/null
+if(ANDROID OR IOS)
+ ocv_module_disable(gpufeatures2d)
+endif()
+
+set(the_description "GPU-accelerated Feature Detection and Description")
+
+ocv_warnings_disable(CMAKE_CXX_FLAGS /wd4127 /wd4100 /wd4324 /wd4512 /wd4515 -Wundef -Wmissing-declarations -Wshadow -Wunused-parameter)
+
+ocv_define_module(gpufeatures2d opencv_features2d opencv_gpufilters opencv_gpuwarping)
--- /dev/null
+****************************************************************
+gpufeatures2d. GPU-accelerated Feature Detection and Description
+****************************************************************
+
+.. toctree::
+ :maxdepth: 1
+
+ feature_detection_and_description
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_GPUFEATURES2D_HPP__
+#define __OPENCV_GPUFEATURES2D_HPP__
+
+#ifndef __cplusplus
+# error gpufeatures2d.hpp header must be compiled as C++
+#endif
+
+#include "opencv2/core/gpumat.hpp"
+#include "opencv2/gpufilters.hpp"
+
+namespace cv { namespace gpu {
+
+class CV_EXPORTS BFMatcher_GPU
+{
+public:
+ explicit BFMatcher_GPU(int norm = cv::NORM_L2);
+
+ // Add descriptors to train descriptor collection
+ void add(const std::vector<GpuMat>& descCollection);
+
+ // Get train descriptors collection
+ const std::vector<GpuMat>& getTrainDescriptors() const;
+
+ // Clear train descriptors collection
+ void clear();
+
+ // Return true if there are not train descriptors in collection
+ bool empty() const;
+
+ // Return true if the matcher supports mask in match methods
+ bool isMaskSupported() const;
+
+ // Find one best match for each query descriptor
+ void matchSingle(const GpuMat& query, const GpuMat& train,
+ GpuMat& trainIdx, GpuMat& distance,
+ const GpuMat& mask = GpuMat(), Stream& stream = Stream::Null());
+
+ // Download trainIdx and distance and convert it to CPU vector with DMatch
+ static void matchDownload(const GpuMat& trainIdx, const GpuMat& distance, std::vector<DMatch>& matches);
+ // Convert trainIdx and distance to vector with DMatch
+ static void matchConvert(const Mat& trainIdx, const Mat& distance, std::vector<DMatch>& matches);
+
+ // Find one best match for each query descriptor
+ void match(const GpuMat& query, const GpuMat& train, std::vector<DMatch>& matches, const GpuMat& mask = GpuMat());
+
+ // Make gpu collection of trains and masks in suitable format for matchCollection function
+ void makeGpuCollection(GpuMat& trainCollection, GpuMat& maskCollection, const std::vector<GpuMat>& masks = std::vector<GpuMat>());
+
+ // Find one best match from train collection for each query descriptor
+ void matchCollection(const GpuMat& query, const GpuMat& trainCollection,
+ GpuMat& trainIdx, GpuMat& imgIdx, GpuMat& distance,
+ const GpuMat& masks = GpuMat(), Stream& stream = Stream::Null());
+
+ // Download trainIdx, imgIdx and distance and convert it to vector with DMatch
+ static void matchDownload(const GpuMat& trainIdx, const GpuMat& imgIdx, const GpuMat& distance, std::vector<DMatch>& matches);
+ // Convert trainIdx, imgIdx and distance to vector with DMatch
+ static void matchConvert(const Mat& trainIdx, const Mat& imgIdx, const Mat& distance, std::vector<DMatch>& matches);
+
+ // Find one best match from train collection for each query descriptor.
+ void match(const GpuMat& query, std::vector<DMatch>& matches, const std::vector<GpuMat>& masks = std::vector<GpuMat>());
+
+ // Find k best matches for each query descriptor (in increasing order of distances)
+ void knnMatchSingle(const GpuMat& query, const GpuMat& train,
+ GpuMat& trainIdx, GpuMat& distance, GpuMat& allDist, int k,
+ const GpuMat& mask = GpuMat(), Stream& stream = Stream::Null());
+
+ // Download trainIdx and distance and convert it to vector with DMatch
+ // compactResult is used when mask is not empty. If compactResult is false matches
+ // vector will have the same size as queryDescriptors rows. If compactResult is true
+ // matches vector will not contain matches for fully masked out query descriptors.
+ static void knnMatchDownload(const GpuMat& trainIdx, const GpuMat& distance,
+ std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
+ // Convert trainIdx and distance to vector with DMatch
+ static void knnMatchConvert(const Mat& trainIdx, const Mat& distance,
+ std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
+
+ // Find k best matches for each query descriptor (in increasing order of distances).
+ // compactResult is used when mask is not empty. If compactResult is false matches
+ // vector will have the same size as queryDescriptors rows. If compactResult is true
+ // matches vector will not contain matches for fully masked out query descriptors.
+ void knnMatch(const GpuMat& query, const GpuMat& train,
+ std::vector< std::vector<DMatch> >& matches, int k, const GpuMat& mask = GpuMat(),
+ bool compactResult = false);
+
+ // Find k best matches from train collection for each query descriptor (in increasing order of distances)
+ void knnMatch2Collection(const GpuMat& query, const GpuMat& trainCollection,
+ GpuMat& trainIdx, GpuMat& imgIdx, GpuMat& distance,
+ const GpuMat& maskCollection = GpuMat(), Stream& stream = Stream::Null());
+
+ // Download trainIdx and distance and convert it to vector with DMatch
+ // compactResult is used when mask is not empty. If compactResult is false matches
+ // vector will have the same size as queryDescriptors rows. If compactResult is true
+ // matches vector will not contain matches for fully masked out query descriptors.
+ static void knnMatch2Download(const GpuMat& trainIdx, const GpuMat& imgIdx, const GpuMat& distance,
+ std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
+ // Convert trainIdx and distance to vector with DMatch
+ static void knnMatch2Convert(const Mat& trainIdx, const Mat& imgIdx, const Mat& distance,
+ std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
+
+ // Find k best matches for each query descriptor (in increasing order of distances).
+ // compactResult is used when mask is not empty. If compactResult is false matches
+ // vector will have the same size as queryDescriptors rows. If compactResult is true
+ // matches vector will not contain matches for fully masked out query descriptors.
+ void knnMatch(const GpuMat& query, std::vector< std::vector<DMatch> >& matches, int k,
+ const std::vector<GpuMat>& masks = std::vector<GpuMat>(), bool compactResult = false);
+
+ // Find best matches for each query descriptor which have distance less than maxDistance.
+ // nMatches.at<int>(0, queryIdx) will contain matches count for queryIdx.
+ // carefully nMatches can be greater than trainIdx.cols - it means that matcher didn't find all matches,
+ // because it didn't have enough memory.
+ // If trainIdx is empty, then trainIdx and distance will be created with size nQuery x max((nTrain / 100), 10),
+ // otherwize user can pass own allocated trainIdx and distance with size nQuery x nMaxMatches
+ // Matches doesn't sorted.
+ void radiusMatchSingle(const GpuMat& query, const GpuMat& train,
+ GpuMat& trainIdx, GpuMat& distance, GpuMat& nMatches, float maxDistance,
+ const GpuMat& mask = GpuMat(), Stream& stream = Stream::Null());
+
+ // Download trainIdx, nMatches and distance and convert it to vector with DMatch.
+ // matches will be sorted in increasing order of distances.
+ // compactResult is used when mask is not empty. If compactResult is false matches
+ // vector will have the same size as queryDescriptors rows. If compactResult is true
+ // matches vector will not contain matches for fully masked out query descriptors.
+ static void radiusMatchDownload(const GpuMat& trainIdx, const GpuMat& distance, const GpuMat& nMatches,
+ std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
+ // Convert trainIdx, nMatches and distance to vector with DMatch.
+ static void radiusMatchConvert(const Mat& trainIdx, const Mat& distance, const Mat& nMatches,
+ std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
+
+ // Find best matches for each query descriptor which have distance less than maxDistance
+ // in increasing order of distances).
+ void radiusMatch(const GpuMat& query, const GpuMat& train,
+ std::vector< std::vector<DMatch> >& matches, float maxDistance,
+ const GpuMat& mask = GpuMat(), bool compactResult = false);
+
+ // Find best matches for each query descriptor which have distance less than maxDistance.
+ // If trainIdx is empty, then trainIdx and distance will be created with size nQuery x max((nQuery / 100), 10),
+ // otherwize user can pass own allocated trainIdx and distance with size nQuery x nMaxMatches
+ // Matches doesn't sorted.
+ void radiusMatchCollection(const GpuMat& query, GpuMat& trainIdx, GpuMat& imgIdx, GpuMat& distance, GpuMat& nMatches, float maxDistance,
+ const std::vector<GpuMat>& masks = std::vector<GpuMat>(), Stream& stream = Stream::Null());
+
+ // Download trainIdx, imgIdx, nMatches and distance and convert it to vector with DMatch.
+ // matches will be sorted in increasing order of distances.
+ // compactResult is used when mask is not empty. If compactResult is false matches
+ // vector will have the same size as queryDescriptors rows. If compactResult is true
+ // matches vector will not contain matches for fully masked out query descriptors.
+ static void radiusMatchDownload(const GpuMat& trainIdx, const GpuMat& imgIdx, const GpuMat& distance, const GpuMat& nMatches,
+ std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
+ // Convert trainIdx, nMatches and distance to vector with DMatch.
+ static void radiusMatchConvert(const Mat& trainIdx, const Mat& imgIdx, const Mat& distance, const Mat& nMatches,
+ std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
+
+ // Find best matches from train collection for each query descriptor which have distance less than
+ // maxDistance (in increasing order of distances).
+ void radiusMatch(const GpuMat& query, std::vector< std::vector<DMatch> >& matches, float maxDistance,
+ const std::vector<GpuMat>& masks = std::vector<GpuMat>(), bool compactResult = false);
+
+ int norm;
+
+private:
+ std::vector<GpuMat> trainDescCollection;
+};
+
+class CV_EXPORTS FAST_GPU
+{
+public:
+ enum
+ {
+ LOCATION_ROW = 0,
+ RESPONSE_ROW,
+ ROWS_COUNT
+ };
+
+ // all features have same size
+ static const int FEATURE_SIZE = 7;
+
+ explicit FAST_GPU(int threshold, bool nonmaxSupression = true, double keypointsRatio = 0.05);
+
+ //! finds the keypoints using FAST detector
+ //! supports only CV_8UC1 images
+ void operator ()(const GpuMat& image, const GpuMat& mask, GpuMat& keypoints);
+ void operator ()(const GpuMat& image, const GpuMat& mask, std::vector<KeyPoint>& keypoints);
+
+ //! download keypoints from device to host memory
+ static void downloadKeypoints(const GpuMat& d_keypoints, std::vector<KeyPoint>& keypoints);
+
+ //! convert keypoints to KeyPoint vector
+ static void convertKeypoints(const Mat& h_keypoints, std::vector<KeyPoint>& keypoints);
+
+ //! release temporary buffer's memory
+ void release();
+
+ bool nonmaxSupression;
+
+ int threshold;
+
+ //! max keypoints = keypointsRatio * img.size().area()
+ double keypointsRatio;
+
+ //! find keypoints and compute it's response if nonmaxSupression is true
+ //! return count of detected keypoints
+ int calcKeyPointsLocation(const GpuMat& image, const GpuMat& mask);
+
+ //! get final array of keypoints
+ //! performs nonmax supression if needed
+ //! return final count of keypoints
+ int getKeyPoints(GpuMat& keypoints);
+
+private:
+ GpuMat kpLoc_;
+ int count_;
+
+ GpuMat score_;
+
+ GpuMat d_keypoints_;
+};
+
+class CV_EXPORTS ORB_GPU
+{
+public:
+ enum
+ {
+ X_ROW = 0,
+ Y_ROW,
+ RESPONSE_ROW,
+ ANGLE_ROW,
+ OCTAVE_ROW,
+ SIZE_ROW,
+ ROWS_COUNT
+ };
+
+ enum
+ {
+ DEFAULT_FAST_THRESHOLD = 20
+ };
+
+ //! Constructor
+ explicit ORB_GPU(int nFeatures = 500, float scaleFactor = 1.2f, int nLevels = 8, int edgeThreshold = 31,
+ int firstLevel = 0, int WTA_K = 2, int scoreType = 0, int patchSize = 31);
+
+ //! Compute the ORB features on an image
+ //! image - the image to compute the features (supports only CV_8UC1 images)
+ //! mask - the mask to apply
+ //! keypoints - the resulting keypoints
+ void operator()(const GpuMat& image, const GpuMat& mask, std::vector<KeyPoint>& keypoints);
+ void operator()(const GpuMat& image, const GpuMat& mask, GpuMat& keypoints);
+
+ //! Compute the ORB features and descriptors on an image
+ //! image - the image to compute the features (supports only CV_8UC1 images)
+ //! mask - the mask to apply
+ //! keypoints - the resulting keypoints
+ //! descriptors - descriptors array
+ void operator()(const GpuMat& image, const GpuMat& mask, std::vector<KeyPoint>& keypoints, GpuMat& descriptors);
+ void operator()(const GpuMat& image, const GpuMat& mask, GpuMat& keypoints, GpuMat& descriptors);
+
+ //! download keypoints from device to host memory
+ static void downloadKeyPoints(const GpuMat& d_keypoints, std::vector<KeyPoint>& keypoints);
+ //! convert keypoints to KeyPoint vector
+ static void convertKeyPoints(const Mat& d_keypoints, std::vector<KeyPoint>& keypoints);
+
+ //! returns the descriptor size in bytes
+ inline int descriptorSize() const { return kBytes; }
+
+ inline void setFastParams(int threshold, bool nonmaxSupression = true)
+ {
+ fastDetector_.threshold = threshold;
+ fastDetector_.nonmaxSupression = nonmaxSupression;
+ }
+
+ //! release temporary buffer's memory
+ void release();
+
+ //! if true, image will be blurred before descriptors calculation
+ bool blurForDescriptor;
+
+private:
+ enum { kBytes = 32 };
+
+ void buildScalePyramids(const GpuMat& image, const GpuMat& mask);
+
+ void computeKeyPointsPyramid();
+
+ void computeDescriptors(GpuMat& descriptors);
+
+ void mergeKeyPoints(GpuMat& keypoints);
+
+ int nFeatures_;
+ float scaleFactor_;
+ int nLevels_;
+ int edgeThreshold_;
+ int firstLevel_;
+ int WTA_K_;
+ int scoreType_;
+ int patchSize_;
+
+ // The number of desired features per scale
+ std::vector<size_t> n_features_per_level_;
+
+ // Points to compute BRIEF descriptors from
+ GpuMat pattern_;
+
+ std::vector<GpuMat> imagePyr_;
+ std::vector<GpuMat> maskPyr_;
+
+ GpuMat buf_;
+
+ std::vector<GpuMat> keyPointsPyr_;
+ std::vector<int> keyPointsCount_;
+
+ FAST_GPU fastDetector_;
+
+ Ptr<FilterEngine_GPU> blurFilter;
+
+ GpuMat d_keypoints_;
+};
+
+}} // namespace cv { namespace gpu {
+
+#endif /* __OPENCV_GPUFEATURES2D_HPP__ */
DEF_PARAM_TEST(Image_Threshold_NonMaxSupression, string, int, bool);
-PERF_TEST_P(Image_Threshold_NonMaxSupression, Features2D_FAST,
+PERF_TEST_P(Image_Threshold_NonMaxSupression, FAST,
Combine(Values<string>("gpu/perf/aloe.png"),
Values(20),
Bool()))
DEF_PARAM_TEST(Image_NFeatures, string, int);
-PERF_TEST_P(Image_NFeatures, Features2D_ORB,
+PERF_TEST_P(Image_NFeatures, ORB,
Combine(Values<string>("gpu/perf/aloe.png"),
Values(4000)))
{
DEF_PARAM_TEST(DescSize_Norm, int, NormType);
-PERF_TEST_P(DescSize_Norm, Features2D_BFMatch,
+PERF_TEST_P(DescSize_Norm, BFMatch,
Combine(Values(64, 128, 256),
Values(NormType(cv::NORM_L1), NormType(cv::NORM_L2), NormType(cv::NORM_HAMMING))))
{
DEF_PARAM_TEST(DescSize_K_Norm, int, int, NormType);
-PERF_TEST_P(DescSize_K_Norm, Features2D_BFKnnMatch,
+PERF_TEST_P(DescSize_K_Norm, BFKnnMatch,
Combine(Values(64, 128, 256),
Values(2, 3),
Values(NormType(cv::NORM_L1), NormType(cv::NORM_L2))))
//////////////////////////////////////////////////////////////////////
// BFRadiusMatch
-PERF_TEST_P(DescSize_Norm, Features2D_BFRadiusMatch,
+PERF_TEST_P(DescSize_Norm, BFRadiusMatch,
Combine(Values(64, 128, 256),
Values(NormType(cv::NORM_L1), NormType(cv::NORM_L2))))
{
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace perf;
+
+CV_PERF_TEST_MAIN(gpufeatures2d, printCudaInfo())
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifdef __GNUC__
+# pragma GCC diagnostic ignored "-Wmissing-declarations"
+# if defined __clang__ || defined __APPLE__
+# pragma GCC diagnostic ignored "-Wmissing-prototypes"
+# pragma GCC diagnostic ignored "-Wextra"
+# endif
+#endif
+
+#ifndef __OPENCV_PERF_PRECOMP_HPP__
+#define __OPENCV_PERF_PRECOMP_HPP__
+
+#include "opencv2/ts.hpp"
+#include "opencv2/ts/gpu_perf.hpp"
+
+#include "opencv2/gpufeatures2d.hpp"
+#include "opencv2/features2d.hpp"
+
+#ifdef GTEST_CREATE_SHARED_LIBRARY
+#error no modules except ts should have GTEST_CREATE_SHARED_LIBRARY defined
+#endif
+
+#endif
{
if (level < firstLevel_)
{
- resize(image, imagePyr_[level], sz, 0, 0, INTER_LINEAR);
+ gpu::resize(image, imagePyr_[level], sz, 0, 0, INTER_LINEAR);
if (!mask.empty())
- resize(mask, maskPyr_[level], sz, 0, 0, INTER_LINEAR);
+ gpu::resize(mask, maskPyr_[level], sz, 0, 0, INTER_LINEAR);
}
else
{
- resize(imagePyr_[level - 1], imagePyr_[level], sz, 0, 0, INTER_LINEAR);
+ gpu::resize(imagePyr_[level - 1], imagePyr_[level], sz, 0, 0, INTER_LINEAR);
if (!mask.empty())
{
- resize(maskPyr_[level - 1], maskPyr_[level], sz, 0, 0, INTER_LINEAR);
- threshold(maskPyr_[level], maskPyr_[level], 254, 0, THRESH_TOZERO);
+ gpu::resize(maskPyr_[level - 1], maskPyr_[level], sz, 0, 0, INTER_LINEAR);
+ gpu::threshold(maskPyr_[level], maskPyr_[level], 254, 0, THRESH_TOZERO);
}
}
}
Rect inner(edgeThreshold_, edgeThreshold_, sz.width - 2 * edgeThreshold_, sz.height - 2 * edgeThreshold_);
buf_(inner).setTo(Scalar::all(255));
- bitwise_and(maskPyr_[level], buf_, maskPyr_[level]);
+ gpu::bitwise_and(maskPyr_[level], buf_, maskPyr_[level]);
}
}
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_PRECOMP_H__
+#define __OPENCV_PRECOMP_H__
+
+#include <algorithm>
+#include <functional>
+#include <iterator>
+
+#include "opencv2/gpufeatures2d.hpp"
+#include "opencv2/gpuarithm.hpp"
+#include "opencv2/gpuwarping.hpp"
+#include "opencv2/features2d.hpp"
+
+#include "opencv2/core/gpu_private.hpp"
+
+#endif /* __OPENCV_PRECOMP_H__ */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+CV_GPU_TEST_MAIN("gpu")
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifdef __GNUC__
+# pragma GCC diagnostic ignored "-Wmissing-declarations"
+# if defined __clang__ || defined __APPLE__
+# pragma GCC diagnostic ignored "-Wmissing-prototypes"
+# pragma GCC diagnostic ignored "-Wextra"
+# endif
+#endif
+
+#ifndef __OPENCV_TEST_PRECOMP_HPP__
+#define __OPENCV_TEST_PRECOMP_HPP__
+
+#include "opencv2/ts.hpp"
+#include "opencv2/ts/gpu_test.hpp"
+
+#include "opencv2/gpufeatures2d.hpp"
+#include "opencv2/features2d.hpp"
+
+#endif
--- /dev/null
+if(ANDROID OR IOS)
+ ocv_module_disable(gpufilters)
+endif()
+
+set(the_description "GPU-accelerated Image Filtering")
+
+ocv_warnings_disable(CMAKE_CXX_FLAGS /wd4127 /wd4324 /wd4512 -Wundef -Wmissing-declarations)
+
+ocv_define_module(gpufilters opencv_imgproc OPTIONAL opencv_gpuarithm)
--- /dev/null
+*******************************************
+gpufilters. GPU-accelerated Image Filtering
+*******************************************
+
+.. toctree::
+ :maxdepth: 1
+
+ filtering
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_GPUFILTERS_HPP__
+#define __OPENCV_GPUFILTERS_HPP__
+
+#ifndef __cplusplus
+# error gpufilters.hpp header must be compiled as C++
+#endif
+
+#include "opencv2/core/gpumat.hpp"
+#include "opencv2/core/base.hpp"
+
+namespace cv { namespace gpu {
+
+/*!
+The Base Class for 1D or Row-wise Filters
+
+This is the base class for linear or non-linear filters that process 1D data.
+In particular, such filters are used for the "horizontal" filtering parts in separable filters.
+*/
+class CV_EXPORTS BaseRowFilter_GPU
+{
+public:
+ BaseRowFilter_GPU(int ksize_, int anchor_) : ksize(ksize_), anchor(anchor_) {}
+ virtual ~BaseRowFilter_GPU() {}
+ virtual void operator()(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null()) = 0;
+ int ksize, anchor;
+};
+
+/*!
+The Base Class for Column-wise Filters
+
+This is the base class for linear or non-linear filters that process columns of 2D arrays.
+Such filters are used for the "vertical" filtering parts in separable filters.
+*/
+class CV_EXPORTS BaseColumnFilter_GPU
+{
+public:
+ BaseColumnFilter_GPU(int ksize_, int anchor_) : ksize(ksize_), anchor(anchor_) {}
+ virtual ~BaseColumnFilter_GPU() {}
+ virtual void operator()(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null()) = 0;
+ int ksize, anchor;
+};
+
+/*!
+The Base Class for Non-Separable 2D Filters.
+
+This is the base class for linear or non-linear 2D filters.
+*/
+class CV_EXPORTS BaseFilter_GPU
+{
+public:
+ BaseFilter_GPU(const Size& ksize_, const Point& anchor_) : ksize(ksize_), anchor(anchor_) {}
+ virtual ~BaseFilter_GPU() {}
+ virtual void operator()(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null()) = 0;
+ Size ksize;
+ Point anchor;
+};
+
+/*!
+The Base Class for Filter Engine.
+
+The class can be used to apply an arbitrary filtering operation to an image.
+It contains all the necessary intermediate buffers.
+*/
+class CV_EXPORTS FilterEngine_GPU
+{
+public:
+ virtual ~FilterEngine_GPU() {}
+
+ virtual void apply(const GpuMat& src, GpuMat& dst, Rect roi = Rect(0,0,-1,-1), Stream& stream = Stream::Null()) = 0;
+};
+
+//! returns the non-separable filter engine with the specified filter
+CV_EXPORTS Ptr<FilterEngine_GPU> createFilter2D_GPU(const Ptr<BaseFilter_GPU>& filter2D, int srcType, int dstType);
+
+//! returns the separable filter engine with the specified filters
+CV_EXPORTS Ptr<FilterEngine_GPU> createSeparableFilter_GPU(const Ptr<BaseRowFilter_GPU>& rowFilter,
+ const Ptr<BaseColumnFilter_GPU>& columnFilter, int srcType, int bufType, int dstType);
+CV_EXPORTS Ptr<FilterEngine_GPU> createSeparableFilter_GPU(const Ptr<BaseRowFilter_GPU>& rowFilter,
+ const Ptr<BaseColumnFilter_GPU>& columnFilter, int srcType, int bufType, int dstType, GpuMat& buf);
+
+//! returns horizontal 1D box filter
+//! supports only CV_8UC1 source type and CV_32FC1 sum type
+CV_EXPORTS Ptr<BaseRowFilter_GPU> getRowSumFilter_GPU(int srcType, int sumType, int ksize, int anchor = -1);
+
+//! returns vertical 1D box filter
+//! supports only CV_8UC1 sum type and CV_32FC1 dst type
+CV_EXPORTS Ptr<BaseColumnFilter_GPU> getColumnSumFilter_GPU(int sumType, int dstType, int ksize, int anchor = -1);
+
+//! returns 2D box filter
+//! supports CV_8UC1 and CV_8UC4 source type, dst type must be the same as source type
+CV_EXPORTS Ptr<BaseFilter_GPU> getBoxFilter_GPU(int srcType, int dstType, const Size& ksize, Point anchor = Point(-1, -1));
+
+//! returns box filter engine
+CV_EXPORTS Ptr<FilterEngine_GPU> createBoxFilter_GPU(int srcType, int dstType, const Size& ksize,
+ const Point& anchor = Point(-1,-1));
+
+//! returns 2D morphological filter
+//! only MORPH_ERODE and MORPH_DILATE are supported
+//! supports CV_8UC1 and CV_8UC4 types
+//! kernel must have CV_8UC1 type, one rows and cols == ksize.width * ksize.height
+CV_EXPORTS Ptr<BaseFilter_GPU> getMorphologyFilter_GPU(int op, int type, const Mat& kernel, const Size& ksize,
+ Point anchor=Point(-1,-1));
+
+//! returns morphological filter engine. Only MORPH_ERODE and MORPH_DILATE are supported.
+CV_EXPORTS Ptr<FilterEngine_GPU> createMorphologyFilter_GPU(int op, int type, const Mat& kernel,
+ const Point& anchor = Point(-1,-1), int iterations = 1);
+CV_EXPORTS Ptr<FilterEngine_GPU> createMorphologyFilter_GPU(int op, int type, const Mat& kernel, GpuMat& buf,
+ const Point& anchor = Point(-1,-1), int iterations = 1);
+
+//! returns 2D filter with the specified kernel
+//! supports CV_8U, CV_16U and CV_32F one and four channel image
+CV_EXPORTS Ptr<BaseFilter_GPU> getLinearFilter_GPU(int srcType, int dstType, const Mat& kernel, Point anchor = Point(-1, -1), int borderType = BORDER_DEFAULT);
+
+//! returns the non-separable linear filter engine
+CV_EXPORTS Ptr<FilterEngine_GPU> createLinearFilter_GPU(int srcType, int dstType, const Mat& kernel,
+ Point anchor = Point(-1,-1), int borderType = BORDER_DEFAULT);
+
+//! returns the primitive row filter with the specified kernel.
+//! supports only CV_8UC1, CV_8UC4, CV_16SC1, CV_16SC2, CV_32SC1, CV_32FC1 source type.
+//! there are two version of algorithm: NPP and OpenCV.
+//! NPP calls when srcType == CV_8UC1 or srcType == CV_8UC4 and bufType == srcType,
+//! otherwise calls OpenCV version.
+//! NPP supports only BORDER_CONSTANT border type.
+//! OpenCV version supports only CV_32F as buffer depth and
+//! BORDER_REFLECT101, BORDER_REPLICATE and BORDER_CONSTANT border types.
+CV_EXPORTS Ptr<BaseRowFilter_GPU> getLinearRowFilter_GPU(int srcType, int bufType, const Mat& rowKernel,
+ int anchor = -1, int borderType = BORDER_DEFAULT);
+
+//! returns the primitive column filter with the specified kernel.
+//! supports only CV_8UC1, CV_8UC4, CV_16SC1, CV_16SC2, CV_32SC1, CV_32FC1 dst type.
+//! there are two version of algorithm: NPP and OpenCV.
+//! NPP calls when dstType == CV_8UC1 or dstType == CV_8UC4 and bufType == dstType,
+//! otherwise calls OpenCV version.
+//! NPP supports only BORDER_CONSTANT border type.
+//! OpenCV version supports only CV_32F as buffer depth and
+//! BORDER_REFLECT101, BORDER_REPLICATE and BORDER_CONSTANT border types.
+CV_EXPORTS Ptr<BaseColumnFilter_GPU> getLinearColumnFilter_GPU(int bufType, int dstType, const Mat& columnKernel,
+ int anchor = -1, int borderType = BORDER_DEFAULT);
+
+//! returns the separable linear filter engine
+CV_EXPORTS Ptr<FilterEngine_GPU> createSeparableLinearFilter_GPU(int srcType, int dstType, const Mat& rowKernel,
+ const Mat& columnKernel, const Point& anchor = Point(-1,-1), int rowBorderType = BORDER_DEFAULT,
+ int columnBorderType = -1);
+CV_EXPORTS Ptr<FilterEngine_GPU> createSeparableLinearFilter_GPU(int srcType, int dstType, const Mat& rowKernel,
+ const Mat& columnKernel, GpuMat& buf, const Point& anchor = Point(-1,-1), int rowBorderType = BORDER_DEFAULT,
+ int columnBorderType = -1);
+
+//! returns filter engine for the generalized Sobel operator
+CV_EXPORTS Ptr<FilterEngine_GPU> createDerivFilter_GPU(int srcType, int dstType, int dx, int dy, int ksize,
+ int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
+CV_EXPORTS Ptr<FilterEngine_GPU> createDerivFilter_GPU(int srcType, int dstType, int dx, int dy, int ksize, GpuMat& buf,
+ int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
+
+//! returns the Gaussian filter engine
+CV_EXPORTS Ptr<FilterEngine_GPU> createGaussianFilter_GPU(int type, Size ksize, double sigma1, double sigma2 = 0,
+ int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
+CV_EXPORTS Ptr<FilterEngine_GPU> createGaussianFilter_GPU(int type, Size ksize, GpuMat& buf, double sigma1, double sigma2 = 0,
+ int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
+
+//! returns maximum filter
+CV_EXPORTS Ptr<BaseFilter_GPU> getMaxFilter_GPU(int srcType, int dstType, const Size& ksize, Point anchor = Point(-1,-1));
+
+//! returns minimum filter
+CV_EXPORTS Ptr<BaseFilter_GPU> getMinFilter_GPU(int srcType, int dstType, const Size& ksize, Point anchor = Point(-1,-1));
+
+//! smooths the image using the normalized box filter
+//! supports CV_8UC1, CV_8UC4 types
+CV_EXPORTS void boxFilter(const GpuMat& src, GpuMat& dst, int ddepth, Size ksize, Point anchor = Point(-1,-1), Stream& stream = Stream::Null());
+
+//! a synonym for normalized box filter
+static inline void blur(const GpuMat& src, GpuMat& dst, Size ksize, Point anchor = Point(-1,-1), Stream& stream = Stream::Null())
+{
+ boxFilter(src, dst, -1, ksize, anchor, stream);
+}
+
+//! erodes the image (applies the local minimum operator)
+CV_EXPORTS void erode(const GpuMat& src, GpuMat& dst, const Mat& kernel, Point anchor = Point(-1, -1), int iterations = 1);
+CV_EXPORTS void erode(const GpuMat& src, GpuMat& dst, const Mat& kernel, GpuMat& buf,
+ Point anchor = Point(-1, -1), int iterations = 1,
+ Stream& stream = Stream::Null());
+
+//! dilates the image (applies the local maximum operator)
+CV_EXPORTS void dilate(const GpuMat& src, GpuMat& dst, const Mat& kernel, Point anchor = Point(-1, -1), int iterations = 1);
+CV_EXPORTS void dilate(const GpuMat& src, GpuMat& dst, const Mat& kernel, GpuMat& buf,
+ Point anchor = Point(-1, -1), int iterations = 1,
+ Stream& stream = Stream::Null());
+
+//! applies an advanced morphological operation to the image
+CV_EXPORTS void morphologyEx(const GpuMat& src, GpuMat& dst, int op, const Mat& kernel, Point anchor = Point(-1, -1), int iterations = 1);
+CV_EXPORTS void morphologyEx(const GpuMat& src, GpuMat& dst, int op, const Mat& kernel, GpuMat& buf1, GpuMat& buf2,
+ Point anchor = Point(-1, -1), int iterations = 1, Stream& stream = Stream::Null());
+
+//! applies non-separable 2D linear filter to the image
+CV_EXPORTS void filter2D(const GpuMat& src, GpuMat& dst, int ddepth, const Mat& kernel, Point anchor=Point(-1,-1), int borderType = BORDER_DEFAULT, Stream& stream = Stream::Null());
+
+//! applies separable 2D linear filter to the image
+CV_EXPORTS void sepFilter2D(const GpuMat& src, GpuMat& dst, int ddepth, const Mat& kernelX, const Mat& kernelY,
+ Point anchor = Point(-1,-1), int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
+CV_EXPORTS void sepFilter2D(const GpuMat& src, GpuMat& dst, int ddepth, const Mat& kernelX, const Mat& kernelY, GpuMat& buf,
+ Point anchor = Point(-1,-1), int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1,
+ Stream& stream = Stream::Null());
+
+//! applies generalized Sobel operator to the image
+CV_EXPORTS void Sobel(const GpuMat& src, GpuMat& dst, int ddepth, int dx, int dy, int ksize = 3, double scale = 1,
+ int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
+CV_EXPORTS void Sobel(const GpuMat& src, GpuMat& dst, int ddepth, int dx, int dy, GpuMat& buf, int ksize = 3, double scale = 1,
+ int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1, Stream& stream = Stream::Null());
+
+//! applies the vertical or horizontal Scharr operator to the image
+CV_EXPORTS void Scharr(const GpuMat& src, GpuMat& dst, int ddepth, int dx, int dy, double scale = 1,
+ int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
+CV_EXPORTS void Scharr(const GpuMat& src, GpuMat& dst, int ddepth, int dx, int dy, GpuMat& buf, double scale = 1,
+ int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1, Stream& stream = Stream::Null());
+
+//! smooths the image using Gaussian filter.
+CV_EXPORTS void GaussianBlur(const GpuMat& src, GpuMat& dst, Size ksize, double sigma1, double sigma2 = 0,
+ int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
+CV_EXPORTS void GaussianBlur(const GpuMat& src, GpuMat& dst, Size ksize, GpuMat& buf, double sigma1, double sigma2 = 0,
+ int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1, Stream& stream = Stream::Null());
+
+//! applies Laplacian operator to the image
+//! supports only ksize = 1 and ksize = 3
+CV_EXPORTS void Laplacian(const GpuMat& src, GpuMat& dst, int ddepth, int ksize = 1, double scale = 1, int borderType = BORDER_DEFAULT, Stream& stream = Stream::Null());
+
+}} // namespace cv { namespace gpu {
+
+#endif /* __OPENCV_GPUFILTERS_HPP__ */
DEF_PARAM_TEST(Sz_Type_KernelSz, cv::Size, MatType, int);
-PERF_TEST_P(Sz_Type_KernelSz, Filters_Blur,
+PERF_TEST_P(Sz_Type_KernelSz, Blur,
Combine(GPU_TYPICAL_MAT_SIZES,
Values(CV_8UC1, CV_8UC4),
Values(3, 5, 7)))
//////////////////////////////////////////////////////////////////////
// Sobel
-PERF_TEST_P(Sz_Type_KernelSz, Filters_Sobel, Combine(GPU_TYPICAL_MAT_SIZES, Values(CV_8UC1, CV_8UC4, CV_32FC1), Values(3, 5, 7, 9, 11, 13, 15)))
+PERF_TEST_P(Sz_Type_KernelSz, Sobel, Combine(GPU_TYPICAL_MAT_SIZES, Values(CV_8UC1, CV_8UC4, CV_32FC1), Values(3, 5, 7, 9, 11, 13, 15)))
{
declare.time(20.0);
//////////////////////////////////////////////////////////////////////
// Scharr
-PERF_TEST_P(Sz_Type, Filters_Scharr, Combine(GPU_TYPICAL_MAT_SIZES, Values(CV_8UC1, CV_8UC4, CV_32FC1)))
+PERF_TEST_P(Sz_Type, Scharr, Combine(GPU_TYPICAL_MAT_SIZES, Values(CV_8UC1, CV_8UC4, CV_32FC1)))
{
declare.time(20.0);
//////////////////////////////////////////////////////////////////////
// GaussianBlur
-PERF_TEST_P(Sz_Type_KernelSz, Filters_GaussianBlur, Combine(GPU_TYPICAL_MAT_SIZES, Values(CV_8UC1, CV_8UC4, CV_32FC1), Values(3, 5, 7, 9, 11, 13, 15)))
+PERF_TEST_P(Sz_Type_KernelSz, GaussianBlur, Combine(GPU_TYPICAL_MAT_SIZES, Values(CV_8UC1, CV_8UC4, CV_32FC1), Values(3, 5, 7, 9, 11, 13, 15)))
{
declare.time(20.0);
//////////////////////////////////////////////////////////////////////
// Laplacian
-PERF_TEST_P(Sz_Type_KernelSz, Filters_Laplacian, Combine(GPU_TYPICAL_MAT_SIZES, Values(CV_8UC1, CV_8UC4, CV_32FC1, CV_32FC4), Values(1, 3)))
+PERF_TEST_P(Sz_Type_KernelSz, Laplacian, Combine(GPU_TYPICAL_MAT_SIZES, Values(CV_8UC1, CV_8UC4, CV_32FC1, CV_32FC4), Values(1, 3)))
{
declare.time(20.0);
//////////////////////////////////////////////////////////////////////
// Erode
-PERF_TEST_P(Sz_Type, Filters_Erode, Combine(GPU_TYPICAL_MAT_SIZES, Values(CV_8UC1, CV_8UC4)))
+PERF_TEST_P(Sz_Type, Erode, Combine(GPU_TYPICAL_MAT_SIZES, Values(CV_8UC1, CV_8UC4)))
{
declare.time(20.0);
//////////////////////////////////////////////////////////////////////
// Dilate
-PERF_TEST_P(Sz_Type, Filters_Dilate, Combine(GPU_TYPICAL_MAT_SIZES, Values(CV_8UC1, CV_8UC4)))
+PERF_TEST_P(Sz_Type, Dilate, Combine(GPU_TYPICAL_MAT_SIZES, Values(CV_8UC1, CV_8UC4)))
{
declare.time(20.0);
DEF_PARAM_TEST(Sz_Type_Op, cv::Size, MatType, MorphOp);
-PERF_TEST_P(Sz_Type_Op, Filters_MorphologyEx, Combine(GPU_TYPICAL_MAT_SIZES, Values(CV_8UC1, CV_8UC4), MorphOp::all()))
+PERF_TEST_P(Sz_Type_Op, MorphologyEx, Combine(GPU_TYPICAL_MAT_SIZES, Values(CV_8UC1, CV_8UC4), MorphOp::all()))
{
declare.time(20.0);
//////////////////////////////////////////////////////////////////////
// Filter2D
-PERF_TEST_P(Sz_Type_KernelSz, Filters_Filter2D, Combine(GPU_TYPICAL_MAT_SIZES, Values(CV_8UC1, CV_8UC4, CV_32FC1, CV_32FC4), Values(3, 5, 7, 9, 11, 13, 15)))
+PERF_TEST_P(Sz_Type_KernelSz, Filter2D, Combine(GPU_TYPICAL_MAT_SIZES, Values(CV_8UC1, CV_8UC4, CV_32FC1, CV_32FC4), Values(3, 5, 7, 9, 11, 13, 15)))
{
declare.time(20.0);
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace perf;
+
+CV_PERF_TEST_MAIN(gpufilters, printCudaInfo())
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifdef __GNUC__
+# pragma GCC diagnostic ignored "-Wmissing-declarations"
+# if defined __clang__ || defined __APPLE__
+# pragma GCC diagnostic ignored "-Wmissing-prototypes"
+# pragma GCC diagnostic ignored "-Wextra"
+# endif
+#endif
+
+#ifndef __OPENCV_PERF_PRECOMP_HPP__
+#define __OPENCV_PERF_PRECOMP_HPP__
+
+#include "opencv2/ts.hpp"
+#include "opencv2/ts/gpu_perf.hpp"
+
+#include "opencv2/gpufilters.hpp"
+#include "opencv2/imgproc.hpp"
+
+#ifdef GTEST_CREATE_SHARED_LIBRARY
+#error no modules except ts should have GTEST_CREATE_SHARED_LIBRARY defined
+#endif
+
+#endif
#if !defined CUDA_DISABLER
-#include "column_filter.h"
+#include "column_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "column_filter.h"
+#include "column_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "column_filter.h"
+#include "column_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "column_filter.h"
+#include "column_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "column_filter.h"
+#include "column_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "column_filter.h"
+#include "column_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "column_filter.h"
+#include "column_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "column_filter.h"
+#include "column_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "column_filter.h"
+#include "column_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "column_filter.h"
+#include "column_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "column_filter.h"
+#include "column_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "column_filter.h"
+#include "column_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "column_filter.h"
+#include "column_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "column_filter.h"
+#include "column_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "column_filter.h"
+#include "column_filter.hpp"
namespace filter
{
{
{
0,
- column_filter::caller< 1, T, D, BrdColReflect101>,
- column_filter::caller< 2, T, D, BrdColReflect101>,
- column_filter::caller< 3, T, D, BrdColReflect101>,
- column_filter::caller< 4, T, D, BrdColReflect101>,
- column_filter::caller< 5, T, D, BrdColReflect101>,
- column_filter::caller< 6, T, D, BrdColReflect101>,
- column_filter::caller< 7, T, D, BrdColReflect101>,
- column_filter::caller< 8, T, D, BrdColReflect101>,
- column_filter::caller< 9, T, D, BrdColReflect101>,
- column_filter::caller<10, T, D, BrdColReflect101>,
- column_filter::caller<11, T, D, BrdColReflect101>,
- column_filter::caller<12, T, D, BrdColReflect101>,
- column_filter::caller<13, T, D, BrdColReflect101>,
- column_filter::caller<14, T, D, BrdColReflect101>,
- column_filter::caller<15, T, D, BrdColReflect101>,
- column_filter::caller<16, T, D, BrdColReflect101>,
- column_filter::caller<17, T, D, BrdColReflect101>,
- column_filter::caller<18, T, D, BrdColReflect101>,
- column_filter::caller<19, T, D, BrdColReflect101>,
- column_filter::caller<20, T, D, BrdColReflect101>,
- column_filter::caller<21, T, D, BrdColReflect101>,
- column_filter::caller<22, T, D, BrdColReflect101>,
- column_filter::caller<23, T, D, BrdColReflect101>,
- column_filter::caller<24, T, D, BrdColReflect101>,
- column_filter::caller<25, T, D, BrdColReflect101>,
- column_filter::caller<26, T, D, BrdColReflect101>,
- column_filter::caller<27, T, D, BrdColReflect101>,
- column_filter::caller<28, T, D, BrdColReflect101>,
- column_filter::caller<29, T, D, BrdColReflect101>,
- column_filter::caller<30, T, D, BrdColReflect101>,
- column_filter::caller<31, T, D, BrdColReflect101>,
- column_filter::caller<32, T, D, BrdColReflect101>
+ column_filter::caller< 1, T, D, BrdColConstant>,
+ column_filter::caller< 2, T, D, BrdColConstant>,
+ column_filter::caller< 3, T, D, BrdColConstant>,
+ column_filter::caller< 4, T, D, BrdColConstant>,
+ column_filter::caller< 5, T, D, BrdColConstant>,
+ column_filter::caller< 6, T, D, BrdColConstant>,
+ column_filter::caller< 7, T, D, BrdColConstant>,
+ column_filter::caller< 8, T, D, BrdColConstant>,
+ column_filter::caller< 9, T, D, BrdColConstant>,
+ column_filter::caller<10, T, D, BrdColConstant>,
+ column_filter::caller<11, T, D, BrdColConstant>,
+ column_filter::caller<12, T, D, BrdColConstant>,
+ column_filter::caller<13, T, D, BrdColConstant>,
+ column_filter::caller<14, T, D, BrdColConstant>,
+ column_filter::caller<15, T, D, BrdColConstant>,
+ column_filter::caller<16, T, D, BrdColConstant>,
+ column_filter::caller<17, T, D, BrdColConstant>,
+ column_filter::caller<18, T, D, BrdColConstant>,
+ column_filter::caller<19, T, D, BrdColConstant>,
+ column_filter::caller<20, T, D, BrdColConstant>,
+ column_filter::caller<21, T, D, BrdColConstant>,
+ column_filter::caller<22, T, D, BrdColConstant>,
+ column_filter::caller<23, T, D, BrdColConstant>,
+ column_filter::caller<24, T, D, BrdColConstant>,
+ column_filter::caller<25, T, D, BrdColConstant>,
+ column_filter::caller<26, T, D, BrdColConstant>,
+ column_filter::caller<27, T, D, BrdColConstant>,
+ column_filter::caller<28, T, D, BrdColConstant>,
+ column_filter::caller<29, T, D, BrdColConstant>,
+ column_filter::caller<30, T, D, BrdColConstant>,
+ column_filter::caller<31, T, D, BrdColConstant>,
+ column_filter::caller<32, T, D, BrdColConstant>
},
{
0,
},
{
0,
- column_filter::caller< 1, T, D, BrdColConstant>,
- column_filter::caller< 2, T, D, BrdColConstant>,
- column_filter::caller< 3, T, D, BrdColConstant>,
- column_filter::caller< 4, T, D, BrdColConstant>,
- column_filter::caller< 5, T, D, BrdColConstant>,
- column_filter::caller< 6, T, D, BrdColConstant>,
- column_filter::caller< 7, T, D, BrdColConstant>,
- column_filter::caller< 8, T, D, BrdColConstant>,
- column_filter::caller< 9, T, D, BrdColConstant>,
- column_filter::caller<10, T, D, BrdColConstant>,
- column_filter::caller<11, T, D, BrdColConstant>,
- column_filter::caller<12, T, D, BrdColConstant>,
- column_filter::caller<13, T, D, BrdColConstant>,
- column_filter::caller<14, T, D, BrdColConstant>,
- column_filter::caller<15, T, D, BrdColConstant>,
- column_filter::caller<16, T, D, BrdColConstant>,
- column_filter::caller<17, T, D, BrdColConstant>,
- column_filter::caller<18, T, D, BrdColConstant>,
- column_filter::caller<19, T, D, BrdColConstant>,
- column_filter::caller<20, T, D, BrdColConstant>,
- column_filter::caller<21, T, D, BrdColConstant>,
- column_filter::caller<22, T, D, BrdColConstant>,
- column_filter::caller<23, T, D, BrdColConstant>,
- column_filter::caller<24, T, D, BrdColConstant>,
- column_filter::caller<25, T, D, BrdColConstant>,
- column_filter::caller<26, T, D, BrdColConstant>,
- column_filter::caller<27, T, D, BrdColConstant>,
- column_filter::caller<28, T, D, BrdColConstant>,
- column_filter::caller<29, T, D, BrdColConstant>,
- column_filter::caller<30, T, D, BrdColConstant>,
- column_filter::caller<31, T, D, BrdColConstant>,
- column_filter::caller<32, T, D, BrdColConstant>
- },
- {
- 0,
column_filter::caller< 1, T, D, BrdColReflect>,
column_filter::caller< 2, T, D, BrdColReflect>,
column_filter::caller< 3, T, D, BrdColReflect>,
column_filter::caller<30, T, D, BrdColWrap>,
column_filter::caller<31, T, D, BrdColWrap>,
column_filter::caller<32, T, D, BrdColWrap>
+ },
+ {
+ 0,
+ column_filter::caller< 1, T, D, BrdColReflect101>,
+ column_filter::caller< 2, T, D, BrdColReflect101>,
+ column_filter::caller< 3, T, D, BrdColReflect101>,
+ column_filter::caller< 4, T, D, BrdColReflect101>,
+ column_filter::caller< 5, T, D, BrdColReflect101>,
+ column_filter::caller< 6, T, D, BrdColReflect101>,
+ column_filter::caller< 7, T, D, BrdColReflect101>,
+ column_filter::caller< 8, T, D, BrdColReflect101>,
+ column_filter::caller< 9, T, D, BrdColReflect101>,
+ column_filter::caller<10, T, D, BrdColReflect101>,
+ column_filter::caller<11, T, D, BrdColReflect101>,
+ column_filter::caller<12, T, D, BrdColReflect101>,
+ column_filter::caller<13, T, D, BrdColReflect101>,
+ column_filter::caller<14, T, D, BrdColReflect101>,
+ column_filter::caller<15, T, D, BrdColReflect101>,
+ column_filter::caller<16, T, D, BrdColReflect101>,
+ column_filter::caller<17, T, D, BrdColReflect101>,
+ column_filter::caller<18, T, D, BrdColReflect101>,
+ column_filter::caller<19, T, D, BrdColReflect101>,
+ column_filter::caller<20, T, D, BrdColReflect101>,
+ column_filter::caller<21, T, D, BrdColReflect101>,
+ column_filter::caller<22, T, D, BrdColReflect101>,
+ column_filter::caller<23, T, D, BrdColReflect101>,
+ column_filter::caller<24, T, D, BrdColReflect101>,
+ column_filter::caller<25, T, D, BrdColReflect101>,
+ column_filter::caller<26, T, D, BrdColReflect101>,
+ column_filter::caller<27, T, D, BrdColReflect101>,
+ column_filter::caller<28, T, D, BrdColReflect101>,
+ column_filter::caller<29, T, D, BrdColReflect101>,
+ column_filter::caller<30, T, D, BrdColReflect101>,
+ column_filter::caller<31, T, D, BrdColReflect101>,
+ column_filter::caller<32, T, D, BrdColReflect101>
}
};
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/border_interpolate.hpp"
+
+namespace cv { namespace gpu { namespace cudev
+{
+ namespace imgproc
+ {
+ #define FILTER2D_MAX_KERNEL_SIZE 16
+
+ __constant__ float c_filter2DKernel[FILTER2D_MAX_KERNEL_SIZE * FILTER2D_MAX_KERNEL_SIZE];
+
+ template <class SrcT, typename D>
+ __global__ void filter2D(const SrcT src, PtrStepSz<D> dst, const int kWidth, const int kHeight, const int anchorX, const int anchorY)
+ {
+ typedef typename TypeVec<float, VecTraits<D>::cn>::vec_type sum_t;
+
+ const int x = blockIdx.x * blockDim.x + threadIdx.x;
+ const int y = blockIdx.y * blockDim.y + threadIdx.y;
+
+ if (x >= dst.cols || y >= dst.rows)
+ return;
+
+ sum_t res = VecTraits<sum_t>::all(0);
+ int kInd = 0;
+
+ for (int i = 0; i < kHeight; ++i)
+ {
+ for (int j = 0; j < kWidth; ++j)
+ res = res + src(y - anchorY + i, x - anchorX + j) * c_filter2DKernel[kInd++];
+ }
+
+ dst(y, x) = saturate_cast<D>(res);
+ }
+
+ template <typename T, typename D, template <typename> class Brd> struct Filter2DCaller;
+
+ #define IMPLEMENT_FILTER2D_TEX_READER(type) \
+ texture< type , cudaTextureType2D, cudaReadModeElementType> tex_filter2D_ ## type (0, cudaFilterModePoint, cudaAddressModeClamp); \
+ struct tex_filter2D_ ## type ## _reader \
+ { \
+ typedef type elem_type; \
+ typedef int index_type; \
+ const int xoff; \
+ const int yoff; \
+ tex_filter2D_ ## type ## _reader (int xoff_, int yoff_) : xoff(xoff_), yoff(yoff_) {} \
+ __device__ __forceinline__ elem_type operator ()(index_type y, index_type x) const \
+ { \
+ return tex2D(tex_filter2D_ ## type , x + xoff, y + yoff); \
+ } \
+ }; \
+ template <typename D, template <typename> class Brd> struct Filter2DCaller< type , D, Brd> \
+ { \
+ static void call(const PtrStepSz< type > srcWhole, int xoff, int yoff, PtrStepSz<D> dst, \
+ int kWidth, int kHeight, int anchorX, int anchorY, const float* borderValue, cudaStream_t stream) \
+ { \
+ typedef typename TypeVec<float, VecTraits< type >::cn>::vec_type work_type; \
+ dim3 block(16, 16); \
+ dim3 grid(divUp(dst.cols, block.x), divUp(dst.rows, block.y)); \
+ bindTexture(&tex_filter2D_ ## type , srcWhole); \
+ tex_filter2D_ ## type ##_reader texSrc(xoff, yoff); \
+ Brd<work_type> brd(dst.rows, dst.cols, VecTraits<work_type>::make(borderValue)); \
+ BorderReader< tex_filter2D_ ## type ##_reader, Brd<work_type> > brdSrc(texSrc, brd); \
+ filter2D<<<grid, block, 0, stream>>>(brdSrc, dst, kWidth, kHeight, anchorX, anchorY); \
+ cudaSafeCall( cudaGetLastError() ); \
+ if (stream == 0) \
+ cudaSafeCall( cudaDeviceSynchronize() ); \
+ } \
+ };
+
+ IMPLEMENT_FILTER2D_TEX_READER(uchar);
+ IMPLEMENT_FILTER2D_TEX_READER(uchar4);
+
+ IMPLEMENT_FILTER2D_TEX_READER(ushort);
+ IMPLEMENT_FILTER2D_TEX_READER(ushort4);
+
+ IMPLEMENT_FILTER2D_TEX_READER(float);
+ IMPLEMENT_FILTER2D_TEX_READER(float4);
+
+ #undef IMPLEMENT_FILTER2D_TEX_READER
+
+ template <typename T, typename D>
+ void filter2D_gpu(PtrStepSzb srcWhole, int ofsX, int ofsY, PtrStepSzb dst,
+ int kWidth, int kHeight, int anchorX, int anchorY, const float* kernel,
+ int borderMode, const float* borderValue, cudaStream_t stream)
+ {
+ typedef void (*func_t)(const PtrStepSz<T> srcWhole, int xoff, int yoff, PtrStepSz<D> dst, int kWidth, int kHeight, int anchorX, int anchorY, const float* borderValue, cudaStream_t stream);
+ static const func_t funcs[] =
+ {
+ Filter2DCaller<T, D, BrdConstant>::call,
+ Filter2DCaller<T, D, BrdReplicate>::call,
+ Filter2DCaller<T, D, BrdReflect>::call,
+ Filter2DCaller<T, D, BrdWrap>::call,
+ Filter2DCaller<T, D, BrdReflect101>::call
+ };
+
+ if (stream == 0)
+ cudaSafeCall( cudaMemcpyToSymbol(c_filter2DKernel, kernel, kWidth * kHeight * sizeof(float), 0, cudaMemcpyDeviceToDevice) );
+ else
+ cudaSafeCall( cudaMemcpyToSymbolAsync(c_filter2DKernel, kernel, kWidth * kHeight * sizeof(float), 0, cudaMemcpyDeviceToDevice, stream) );
+
+ funcs[borderMode](static_cast< PtrStepSz<T> >(srcWhole), ofsX, ofsY, static_cast< PtrStepSz<D> >(dst), kWidth, kHeight, anchorX, anchorY, borderValue, stream);
+ }
+
+ template void filter2D_gpu<uchar, uchar>(PtrStepSzb srcWhole, int ofsX, int ofsY, PtrStepSzb dst, int kWidth, int kHeight, int anchorX, int anchorY, const float* kernel, int borderMode, const float* borderValue, cudaStream_t stream);
+ template void filter2D_gpu<uchar4, uchar4>(PtrStepSzb srcWhole, int ofsX, int ofsY, PtrStepSzb dst, int kWidth, int kHeight, int anchorX, int anchorY, const float* kernel, int borderMode, const float* borderValue, cudaStream_t stream);
+ template void filter2D_gpu<ushort, ushort>(PtrStepSzb srcWhole, int ofsX, int ofsY, PtrStepSzb dst, int kWidth, int kHeight, int anchorX, int anchorY, const float* kernel, int borderMode, const float* borderValue, cudaStream_t stream);
+ template void filter2D_gpu<ushort4, ushort4>(PtrStepSzb srcWhole, int ofsX, int ofsY, PtrStepSzb dst, int kWidth, int kHeight, int anchorX, int anchorY, const float* kernel, int borderMode, const float* borderValue, cudaStream_t stream);
+ template void filter2D_gpu<float, float>(PtrStepSzb srcWhole, int ofsX, int ofsY, PtrStepSzb dst, int kWidth, int kHeight, int anchorX, int anchorY, const float* kernel, int borderMode, const float* borderValue, cudaStream_t stream);
+ template void filter2D_gpu<float4, float4>(PtrStepSzb srcWhole, int ofsX, int ofsY, PtrStepSzb dst, int kWidth, int kHeight, int anchorX, int anchorY, const float* kernel, int borderMode, const float* borderValue, cudaStream_t stream);
+ }
+}}}
+
+#endif // CUDA_DISABLER
#if !defined CUDA_DISABLER
-#include "row_filter.h"
+#include "row_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "row_filter.h"
+#include "row_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "row_filter.h"
+#include "row_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "row_filter.h"
+#include "row_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "row_filter.h"
+#include "row_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "row_filter.h"
+#include "row_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "row_filter.h"
+#include "row_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "row_filter.h"
+#include "row_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "row_filter.h"
+#include "row_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "row_filter.h"
+#include "row_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "row_filter.h"
+#include "row_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "row_filter.h"
+#include "row_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "row_filter.h"
+#include "row_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "row_filter.h"
+#include "row_filter.hpp"
namespace filter
{
#if !defined CUDA_DISABLER
-#include "row_filter.h"
+#include "row_filter.hpp"
namespace filter
{
{
{
0,
- row_filter::caller< 1, T, D, BrdRowReflect101>,
- row_filter::caller< 2, T, D, BrdRowReflect101>,
- row_filter::caller< 3, T, D, BrdRowReflect101>,
- row_filter::caller< 4, T, D, BrdRowReflect101>,
- row_filter::caller< 5, T, D, BrdRowReflect101>,
- row_filter::caller< 6, T, D, BrdRowReflect101>,
- row_filter::caller< 7, T, D, BrdRowReflect101>,
- row_filter::caller< 8, T, D, BrdRowReflect101>,
- row_filter::caller< 9, T, D, BrdRowReflect101>,
- row_filter::caller<10, T, D, BrdRowReflect101>,
- row_filter::caller<11, T, D, BrdRowReflect101>,
- row_filter::caller<12, T, D, BrdRowReflect101>,
- row_filter::caller<13, T, D, BrdRowReflect101>,
- row_filter::caller<14, T, D, BrdRowReflect101>,
- row_filter::caller<15, T, D, BrdRowReflect101>,
- row_filter::caller<16, T, D, BrdRowReflect101>,
- row_filter::caller<17, T, D, BrdRowReflect101>,
- row_filter::caller<18, T, D, BrdRowReflect101>,
- row_filter::caller<19, T, D, BrdRowReflect101>,
- row_filter::caller<20, T, D, BrdRowReflect101>,
- row_filter::caller<21, T, D, BrdRowReflect101>,
- row_filter::caller<22, T, D, BrdRowReflect101>,
- row_filter::caller<23, T, D, BrdRowReflect101>,
- row_filter::caller<24, T, D, BrdRowReflect101>,
- row_filter::caller<25, T, D, BrdRowReflect101>,
- row_filter::caller<26, T, D, BrdRowReflect101>,
- row_filter::caller<27, T, D, BrdRowReflect101>,
- row_filter::caller<28, T, D, BrdRowReflect101>,
- row_filter::caller<29, T, D, BrdRowReflect101>,
- row_filter::caller<30, T, D, BrdRowReflect101>,
- row_filter::caller<31, T, D, BrdRowReflect101>,
- row_filter::caller<32, T, D, BrdRowReflect101>
+ row_filter::caller< 1, T, D, BrdRowConstant>,
+ row_filter::caller< 2, T, D, BrdRowConstant>,
+ row_filter::caller< 3, T, D, BrdRowConstant>,
+ row_filter::caller< 4, T, D, BrdRowConstant>,
+ row_filter::caller< 5, T, D, BrdRowConstant>,
+ row_filter::caller< 6, T, D, BrdRowConstant>,
+ row_filter::caller< 7, T, D, BrdRowConstant>,
+ row_filter::caller< 8, T, D, BrdRowConstant>,
+ row_filter::caller< 9, T, D, BrdRowConstant>,
+ row_filter::caller<10, T, D, BrdRowConstant>,
+ row_filter::caller<11, T, D, BrdRowConstant>,
+ row_filter::caller<12, T, D, BrdRowConstant>,
+ row_filter::caller<13, T, D, BrdRowConstant>,
+ row_filter::caller<14, T, D, BrdRowConstant>,
+ row_filter::caller<15, T, D, BrdRowConstant>,
+ row_filter::caller<16, T, D, BrdRowConstant>,
+ row_filter::caller<17, T, D, BrdRowConstant>,
+ row_filter::caller<18, T, D, BrdRowConstant>,
+ row_filter::caller<19, T, D, BrdRowConstant>,
+ row_filter::caller<20, T, D, BrdRowConstant>,
+ row_filter::caller<21, T, D, BrdRowConstant>,
+ row_filter::caller<22, T, D, BrdRowConstant>,
+ row_filter::caller<23, T, D, BrdRowConstant>,
+ row_filter::caller<24, T, D, BrdRowConstant>,
+ row_filter::caller<25, T, D, BrdRowConstant>,
+ row_filter::caller<26, T, D, BrdRowConstant>,
+ row_filter::caller<27, T, D, BrdRowConstant>,
+ row_filter::caller<28, T, D, BrdRowConstant>,
+ row_filter::caller<29, T, D, BrdRowConstant>,
+ row_filter::caller<30, T, D, BrdRowConstant>,
+ row_filter::caller<31, T, D, BrdRowConstant>,
+ row_filter::caller<32, T, D, BrdRowConstant>
},
{
0,
},
{
0,
- row_filter::caller< 1, T, D, BrdRowConstant>,
- row_filter::caller< 2, T, D, BrdRowConstant>,
- row_filter::caller< 3, T, D, BrdRowConstant>,
- row_filter::caller< 4, T, D, BrdRowConstant>,
- row_filter::caller< 5, T, D, BrdRowConstant>,
- row_filter::caller< 6, T, D, BrdRowConstant>,
- row_filter::caller< 7, T, D, BrdRowConstant>,
- row_filter::caller< 8, T, D, BrdRowConstant>,
- row_filter::caller< 9, T, D, BrdRowConstant>,
- row_filter::caller<10, T, D, BrdRowConstant>,
- row_filter::caller<11, T, D, BrdRowConstant>,
- row_filter::caller<12, T, D, BrdRowConstant>,
- row_filter::caller<13, T, D, BrdRowConstant>,
- row_filter::caller<14, T, D, BrdRowConstant>,
- row_filter::caller<15, T, D, BrdRowConstant>,
- row_filter::caller<16, T, D, BrdRowConstant>,
- row_filter::caller<17, T, D, BrdRowConstant>,
- row_filter::caller<18, T, D, BrdRowConstant>,
- row_filter::caller<19, T, D, BrdRowConstant>,
- row_filter::caller<20, T, D, BrdRowConstant>,
- row_filter::caller<21, T, D, BrdRowConstant>,
- row_filter::caller<22, T, D, BrdRowConstant>,
- row_filter::caller<23, T, D, BrdRowConstant>,
- row_filter::caller<24, T, D, BrdRowConstant>,
- row_filter::caller<25, T, D, BrdRowConstant>,
- row_filter::caller<26, T, D, BrdRowConstant>,
- row_filter::caller<27, T, D, BrdRowConstant>,
- row_filter::caller<28, T, D, BrdRowConstant>,
- row_filter::caller<29, T, D, BrdRowConstant>,
- row_filter::caller<30, T, D, BrdRowConstant>,
- row_filter::caller<31, T, D, BrdRowConstant>,
- row_filter::caller<32, T, D, BrdRowConstant>
- },
- {
- 0,
row_filter::caller< 1, T, D, BrdRowReflect>,
row_filter::caller< 2, T, D, BrdRowReflect>,
row_filter::caller< 3, T, D, BrdRowReflect>,
row_filter::caller<30, T, D, BrdRowWrap>,
row_filter::caller<31, T, D, BrdRowWrap>,
row_filter::caller<32, T, D, BrdRowWrap>
+ },
+ {
+ 0,
+ row_filter::caller< 1, T, D, BrdRowReflect101>,
+ row_filter::caller< 2, T, D, BrdRowReflect101>,
+ row_filter::caller< 3, T, D, BrdRowReflect101>,
+ row_filter::caller< 4, T, D, BrdRowReflect101>,
+ row_filter::caller< 5, T, D, BrdRowReflect101>,
+ row_filter::caller< 6, T, D, BrdRowReflect101>,
+ row_filter::caller< 7, T, D, BrdRowReflect101>,
+ row_filter::caller< 8, T, D, BrdRowReflect101>,
+ row_filter::caller< 9, T, D, BrdRowReflect101>,
+ row_filter::caller<10, T, D, BrdRowReflect101>,
+ row_filter::caller<11, T, D, BrdRowReflect101>,
+ row_filter::caller<12, T, D, BrdRowReflect101>,
+ row_filter::caller<13, T, D, BrdRowReflect101>,
+ row_filter::caller<14, T, D, BrdRowReflect101>,
+ row_filter::caller<15, T, D, BrdRowReflect101>,
+ row_filter::caller<16, T, D, BrdRowReflect101>,
+ row_filter::caller<17, T, D, BrdRowReflect101>,
+ row_filter::caller<18, T, D, BrdRowReflect101>,
+ row_filter::caller<19, T, D, BrdRowReflect101>,
+ row_filter::caller<20, T, D, BrdRowReflect101>,
+ row_filter::caller<21, T, D, BrdRowReflect101>,
+ row_filter::caller<22, T, D, BrdRowReflect101>,
+ row_filter::caller<23, T, D, BrdRowReflect101>,
+ row_filter::caller<24, T, D, BrdRowReflect101>,
+ row_filter::caller<25, T, D, BrdRowReflect101>,
+ row_filter::caller<26, T, D, BrdRowReflect101>,
+ row_filter::caller<27, T, D, BrdRowReflect101>,
+ row_filter::caller<28, T, D, BrdRowReflect101>,
+ row_filter::caller<29, T, D, BrdRowReflect101>,
+ row_filter::caller<30, T, D, BrdRowReflect101>,
+ row_filter::caller<31, T, D, BrdRowReflect101>,
+ row_filter::caller<32, T, D, BrdRowReflect101>
}
};
using namespace cv;
using namespace cv::gpu;
-
#if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
Ptr<FilterEngine_GPU> cv::gpu::createFilter2D_GPU(const Ptr<BaseFilter_GPU>&, int, int) { throw_no_cuda(); return Ptr<FilterEngine_GPU>(0); }
{
switch( op )
{
- case MORPH_ERODE: erode(src, dst, kernel, buf1, anchor, iterations, stream); break;
- case MORPH_DILATE: dilate(src, dst, kernel, buf1, anchor, iterations, stream); break;
+ case MORPH_ERODE:
+ erode(src, dst, kernel, buf1, anchor, iterations, stream);
+ break;
+
+ case MORPH_DILATE:
+ dilate(src, dst, kernel, buf1, anchor, iterations, stream);
+ break;
+
case MORPH_OPEN:
erode(src, buf2, kernel, buf1, anchor, iterations, stream);
dilate(buf2, dst, kernel, buf1, anchor, iterations, stream);
break;
+
case MORPH_CLOSE:
dilate(src, buf2, kernel, buf1, anchor, iterations, stream);
erode(buf2, dst, kernel, buf1, anchor, iterations, stream);
break;
+
+#ifdef HAVE_OPENCV_GPUARITHM
case MORPH_GRADIENT:
erode(src, buf2, kernel, buf1, anchor, iterations, stream);
dilate(src, dst, kernel, buf1, anchor, iterations, stream);
- subtract(dst, buf2, dst, GpuMat(), -1, stream);
+ gpu::subtract(dst, buf2, dst, GpuMat(), -1, stream);
break;
+
case MORPH_TOPHAT:
erode(src, dst, kernel, buf1, anchor, iterations, stream);
dilate(dst, buf2, kernel, buf1, anchor, iterations, stream);
- subtract(src, buf2, dst, GpuMat(), -1, stream);
+ gpu::subtract(src, buf2, dst, GpuMat(), -1, stream);
break;
+
case MORPH_BLACKHAT:
dilate(src, dst, kernel, buf1, anchor, iterations, stream);
erode(dst, buf2, kernel, buf1, anchor, iterations, stream);
- subtract(buf2, src, dst, GpuMat(), -1, stream);
+ gpu::subtract(buf2, src, dst, GpuMat(), -1, stream);
break;
+#endif
+
default:
CV_Error(cv::Error::StsBadArg, "unknown morphological operation");
}
CV_Assert(ksize.width * ksize.height <= 16 * 16);
- int gpuBorderType;
- CV_Assert( tryConvertToGpuBorderType(brd_type, gpuBorderType) );
-
GpuMat gpu_krnl;
normalizeKernel(kernel, gpu_krnl, CV_32F);
break;
}
- return Ptr<BaseFilter_GPU>(new GpuFilter2D(ksize, anchor, func, gpu_krnl, gpuBorderType));
+ return Ptr<BaseFilter_GPU>(new GpuFilter2D(ksize, anchor, func, gpu_krnl, brd_type));
}
Ptr<FilterEngine_GPU> cv::gpu::createLinearFilter_GPU(int srcType, int dstType, const Mat& kernel, Point anchor, int borderType)
CV_Assert( borderType == BORDER_REFLECT101 || borderType == BORDER_REPLICATE || borderType == BORDER_CONSTANT || borderType == BORDER_REFLECT || borderType == BORDER_WRAP );
- int gpuBorderType;
- CV_Assert( tryConvertToGpuBorderType(borderType, gpuBorderType) );
-
const int sdepth = CV_MAT_DEPTH(srcType);
const int cn = CV_MAT_CN(srcType);
CV_Assert( sdepth <= CV_64F && cn <= 4 );
normalizeAnchor(anchor, ksize);
- return Ptr<BaseRowFilter_GPU>(new GpuLinearRowFilter(ksize, anchor, gpu_row_krnl, func, gpuBorderType));
+ return Ptr<BaseRowFilter_GPU>(new GpuLinearRowFilter(ksize, anchor, gpu_row_krnl, func, borderType));
}
namespace
CV_Assert( borderType == BORDER_REFLECT101 || borderType == BORDER_REPLICATE || borderType == BORDER_CONSTANT || borderType == BORDER_REFLECT || borderType == BORDER_WRAP );
- int gpuBorderType;
- CV_Assert( tryConvertToGpuBorderType(borderType, gpuBorderType) );
-
const int ddepth = CV_MAT_DEPTH(dstType);
const int cn = CV_MAT_CN(dstType);
CV_Assert( ddepth <= CV_64F && cn <= 4 );
normalizeAnchor(anchor, ksize);
- return Ptr<BaseColumnFilter_GPU>(new GpuLinearColumnFilter(ksize, anchor, gpu_col_krnl, func, gpuBorderType));
+ return Ptr<BaseColumnFilter_GPU>(new GpuLinearColumnFilter(ksize, anchor, gpu_col_krnl, func, borderType));
}
Ptr<FilterEngine_GPU> cv::gpu::createSeparableLinearFilter_GPU(int srcType, int dstType, const Mat& rowKernel, const Mat& columnKernel,
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_PRECOMP_H__
+#define __OPENCV_PRECOMP_H__
+
+#include <limits>
+
+#include "opencv2/gpufilters.hpp"
+#include "opencv2/imgproc.hpp"
+
+#include "opencv2/core/gpu_private.hpp"
+
+#include "opencv2/opencv_modules.hpp"
+
+#ifdef HAVE_OPENCV_GPUARITHM
+# include "opencv2/gpuarithm.hpp"
+#endif
+
+#endif /* __OPENCV_PRECOMP_H__ */
EXPECT_MAT_NEAR(getInnerROI(dst_gold, ksize), getInnerROI(dst, ksize), 1.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_Filter, Blur, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Filters, Blur, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatType(CV_8UC1), MatType(CV_8UC4)),
EXPECT_MAT_NEAR(getInnerROI(dst_gold, ksize), getInnerROI(dst, ksize), CV_MAT_DEPTH(type) < CV_32F ? 0.0 : 0.1);
}
-INSTANTIATE_TEST_CASE_P(GPU_Filter, Sobel, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Filters, Sobel, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatDepth(CV_8U), MatDepth(CV_16U), MatDepth(CV_16S), MatDepth(CV_32F)),
EXPECT_MAT_NEAR(getInnerROI(dst_gold, cv::Size(3, 3)), getInnerROI(dst, cv::Size(3, 3)), CV_MAT_DEPTH(type) < CV_32F ? 0.0 : 0.1);
}
-INSTANTIATE_TEST_CASE_P(GPU_Filter, Scharr, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Filters, Scharr, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatDepth(CV_8U), MatDepth(CV_16U), MatDepth(CV_16S), MatDepth(CV_32F)),
}
}
-INSTANTIATE_TEST_CASE_P(GPU_Filter, GaussianBlur, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Filters, GaussianBlur, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatDepth(CV_8U), MatDepth(CV_16U), MatDepth(CV_16S), MatDepth(CV_32F)),
EXPECT_MAT_NEAR(dst_gold, dst, src.depth() < CV_32F ? 0.0 : 1e-3);
}
-INSTANTIATE_TEST_CASE_P(GPU_Filter, Laplacian, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Filters, Laplacian, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatType(CV_8UC1), MatType(CV_8UC4), MatType(CV_32FC1)),
EXPECT_MAT_NEAR(getInnerROI(dst_gold, ksize), getInnerROI(dst, ksize), 0.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_Filter, Erode, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Filters, Erode, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatType(CV_8UC1), MatType(CV_8UC4)),
EXPECT_MAT_NEAR(getInnerROI(dst_gold, ksize), getInnerROI(dst, ksize), 0.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_Filter, Dilate, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Filters, Dilate, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatType(CV_8UC1), MatType(CV_8UC4)),
EXPECT_MAT_NEAR(getInnerROI(dst_gold, border), getInnerROI(dst, border), 0.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_Filter, MorphEx, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Filters, MorphEx, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatType(CV_8UC1), MatType(CV_8UC4)),
EXPECT_MAT_NEAR(dst_gold, dst, CV_MAT_DEPTH(type) == CV_32F ? 1e-1 : 1.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_Filter, Filter2D, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Filters, Filter2D, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatType(CV_8UC1), MatType(CV_8UC4), MatType(CV_16UC1), MatType(CV_16UC4), MatType(CV_32FC1), MatType(CV_32FC4)),
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+CV_GPU_TEST_MAIN("gpu")
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifdef __GNUC__
+# pragma GCC diagnostic ignored "-Wmissing-declarations"
+# if defined __clang__ || defined __APPLE__
+# pragma GCC diagnostic ignored "-Wmissing-prototypes"
+# pragma GCC diagnostic ignored "-Wextra"
+# endif
+#endif
+
+#ifndef __OPENCV_TEST_PRECOMP_HPP__
+#define __OPENCV_TEST_PRECOMP_HPP__
+
+#include "opencv2/ts.hpp"
+#include "opencv2/ts/gpu_test.hpp"
+
+#include "opencv2/gpufilters.hpp"
+#include "opencv2/imgproc.hpp"
+
+#endif
--- /dev/null
+if(ANDROID OR IOS)
+ ocv_module_disable(gpuimgproc)
+endif()
+
+set(the_description "GPU-accelerated Image Processing")
+
+ocv_warnings_disable(CMAKE_CXX_FLAGS /wd4127 /wd4100 /wd4324 /wd4512 /wd4515 -Wundef -Wmissing-declarations -Wshadow -Wunused-parameter)
+
+ocv_define_module(gpuimgproc opencv_imgproc opencv_gpufilters OPTIONAL opencv_gpuarithm)
--- /dev/null
+Color space processing
+======================
+
+.. highlight:: cpp
+
+
+
+gpu::cvtColor
+-----------------
+Converts an image from one color space to another.
+
+.. ocv:function:: void gpu::cvtColor(const GpuMat& src, GpuMat& dst, int code, int dcn = 0, Stream& stream = Stream::Null())
+
+ :param src: Source image with ``CV_8U`` , ``CV_16U`` , or ``CV_32F`` depth and 1, 3, or 4 channels.
+
+ :param dst: Destination image with the same size and depth as ``src`` .
+
+ :param code: Color space conversion code. For details, see :ocv:func:`cvtColor` . Conversion to/from Luv and Bayer color spaces is not supported.
+
+ :param dcn: Number of channels in the destination image. If the parameter is 0, the number of the channels is derived automatically from ``src`` and the ``code`` .
+
+ :param stream: Stream for the asynchronous version.
+
+3-channel color spaces (like ``HSV``, ``XYZ``, and so on) can be stored in a 4-channel image for better performance.
+
+.. seealso:: :ocv:func:`cvtColor`
+
+
+
+gpu::swapChannels
+-----------------
+Exchanges the color channels of an image in-place.
+
+.. ocv:function:: void gpu::swapChannels(GpuMat& image, const int dstOrder[4], Stream& stream = Stream::Null())
+
+ :param image: Source image. Supports only ``CV_8UC4`` type.
+
+ :param dstOrder: Integer array describing how channel values are permutated. The n-th entry of the array contains the number of the channel that is stored in the n-th channel of the output image. E.g. Given an RGBA image, aDstOrder = [3,2,1,0] converts this to ABGR channel order.
+
+ :param stream: Stream for the asynchronous version.
+
+The methods support arbitrary permutations of the original channels, including replication.
+
+
+
+gpu::alphaComp
+-------------------
+Composites two images using alpha opacity values contained in each image.
+
+.. ocv:function:: void gpu::alphaComp(const GpuMat& img1, const GpuMat& img2, GpuMat& dst, int alpha_op, Stream& stream = Stream::Null())
+
+ :param img1: First image. Supports ``CV_8UC4`` , ``CV_16UC4`` , ``CV_32SC4`` and ``CV_32FC4`` types.
+
+ :param img2: Second image. Must have the same size and the same type as ``img1`` .
+
+ :param dst: Destination image.
+
+ :param alpha_op: Flag specifying the alpha-blending operation:
+
+ * **ALPHA_OVER**
+ * **ALPHA_IN**
+ * **ALPHA_OUT**
+ * **ALPHA_ATOP**
+ * **ALPHA_XOR**
+ * **ALPHA_PLUS**
+ * **ALPHA_OVER_PREMUL**
+ * **ALPHA_IN_PREMUL**
+ * **ALPHA_OUT_PREMUL**
+ * **ALPHA_ATOP_PREMUL**
+ * **ALPHA_XOR_PREMUL**
+ * **ALPHA_PLUS_PREMUL**
+ * **ALPHA_PREMUL**
+
+ :param stream: Stream for the asynchronous version.
--- /dev/null
+Feature Detection
+=================
+
+.. highlight:: cpp
+
+
+
+gpu::cornerHarris
+---------------------
+Computes the Harris cornerness criteria at each image pixel.
+
+.. ocv:function:: void gpu::cornerHarris(const GpuMat& src, GpuMat& dst, int blockSize, int ksize, double k, int borderType=BORDER_REFLECT101)
+
+ :param src: Source image. Only ``CV_8UC1`` and ``CV_32FC1`` images are supported for now.
+
+ :param dst: Destination image containing cornerness values. It has the same size as ``src`` and ``CV_32FC1`` type.
+
+ :param blockSize: Neighborhood size.
+
+ :param ksize: Aperture parameter for the Sobel operator.
+
+ :param k: Harris detector free parameter.
+
+ :param borderType: Pixel extrapolation method. Only ``BORDER_REFLECT101`` and ``BORDER_REPLICATE`` are supported for now.
+
+.. seealso:: :ocv:func:`cornerHarris`
+
+
+
+gpu::cornerMinEigenVal
+--------------------------
+Computes the minimum eigen value of a 2x2 derivative covariation matrix at each pixel (the cornerness criteria).
+
+.. ocv:function:: void gpu::cornerMinEigenVal(const GpuMat& src, GpuMat& dst, int blockSize, int ksize, int borderType=BORDER_REFLECT101)
+
+.. ocv:function:: void gpu::cornerMinEigenVal(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, int blockSize, int ksize, int borderType=BORDER_REFLECT101)
+
+.. ocv:function:: void gpu::cornerMinEigenVal(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, GpuMat& buf, int blockSize, int ksize, int borderType=BORDER_REFLECT101, Stream& stream = Stream::Null())
+
+ :param src: Source image. Only ``CV_8UC1`` and ``CV_32FC1`` images are supported for now.
+
+ :param dst: Destination image containing cornerness values. The size is the same. The type is ``CV_32FC1`` .
+
+ :param blockSize: Neighborhood size.
+
+ :param ksize: Aperture parameter for the Sobel operator.
+
+ :param borderType: Pixel extrapolation method. Only ``BORDER_REFLECT101`` and ``BORDER_REPLICATE`` are supported for now.
+
+.. seealso:: :ocv:func:`cornerMinEigenVal`
+
+
+
+gpu::GoodFeaturesToTrackDetector_GPU
+------------------------------------
+.. ocv:class:: gpu::GoodFeaturesToTrackDetector_GPU
+
+Class used for strong corners detection on an image. ::
+
+ class GoodFeaturesToTrackDetector_GPU
+ {
+ public:
+ explicit GoodFeaturesToTrackDetector_GPU(int maxCorners_ = 1000, double qualityLevel_ = 0.01, double minDistance_ = 0.0,
+ int blockSize_ = 3, bool useHarrisDetector_ = false, double harrisK_ = 0.04);
+
+ void operator ()(const GpuMat& image, GpuMat& corners, const GpuMat& mask = GpuMat());
+
+ int maxCorners;
+ double qualityLevel;
+ double minDistance;
+
+ int blockSize;
+ bool useHarrisDetector;
+ double harrisK;
+
+ void releaseMemory();
+ };
+
+The class finds the most prominent corners in the image.
+
+.. seealso:: :ocv:func:`goodFeaturesToTrack`
--- /dev/null
+********************************************
+gpuimgproc. GPU-accelerated Image Processing
+********************************************
+
+.. toctree::
+ :maxdepth: 1
+
+ color
+ histogram
+ hough
+ feature_detection
+ imgproc
--- /dev/null
+Histogram Calculation
+=====================
+
+.. highlight:: cpp
+
+
+
+gpu::evenLevels
+-------------------
+Computes levels with even distribution.
+
+.. ocv:function:: void gpu::evenLevels(GpuMat& levels, int nLevels, int lowerLevel, int upperLevel)
+
+ :param levels: Destination array. ``levels`` has 1 row, ``nLevels`` columns, and the ``CV_32SC1`` type.
+
+ :param nLevels: Number of computed levels. ``nLevels`` must be at least 2.
+
+ :param lowerLevel: Lower boundary value of the lowest level.
+
+ :param upperLevel: Upper boundary value of the greatest level.
+
+
+
+gpu::histEven
+-----------------
+Calculates a histogram with evenly distributed bins.
+
+.. ocv:function:: void gpu::histEven(const GpuMat& src, GpuMat& hist, int histSize, int lowerLevel, int upperLevel, Stream& stream = Stream::Null())
+
+.. ocv:function:: void gpu::histEven(const GpuMat& src, GpuMat& hist, GpuMat& buf, int histSize, int lowerLevel, int upperLevel, Stream& stream = Stream::Null())
+
+.. ocv:function:: void gpu::histEven( const GpuMat& src, GpuMat hist[4], int histSize[4], int lowerLevel[4], int upperLevel[4], Stream& stream=Stream::Null() )
+
+.. ocv:function:: void gpu::histEven( const GpuMat& src, GpuMat hist[4], GpuMat& buf, int histSize[4], int lowerLevel[4], int upperLevel[4], Stream& stream=Stream::Null() )
+
+ :param src: Source image. ``CV_8U``, ``CV_16U``, or ``CV_16S`` depth and 1 or 4 channels are supported. For a four-channel image, all channels are processed separately.
+
+ :param hist: Destination histogram with one row, ``histSize`` columns, and the ``CV_32S`` type.
+
+ :param histSize: Size of the histogram.
+
+ :param lowerLevel: Lower boundary of lowest-level bin.
+
+ :param upperLevel: Upper boundary of highest-level bin.
+
+ :param buf: Optional buffer to avoid extra memory allocations (for many calls with the same sizes).
+
+ :param stream: Stream for the asynchronous version.
+
+
+
+gpu::histRange
+------------------
+Calculates a histogram with bins determined by the ``levels`` array.
+
+.. ocv:function:: void gpu::histRange(const GpuMat& src, GpuMat& hist, const GpuMat& levels, Stream& stream = Stream::Null())
+
+.. ocv:function:: void gpu::histRange(const GpuMat& src, GpuMat& hist, const GpuMat& levels, GpuMat& buf, Stream& stream = Stream::Null())
+
+ :param src: Source image. ``CV_8U`` , ``CV_16U`` , or ``CV_16S`` depth and 1 or 4 channels are supported. For a four-channel image, all channels are processed separately.
+
+ :param hist: Destination histogram with one row, ``(levels.cols-1)`` columns, and the ``CV_32SC1`` type.
+
+ :param levels: Number of levels in the histogram.
+
+ :param buf: Optional buffer to avoid extra memory allocations (for many calls with the same sizes).
+
+ :param stream: Stream for the asynchronous version.
+
+
+
+gpu::calcHist
+------------------
+Calculates histogram for one channel 8-bit image.
+
+.. ocv:function:: void gpu::calcHist(const GpuMat& src, GpuMat& hist, Stream& stream = Stream::Null())
+
+ :param src: Source image.
+
+ :param hist: Destination histogram with one row, 256 columns, and the ``CV_32SC1`` type.
+
+ :param stream: Stream for the asynchronous version.
+
+
+
+gpu::equalizeHist
+------------------
+Equalizes the histogram of a grayscale image.
+
+.. ocv:function:: void gpu::equalizeHist(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null())
+
+.. ocv:function:: void gpu::equalizeHist(const GpuMat& src, GpuMat& dst, GpuMat& hist, GpuMat& buf, Stream& stream = Stream::Null())
+
+ :param src: Source image.
+
+ :param dst: Destination image.
+
+ :param hist: Destination histogram with one row, 256 columns, and the ``CV_32SC1`` type.
+
+ :param buf: Optional buffer to avoid extra memory allocations (for many calls with the same sizes).
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`equalizeHist`
--- /dev/null
+Hough Transform
+===============
+
+.. highlight:: cpp
+
+
+
+gpu::HoughLines
+---------------
+Finds lines in a binary image using the classical Hough transform.
+
+.. ocv:function:: void gpu::HoughLines(const GpuMat& src, GpuMat& lines, float rho, float theta, int threshold, bool doSort = false, int maxLines = 4096)
+
+.. ocv:function:: void gpu::HoughLines(const GpuMat& src, GpuMat& lines, HoughLinesBuf& buf, float rho, float theta, int threshold, bool doSort = false, int maxLines = 4096)
+
+ :param src: 8-bit, single-channel binary source image.
+
+ :param lines: Output vector of lines. Each line is represented by a two-element vector :math:`(\rho, \theta)` . :math:`\rho` is the distance from the coordinate origin :math:`(0,0)` (top-left corner of the image). :math:`\theta` is the line rotation angle in radians ( :math:`0 \sim \textrm{vertical line}, \pi/2 \sim \textrm{horizontal line}` ).
+
+ :param rho: Distance resolution of the accumulator in pixels.
+
+ :param theta: Angle resolution of the accumulator in radians.
+
+ :param threshold: Accumulator threshold parameter. Only those lines are returned that get enough votes ( :math:`>\texttt{threshold}` ).
+
+ :param doSort: Performs lines sort by votes.
+
+ :param maxLines: Maximum number of output lines.
+
+ :param buf: Optional buffer to avoid extra memory allocations (for many calls with the same sizes).
+
+.. seealso:: :ocv:func:`HoughLines`
+
+
+
+gpu::HoughLinesDownload
+-----------------------
+Downloads results from :ocv:func:`gpu::HoughLines` to host memory.
+
+.. ocv:function:: void gpu::HoughLinesDownload(const GpuMat& d_lines, OutputArray h_lines, OutputArray h_votes = noArray())
+
+ :param d_lines: Result of :ocv:func:`gpu::HoughLines` .
+
+ :param h_lines: Output host array.
+
+ :param h_votes: Optional output array for line's votes.
+
+.. seealso:: :ocv:func:`gpu::HoughLines`
+
+
+
+gpu::HoughCircles
+-----------------
+Finds circles in a grayscale image using the Hough transform.
+
+.. ocv:function:: void gpu::HoughCircles(const GpuMat& src, GpuMat& circles, int method, float dp, float minDist, int cannyThreshold, int votesThreshold, int minRadius, int maxRadius, int maxCircles = 4096)
+
+.. ocv:function:: void gpu::HoughCircles(const GpuMat& src, GpuMat& circles, HoughCirclesBuf& buf, int method, float dp, float minDist, int cannyThreshold, int votesThreshold, int minRadius, int maxRadius, int maxCircles = 4096)
+
+ :param src: 8-bit, single-channel grayscale input image.
+
+ :param circles: Output vector of found circles. Each vector is encoded as a 3-element floating-point vector :math:`(x, y, radius)` .
+
+ :param method: Detection method to use. Currently, the only implemented method is ``CV_HOUGH_GRADIENT`` , which is basically *21HT* , described in [Yuen90]_.
+
+ :param dp: Inverse ratio of the accumulator resolution to the image resolution. For example, if ``dp=1`` , the accumulator has the same resolution as the input image. If ``dp=2`` , the accumulator has half as big width and height.
+
+ :param minDist: Minimum distance between the centers of the detected circles. If the parameter is too small, multiple neighbor circles may be falsely detected in addition to a true one. If it is too large, some circles may be missed.
+
+ :param cannyThreshold: The higher threshold of the two passed to the :ocv:func:`gpu::Canny` edge detector (the lower one is twice smaller).
+
+ :param votesThreshold: The accumulator threshold for the circle centers at the detection stage. The smaller it is, the more false circles may be detected.
+
+ :param minRadius: Minimum circle radius.
+
+ :param maxRadius: Maximum circle radius.
+
+ :param maxCircles: Maximum number of output circles.
+
+ :param buf: Optional buffer to avoid extra memory allocations (for many calls with the same sizes).
+
+.. seealso:: :ocv:func:`HoughCircles`
+
+
+
+gpu::HoughCirclesDownload
+-------------------------
+Downloads results from :ocv:func:`gpu::HoughCircles` to host memory.
+
+.. ocv:function:: void gpu::HoughCirclesDownload(const GpuMat& d_circles, OutputArray h_circles)
+
+ :param d_circles: Result of :ocv:func:`gpu::HoughCircles` .
+
+ :param h_circles: Output host array.
+
+.. seealso:: :ocv:func:`gpu::HoughCircles`
--- /dev/null
+Image Processing
+================
+
+.. highlight:: cpp
+
+
+
+gpu::meanShiftFiltering
+---------------------------
+Performs mean-shift filtering for each point of the source image.
+
+.. ocv:function:: void gpu::meanShiftFiltering( const GpuMat& src, GpuMat& dst, int sp, int sr, TermCriteria criteria=TermCriteria(TermCriteria::MAX_ITER + TermCriteria::EPS, 5, 1), Stream& stream=Stream::Null() )
+
+ :param src: Source image. Only ``CV_8UC4`` images are supported for now.
+
+ :param dst: Destination image containing the color of mapped points. It has the same size and type as ``src`` .
+
+ :param sp: Spatial window radius.
+
+ :param sr: Color window radius.
+
+ :param criteria: Termination criteria. See :ocv:class:`TermCriteria`.
+
+It maps each point of the source image into another point. As a result, you have a new color and new position of each point.
+
+
+
+gpu::meanShiftProc
+----------------------
+Performs a mean-shift procedure and stores information about processed points (their colors and positions) in two images.
+
+.. ocv:function:: void gpu::meanShiftProc( const GpuMat& src, GpuMat& dstr, GpuMat& dstsp, int sp, int sr, TermCriteria criteria=TermCriteria(TermCriteria::MAX_ITER + TermCriteria::EPS, 5, 1), Stream& stream=Stream::Null() )
+
+ :param src: Source image. Only ``CV_8UC4`` images are supported for now.
+
+ :param dstr: Destination image containing the color of mapped points. The size and type is the same as ``src`` .
+
+ :param dstsp: Destination image containing the position of mapped points. The size is the same as ``src`` size. The type is ``CV_16SC2`` .
+
+ :param sp: Spatial window radius.
+
+ :param sr: Color window radius.
+
+ :param criteria: Termination criteria. See :ocv:class:`TermCriteria`.
+
+.. seealso:: :ocv:func:`gpu::meanShiftFiltering`
+
+
+
+gpu::meanShiftSegmentation
+------------------------------
+Performs a mean-shift segmentation of the source image and eliminates small segments.
+
+.. ocv:function:: void gpu::meanShiftSegmentation(const GpuMat& src, Mat& dst, int sp, int sr, int minsize, TermCriteria criteria = TermCriteria(TermCriteria::MAX_ITER + TermCriteria::EPS, 5, 1))
+
+ :param src: Source image. Only ``CV_8UC4`` images are supported for now.
+
+ :param dst: Segmented image with the same size and type as ``src`` .
+
+ :param sp: Spatial window radius.
+
+ :param sr: Color window radius.
+
+ :param minsize: Minimum segment size. Smaller segments are merged.
+
+ :param criteria: Termination criteria. See :ocv:class:`TermCriteria`.
+
+
+
+gpu::MatchTemplateBuf
+---------------------
+.. ocv:struct:: gpu::MatchTemplateBuf
+
+Class providing memory buffers for :ocv:func:`gpu::matchTemplate` function, plus it allows to adjust some specific parameters. ::
+
+ struct CV_EXPORTS MatchTemplateBuf
+ {
+ Size user_block_size;
+ GpuMat imagef, templf;
+ std::vector<GpuMat> images;
+ std::vector<GpuMat> image_sums;
+ std::vector<GpuMat> image_sqsums;
+ };
+
+You can use field `user_block_size` to set specific block size for :ocv:func:`gpu::matchTemplate` function. If you leave its default value `Size(0,0)` then automatic estimation of block size will be used (which is optimized for speed). By varying `user_block_size` you can reduce memory requirements at the cost of speed.
+
+
+
+gpu::matchTemplate
+----------------------
+Computes a proximity map for a raster template and an image where the template is searched for.
+
+.. ocv:function:: void gpu::matchTemplate(const GpuMat& image, const GpuMat& templ, GpuMat& result, int method, Stream &stream = Stream::Null())
+
+.. ocv:function:: void gpu::matchTemplate(const GpuMat& image, const GpuMat& templ, GpuMat& result, int method, MatchTemplateBuf &buf, Stream& stream = Stream::Null())
+
+ :param image: Source image. ``CV_32F`` and ``CV_8U`` depth images (1..4 channels) are supported for now.
+
+ :param templ: Template image with the size and type the same as ``image`` .
+
+ :param result: Map containing comparison results ( ``CV_32FC1`` ). If ``image`` is *W x H* and ``templ`` is *w x h*, then ``result`` must be *W-w+1 x H-h+1*.
+
+ :param method: Specifies the way to compare the template with the image.
+
+ :param buf: Optional buffer to avoid extra memory allocations and to adjust some specific parameters. See :ocv:struct:`gpu::MatchTemplateBuf`.
+
+ :param stream: Stream for the asynchronous version.
+
+ The following methods are supported for the ``CV_8U`` depth images for now:
+
+ * ``CV_TM_SQDIFF``
+ * ``CV_TM_SQDIFF_NORMED``
+ * ``CV_TM_CCORR``
+ * ``CV_TM_CCORR_NORMED``
+ * ``CV_TM_CCOEFF``
+ * ``CV_TM_CCOEFF_NORMED``
+
+ The following methods are supported for the ``CV_32F`` images for now:
+
+ * ``CV_TM_SQDIFF``
+ * ``CV_TM_CCORR``
+
+.. seealso:: :ocv:func:`matchTemplate`
+
+
+
+gpu::Canny
+-------------------
+Finds edges in an image using the [Canny86]_ algorithm.
+
+.. ocv:function:: void gpu::Canny(const GpuMat& image, GpuMat& edges, double low_thresh, double high_thresh, int apperture_size = 3, bool L2gradient = false)
+
+.. ocv:function:: void gpu::Canny(const GpuMat& image, CannyBuf& buf, GpuMat& edges, double low_thresh, double high_thresh, int apperture_size = 3, bool L2gradient = false)
+
+.. ocv:function:: void gpu::Canny(const GpuMat& dx, const GpuMat& dy, GpuMat& edges, double low_thresh, double high_thresh, bool L2gradient = false)
+
+.. ocv:function:: void gpu::Canny(const GpuMat& dx, const GpuMat& dy, CannyBuf& buf, GpuMat& edges, double low_thresh, double high_thresh, bool L2gradient = false)
+
+ :param image: Single-channel 8-bit input image.
+
+ :param dx: First derivative of image in the vertical direction. Support only ``CV_32S`` type.
+
+ :param dy: First derivative of image in the horizontal direction. Support only ``CV_32S`` type.
+
+ :param edges: Output edge map. It has the same size and type as ``image`` .
+
+ :param low_thresh: First threshold for the hysteresis procedure.
+
+ :param high_thresh: Second threshold for the hysteresis procedure.
+
+ :param apperture_size: Aperture size for the :ocv:func:`Sobel` operator.
+
+ :param L2gradient: Flag indicating whether a more accurate :math:`L_2` norm :math:`=\sqrt{(dI/dx)^2 + (dI/dy)^2}` should be used to compute the image gradient magnitude ( ``L2gradient=true`` ), or a faster default :math:`L_1` norm :math:`=|dI/dx|+|dI/dy|` is enough ( ``L2gradient=false`` ).
+
+ :param buf: Optional buffer to avoid extra memory allocations (for many calls with the same sizes).
+
+.. seealso:: :ocv:func:`Canny`
+
+
+
+gpu::bilateralFilter
+--------------------
+Performs bilateral filtering of passed image
+
+.. ocv:function:: void gpu::bilateralFilter( const GpuMat& src, GpuMat& dst, int kernel_size, float sigma_color, float sigma_spatial, int borderMode=BORDER_DEFAULT, Stream& stream=Stream::Null() )
+
+ :param src: Source image. Supports only (channles != 2 && depth() != CV_8S && depth() != CV_32S && depth() != CV_64F).
+
+ :param dst: Destination imagwe.
+
+ :param kernel_size: Kernel window size.
+
+ :param sigma_color: Filter sigma in the color space.
+
+ :param sigma_spatial: Filter sigma in the coordinate space.
+
+ :param borderMode: Border type. See :ocv:func:`borderInterpolate` for details. ``BORDER_REFLECT101`` , ``BORDER_REPLICATE`` , ``BORDER_CONSTANT`` , ``BORDER_REFLECT`` and ``BORDER_WRAP`` are supported for now.
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso::
+
+ :ocv:func:`bilateralFilter`
+
+
+
+gpu::blendLinear
+-------------------
+Performs linear blending of two images.
+
+.. ocv:function:: void gpu::blendLinear(const GpuMat& img1, const GpuMat& img2, const GpuMat& weights1, const GpuMat& weights2, GpuMat& result, Stream& stream = Stream::Null())
+
+ :param img1: First image. Supports only ``CV_8U`` and ``CV_32F`` depth.
+
+ :param img2: Second image. Must have the same size and the same type as ``img1`` .
+
+ :param weights1: Weights for first image. Must have tha same size as ``img1`` . Supports only ``CV_32F`` type.
+
+ :param weights2: Weights for second image. Must have tha same size as ``img2`` . Supports only ``CV_32F`` type.
+
+ :param result: Destination image.
+
+ :param stream: Stream for the asynchronous version.
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_GPUIMGPROC_HPP__
+#define __OPENCV_GPUIMGPROC_HPP__
+
+#ifndef __cplusplus
+# error gpuimgproc.hpp header must be compiled as C++
+#endif
+
+#include "opencv2/core/gpumat.hpp"
+#include "opencv2/core/base.hpp"
+#include "opencv2/imgproc.hpp"
+#include "opencv2/gpufilters.hpp"
+
+namespace cv { namespace gpu {
+
+/////////////////////////// Color Processing ///////////////////////////
+
+//! converts image from one color space to another
+CV_EXPORTS void cvtColor(const GpuMat& src, GpuMat& dst, int code, int dcn = 0, Stream& stream = Stream::Null());
+
+enum
+{
+ // Bayer Demosaicing (Malvar, He, and Cutler)
+ COLOR_BayerBG2BGR_MHT = 256,
+ COLOR_BayerGB2BGR_MHT = 257,
+ COLOR_BayerRG2BGR_MHT = 258,
+ COLOR_BayerGR2BGR_MHT = 259,
+
+ COLOR_BayerBG2RGB_MHT = COLOR_BayerRG2BGR_MHT,
+ COLOR_BayerGB2RGB_MHT = COLOR_BayerGR2BGR_MHT,
+ COLOR_BayerRG2RGB_MHT = COLOR_BayerBG2BGR_MHT,
+ COLOR_BayerGR2RGB_MHT = COLOR_BayerGB2BGR_MHT,
+
+ COLOR_BayerBG2GRAY_MHT = 260,
+ COLOR_BayerGB2GRAY_MHT = 261,
+ COLOR_BayerRG2GRAY_MHT = 262,
+ COLOR_BayerGR2GRAY_MHT = 263
+};
+CV_EXPORTS void demosaicing(const GpuMat& src, GpuMat& dst, int code, int dcn = -1, Stream& stream = Stream::Null());
+
+//! swap channels
+//! dstOrder - Integer array describing how channel values are permutated. The n-th entry
+//! of the array contains the number of the channel that is stored in the n-th channel of
+//! the output image. E.g. Given an RGBA image, aDstOrder = [3,2,1,0] converts this to ABGR
+//! channel order.
+CV_EXPORTS void swapChannels(GpuMat& image, const int dstOrder[4], Stream& stream = Stream::Null());
+
+//! Routines for correcting image color gamma
+CV_EXPORTS void gammaCorrection(const GpuMat& src, GpuMat& dst, bool forward = true, Stream& stream = Stream::Null());
+
+enum { ALPHA_OVER, ALPHA_IN, ALPHA_OUT, ALPHA_ATOP, ALPHA_XOR, ALPHA_PLUS, ALPHA_OVER_PREMUL, ALPHA_IN_PREMUL, ALPHA_OUT_PREMUL,
+ ALPHA_ATOP_PREMUL, ALPHA_XOR_PREMUL, ALPHA_PLUS_PREMUL, ALPHA_PREMUL};
+
+//! Composite two images using alpha opacity values contained in each image
+//! Supports CV_8UC4, CV_16UC4, CV_32SC4 and CV_32FC4 types
+CV_EXPORTS void alphaComp(const GpuMat& img1, const GpuMat& img2, GpuMat& dst, int alpha_op, Stream& stream = Stream::Null());
+
+////////////////////////////// Histogram ///////////////////////////////
+
+//! Compute levels with even distribution. levels will have 1 row and nLevels cols and CV_32SC1 type.
+CV_EXPORTS void evenLevels(GpuMat& levels, int nLevels, int lowerLevel, int upperLevel);
+
+//! Calculates histogram with evenly distributed bins for signle channel source.
+//! Supports CV_8UC1, CV_16UC1 and CV_16SC1 source types.
+//! Output hist will have one row and histSize cols and CV_32SC1 type.
+CV_EXPORTS void histEven(const GpuMat& src, GpuMat& hist, int histSize, int lowerLevel, int upperLevel, Stream& stream = Stream::Null());
+CV_EXPORTS void histEven(const GpuMat& src, GpuMat& hist, GpuMat& buf, int histSize, int lowerLevel, int upperLevel, Stream& stream = Stream::Null());
+
+//! Calculates histogram with evenly distributed bins for four-channel source.
+//! All channels of source are processed separately.
+//! Supports CV_8UC4, CV_16UC4 and CV_16SC4 source types.
+//! Output hist[i] will have one row and histSize[i] cols and CV_32SC1 type.
+CV_EXPORTS void histEven(const GpuMat& src, GpuMat hist[4], int histSize[4], int lowerLevel[4], int upperLevel[4], Stream& stream = Stream::Null());
+CV_EXPORTS void histEven(const GpuMat& src, GpuMat hist[4], GpuMat& buf, int histSize[4], int lowerLevel[4], int upperLevel[4], Stream& stream = Stream::Null());
+
+//! Calculates histogram with bins determined by levels array.
+//! levels must have one row and CV_32SC1 type if source has integer type or CV_32FC1 otherwise.
+//! Supports CV_8UC1, CV_16UC1, CV_16SC1 and CV_32FC1 source types.
+//! Output hist will have one row and (levels.cols-1) cols and CV_32SC1 type.
+CV_EXPORTS void histRange(const GpuMat& src, GpuMat& hist, const GpuMat& levels, Stream& stream = Stream::Null());
+CV_EXPORTS void histRange(const GpuMat& src, GpuMat& hist, const GpuMat& levels, GpuMat& buf, Stream& stream = Stream::Null());
+
+//! Calculates histogram with bins determined by levels array.
+//! All levels must have one row and CV_32SC1 type if source has integer type or CV_32FC1 otherwise.
+//! All channels of source are processed separately.
+//! Supports CV_8UC4, CV_16UC4, CV_16SC4 and CV_32FC4 source types.
+//! Output hist[i] will have one row and (levels[i].cols-1) cols and CV_32SC1 type.
+CV_EXPORTS void histRange(const GpuMat& src, GpuMat hist[4], const GpuMat levels[4], Stream& stream = Stream::Null());
+CV_EXPORTS void histRange(const GpuMat& src, GpuMat hist[4], const GpuMat levels[4], GpuMat& buf, Stream& stream = Stream::Null());
+
+//! Calculates histogram for 8u one channel image
+//! Output hist will have one row, 256 cols and CV32SC1 type.
+CV_EXPORTS void calcHist(const GpuMat& src, GpuMat& hist, Stream& stream = Stream::Null());
+CV_EXPORTS void calcHist(const GpuMat& src, GpuMat& hist, GpuMat& buf, Stream& stream = Stream::Null());
+
+//! normalizes the grayscale image brightness and contrast by normalizing its histogram
+CV_EXPORTS void equalizeHist(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null());
+CV_EXPORTS void equalizeHist(const GpuMat& src, GpuMat& dst, GpuMat& hist, Stream& stream = Stream::Null());
+CV_EXPORTS void equalizeHist(const GpuMat& src, GpuMat& dst, GpuMat& hist, GpuMat& buf, Stream& stream = Stream::Null());
+
+class CV_EXPORTS CLAHE : public cv::CLAHE
+{
+public:
+ using cv::CLAHE::apply;
+ virtual void apply(InputArray src, OutputArray dst, Stream& stream) = 0;
+};
+CV_EXPORTS Ptr<cv::gpu::CLAHE> createCLAHE(double clipLimit = 40.0, Size tileGridSize = Size(8, 8));
+
+//////////////////////////////// Canny ////////////////////////////////
+
+struct CV_EXPORTS CannyBuf
+{
+ void create(const Size& image_size, int apperture_size = 3);
+ void release();
+
+ GpuMat dx, dy;
+ GpuMat mag;
+ GpuMat map;
+ GpuMat st1, st2;
+ Ptr<FilterEngine_GPU> filterDX, filterDY;
+};
+
+CV_EXPORTS void Canny(const GpuMat& image, GpuMat& edges, double low_thresh, double high_thresh, int apperture_size = 3, bool L2gradient = false);
+CV_EXPORTS void Canny(const GpuMat& image, CannyBuf& buf, GpuMat& edges, double low_thresh, double high_thresh, int apperture_size = 3, bool L2gradient = false);
+CV_EXPORTS void Canny(const GpuMat& dx, const GpuMat& dy, GpuMat& edges, double low_thresh, double high_thresh, bool L2gradient = false);
+CV_EXPORTS void Canny(const GpuMat& dx, const GpuMat& dy, CannyBuf& buf, GpuMat& edges, double low_thresh, double high_thresh, bool L2gradient = false);
+
+/////////////////////////// Hough Transform ////////////////////////////
+
+struct HoughLinesBuf
+{
+ GpuMat accum;
+ GpuMat list;
+};
+
+CV_EXPORTS void HoughLines(const GpuMat& src, GpuMat& lines, float rho, float theta, int threshold, bool doSort = false, int maxLines = 4096);
+CV_EXPORTS void HoughLines(const GpuMat& src, GpuMat& lines, HoughLinesBuf& buf, float rho, float theta, int threshold, bool doSort = false, int maxLines = 4096);
+CV_EXPORTS void HoughLinesDownload(const GpuMat& d_lines, OutputArray h_lines, OutputArray h_votes = noArray());
+
+//! finds line segments in the black-n-white image using probabalistic Hough transform
+CV_EXPORTS void HoughLinesP(const GpuMat& image, GpuMat& lines, HoughLinesBuf& buf, float rho, float theta, int minLineLength, int maxLineGap, int maxLines = 4096);
+
+struct HoughCirclesBuf
+{
+ GpuMat edges;
+ GpuMat accum;
+ GpuMat list;
+ CannyBuf cannyBuf;
+};
+
+CV_EXPORTS void HoughCircles(const GpuMat& src, GpuMat& circles, int method, float dp, float minDist, int cannyThreshold, int votesThreshold, int minRadius, int maxRadius, int maxCircles = 4096);
+CV_EXPORTS void HoughCircles(const GpuMat& src, GpuMat& circles, HoughCirclesBuf& buf, int method, float dp, float minDist, int cannyThreshold, int votesThreshold, int minRadius, int maxRadius, int maxCircles = 4096);
+CV_EXPORTS void HoughCirclesDownload(const GpuMat& d_circles, OutputArray h_circles);
+
+//! finds arbitrary template in the grayscale image using Generalized Hough Transform
+//! Ballard, D.H. (1981). Generalizing the Hough transform to detect arbitrary shapes. Pattern Recognition 13 (2): 111-122.
+//! Guil, N., González-Linares, J.M. and Zapata, E.L. (1999). Bidimensional shape detection using an invariant approach. Pattern Recognition 32 (6): 1025-1038.
+class CV_EXPORTS GeneralizedHough_GPU : public cv::Algorithm
+{
+public:
+ static Ptr<GeneralizedHough_GPU> create(int method);
+
+ virtual ~GeneralizedHough_GPU();
+
+ //! set template to search
+ void setTemplate(const GpuMat& templ, int cannyThreshold = 100, Point templCenter = Point(-1, -1));
+ void setTemplate(const GpuMat& edges, const GpuMat& dx, const GpuMat& dy, Point templCenter = Point(-1, -1));
+
+ //! find template on image
+ void detect(const GpuMat& image, GpuMat& positions, int cannyThreshold = 100);
+ void detect(const GpuMat& edges, const GpuMat& dx, const GpuMat& dy, GpuMat& positions);
+
+ void download(const GpuMat& d_positions, OutputArray h_positions, OutputArray h_votes = noArray());
+
+ void release();
+
+protected:
+ virtual void setTemplateImpl(const GpuMat& edges, const GpuMat& dx, const GpuMat& dy, Point templCenter) = 0;
+ virtual void detectImpl(const GpuMat& edges, const GpuMat& dx, const GpuMat& dy, GpuMat& positions) = 0;
+ virtual void releaseImpl() = 0;
+
+private:
+ GpuMat edges_;
+ CannyBuf cannyBuf_;
+};
+
+////////////////////////// Corners Detection ///////////////////////////
+
+//! computes Harris cornerness criteria at each image pixel
+CV_EXPORTS void cornerHarris(const GpuMat& src, GpuMat& dst, int blockSize, int ksize, double k, int borderType = BORDER_REFLECT101);
+CV_EXPORTS void cornerHarris(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, int blockSize, int ksize, double k, int borderType = BORDER_REFLECT101);
+CV_EXPORTS void cornerHarris(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, GpuMat& buf, int blockSize, int ksize, double k,
+ int borderType = BORDER_REFLECT101, Stream& stream = Stream::Null());
+
+//! computes minimum eigen value of 2x2 derivative covariation matrix at each pixel - the cornerness criteria
+CV_EXPORTS void cornerMinEigenVal(const GpuMat& src, GpuMat& dst, int blockSize, int ksize, int borderType=BORDER_REFLECT101);
+CV_EXPORTS void cornerMinEigenVal(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, int blockSize, int ksize, int borderType=BORDER_REFLECT101);
+CV_EXPORTS void cornerMinEigenVal(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, GpuMat& buf, int blockSize, int ksize,
+ int borderType=BORDER_REFLECT101, Stream& stream = Stream::Null());
+
+////////////////////////// Feature Detection ///////////////////////////
+
+class CV_EXPORTS GoodFeaturesToTrackDetector_GPU
+{
+public:
+ explicit GoodFeaturesToTrackDetector_GPU(int maxCorners = 1000, double qualityLevel = 0.01, double minDistance = 0.0,
+ int blockSize = 3, bool useHarrisDetector = false, double harrisK = 0.04);
+
+ //! return 1 rows matrix with CV_32FC2 type
+ void operator ()(const GpuMat& image, GpuMat& corners, const GpuMat& mask = GpuMat());
+
+ int maxCorners;
+ double qualityLevel;
+ double minDistance;
+
+ int blockSize;
+ bool useHarrisDetector;
+ double harrisK;
+
+ void releaseMemory()
+ {
+ Dx_.release();
+ Dy_.release();
+ buf_.release();
+ eig_.release();
+ minMaxbuf_.release();
+ tmpCorners_.release();
+ }
+
+private:
+ GpuMat Dx_;
+ GpuMat Dy_;
+ GpuMat buf_;
+ GpuMat eig_;
+ GpuMat minMaxbuf_;
+ GpuMat tmpCorners_;
+};
+
+inline GoodFeaturesToTrackDetector_GPU::GoodFeaturesToTrackDetector_GPU(int maxCorners_, double qualityLevel_, double minDistance_,
+ int blockSize_, bool useHarrisDetector_, double harrisK_)
+{
+ maxCorners = maxCorners_;
+ qualityLevel = qualityLevel_;
+ minDistance = minDistance_;
+ blockSize = blockSize_;
+ useHarrisDetector = useHarrisDetector_;
+ harrisK = harrisK_;
+}
+
+///////////////////////////// Mean Shift //////////////////////////////
+
+//! Does mean shift filtering on GPU.
+CV_EXPORTS void meanShiftFiltering(const GpuMat& src, GpuMat& dst, int sp, int sr,
+ TermCriteria criteria = TermCriteria(TermCriteria::MAX_ITER + TermCriteria::EPS, 5, 1),
+ Stream& stream = Stream::Null());
+
+//! Does mean shift procedure on GPU.
+CV_EXPORTS void meanShiftProc(const GpuMat& src, GpuMat& dstr, GpuMat& dstsp, int sp, int sr,
+ TermCriteria criteria = TermCriteria(TermCriteria::MAX_ITER + TermCriteria::EPS, 5, 1),
+ Stream& stream = Stream::Null());
+
+//! Does mean shift segmentation with elimination of small regions.
+CV_EXPORTS void meanShiftSegmentation(const GpuMat& src, Mat& dst, int sp, int sr, int minsize,
+ TermCriteria criteria = TermCriteria(TermCriteria::MAX_ITER + TermCriteria::EPS, 5, 1));
+
+/////////////////////////// Match Template ////////////////////////////
+
+struct CV_EXPORTS MatchTemplateBuf
+{
+ Size user_block_size;
+ GpuMat imagef, templf;
+ std::vector<GpuMat> images;
+ std::vector<GpuMat> image_sums;
+ std::vector<GpuMat> image_sqsums;
+};
+
+//! computes the proximity map for the raster template and the image where the template is searched for
+CV_EXPORTS void matchTemplate(const GpuMat& image, const GpuMat& templ, GpuMat& result, int method, Stream &stream = Stream::Null());
+
+//! computes the proximity map for the raster template and the image where the template is searched for
+CV_EXPORTS void matchTemplate(const GpuMat& image, const GpuMat& templ, GpuMat& result, int method, MatchTemplateBuf &buf, Stream& stream = Stream::Null());
+
+////////////////////////// Bilateral Filter ///////////////////////////
+
+//! Performa bilateral filtering of passsed image
+CV_EXPORTS void bilateralFilter(const GpuMat& src, GpuMat& dst, int kernel_size, float sigma_color, float sigma_spatial,
+ int borderMode = BORDER_DEFAULT, Stream& stream = Stream::Null());
+
+///////////////////////////// Blending ////////////////////////////////
+
+//! performs linear blending of two images
+//! to avoid accuracy errors sum of weigths shouldn't be very close to zero
+CV_EXPORTS void blendLinear(const GpuMat& img1, const GpuMat& img2, const GpuMat& weights1, const GpuMat& weights2,
+ GpuMat& result, Stream& stream = Stream::Null());
+
+}} // namespace cv { namespace gpu {
+
+#endif /* __OPENCV_GPUIMGPROC_HPP__ */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace std;
+using namespace testing;
+using namespace perf;
+
+//////////////////////////////////////////////////////////////////////
+// BilateralFilter
+
+DEF_PARAM_TEST(Sz_Depth_Cn_KernelSz, cv::Size, MatDepth, MatCn, int);
+
+PERF_TEST_P(Sz_Depth_Cn_KernelSz, BilateralFilter,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_32F),
+ GPU_CHANNELS_1_3,
+ Values(3, 5, 9)))
+{
+ declare.time(60.0);
+
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+ const int kernel_size = GET_PARAM(3);
+
+ const float sigma_color = 7;
+ const float sigma_spatial = 5;
+ const int borderMode = cv::BORDER_REFLECT101;
+
+ const int type = CV_MAKE_TYPE(depth, channels);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::bilateralFilter(d_src, dst, kernel_size, sigma_color, sigma_spatial, borderMode);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::bilateralFilter(src, dst, kernel_size, sigma_color, sigma_spatial, borderMode);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace std;
+using namespace testing;
+using namespace perf;
+
+//////////////////////////////////////////////////////////////////////
+// BlendLinear
+
+PERF_TEST_P(Sz_Depth_Cn, BlendLinear,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_32F),
+ GPU_CHANNELS_1_3_4))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+
+ const int type = CV_MAKE_TYPE(depth, channels);
+
+ cv::Mat img1(size, type);
+ cv::Mat img2(size, type);
+ declare.in(img1, img2, WARMUP_RNG);
+
+ const cv::Mat weights1(size, CV_32FC1, cv::Scalar::all(0.5));
+ const cv::Mat weights2(size, CV_32FC1, cv::Scalar::all(0.5));
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_img1(img1);
+ const cv::gpu::GpuMat d_img2(img2);
+ const cv::gpu::GpuMat d_weights1(weights1);
+ const cv::gpu::GpuMat d_weights2(weights2);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::blendLinear(d_img1, d_img2, d_weights1, d_weights2, dst);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace std;
+using namespace testing;
+using namespace perf;
+
+//////////////////////////////////////////////////////////////////////
+// Canny
+
+DEF_PARAM_TEST(Image_AppertureSz_L2gradient, string, int, bool);
+
+PERF_TEST_P(Image_AppertureSz_L2gradient, Canny,
+ Combine(Values("perf/800x600.png", "perf/1280x1024.png", "perf/1680x1050.png"),
+ Values(3, 5),
+ Bool()))
+{
+ const string fileName = GET_PARAM(0);
+ const int apperture_size = GET_PARAM(1);
+ const bool useL2gradient = GET_PARAM(2);
+
+ const cv::Mat image = readImage(fileName, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(image.empty());
+
+ const double low_thresh = 50.0;
+ const double high_thresh = 100.0;
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_image(image);
+ cv::gpu::GpuMat dst;
+ cv::gpu::CannyBuf d_buf;
+
+ TEST_CYCLE() cv::gpu::Canny(d_image, d_buf, dst, low_thresh, high_thresh, apperture_size, useL2gradient);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::Canny(image, dst, low_thresh, high_thresh, apperture_size, useL2gradient);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace std;
+using namespace testing;
+using namespace perf;
+
+//////////////////////////////////////////////////////////////////////
+// CvtColor
+
+DEF_PARAM_TEST(Sz_Depth_Code, cv::Size, MatDepth, CvtColorInfo);
+
+PERF_TEST_P(Sz_Depth_Code, CvtColor,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_32F),
+ Values(CvtColorInfo(4, 4, cv::COLOR_RGBA2BGRA),
+ CvtColorInfo(4, 1, cv::COLOR_BGRA2GRAY),
+ CvtColorInfo(1, 4, cv::COLOR_GRAY2BGRA),
+ CvtColorInfo(3, 3, cv::COLOR_BGR2XYZ),
+ CvtColorInfo(3, 3, cv::COLOR_XYZ2BGR),
+ CvtColorInfo(3, 3, cv::COLOR_BGR2YCrCb),
+ CvtColorInfo(3, 3, cv::COLOR_YCrCb2BGR),
+ CvtColorInfo(3, 3, cv::COLOR_BGR2YUV),
+ CvtColorInfo(3, 3, cv::COLOR_YUV2BGR),
+ CvtColorInfo(3, 3, cv::COLOR_BGR2HSV),
+ CvtColorInfo(3, 3, cv::COLOR_HSV2BGR),
+ CvtColorInfo(3, 3, cv::COLOR_BGR2HLS),
+ CvtColorInfo(3, 3, cv::COLOR_HLS2BGR),
+ CvtColorInfo(3, 3, cv::COLOR_BGR2Lab),
+ CvtColorInfo(3, 3, cv::COLOR_LBGR2Lab),
+ CvtColorInfo(3, 3, cv::COLOR_BGR2Luv),
+ CvtColorInfo(3, 3, cv::COLOR_LBGR2Luv),
+ CvtColorInfo(3, 3, cv::COLOR_Lab2BGR),
+ CvtColorInfo(3, 3, cv::COLOR_Lab2LBGR),
+ CvtColorInfo(3, 3, cv::COLOR_Luv2RGB),
+ CvtColorInfo(3, 3, cv::COLOR_Luv2LRGB))))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const CvtColorInfo info = GET_PARAM(2);
+
+ cv::Mat src(size, CV_MAKETYPE(depth, info.scn));
+ cv::randu(src, 0, depth == CV_8U ? 255.0 : 1.0);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::cvtColor(d_src, dst, info.code, info.dcn);
+
+ GPU_SANITY_CHECK(dst, 1e-4);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::cvtColor(src, dst, info.code, info.dcn);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+PERF_TEST_P(Sz_Depth_Code, CvtColorBayer,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U),
+ Values(CvtColorInfo(1, 3, cv::COLOR_BayerBG2BGR),
+ CvtColorInfo(1, 3, cv::COLOR_BayerGB2BGR),
+ CvtColorInfo(1, 3, cv::COLOR_BayerRG2BGR),
+ CvtColorInfo(1, 3, cv::COLOR_BayerGR2BGR),
+
+ CvtColorInfo(1, 1, cv::COLOR_BayerBG2GRAY),
+ CvtColorInfo(1, 1, cv::COLOR_BayerGB2GRAY),
+ CvtColorInfo(1, 1, cv::COLOR_BayerRG2GRAY),
+ CvtColorInfo(1, 1, cv::COLOR_BayerGR2GRAY))))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const CvtColorInfo info = GET_PARAM(2);
+
+ cv::Mat src(size, CV_MAKETYPE(depth, info.scn));
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::cvtColor(d_src, dst, info.code, info.dcn);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::cvtColor(src, dst, info.code, info.dcn);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// Demosaicing
+
+CV_ENUM(DemosaicingCode,
+ cv::COLOR_BayerBG2BGR, cv::COLOR_BayerGB2BGR, cv::COLOR_BayerRG2BGR, cv::COLOR_BayerGR2BGR,
+ cv::COLOR_BayerBG2GRAY, cv::COLOR_BayerGB2GRAY, cv::COLOR_BayerRG2GRAY, cv::COLOR_BayerGR2GRAY,
+ cv::gpu::COLOR_BayerBG2BGR_MHT, cv::gpu::COLOR_BayerGB2BGR_MHT, cv::gpu::COLOR_BayerRG2BGR_MHT, cv::gpu::COLOR_BayerGR2BGR_MHT,
+ cv::gpu::COLOR_BayerBG2GRAY_MHT, cv::gpu::COLOR_BayerGB2GRAY_MHT, cv::gpu::COLOR_BayerRG2GRAY_MHT, cv::gpu::COLOR_BayerGR2GRAY_MHT)
+
+DEF_PARAM_TEST(Sz_Code, cv::Size, DemosaicingCode);
+
+PERF_TEST_P(Sz_Code, Demosaicing,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ DemosaicingCode::all()))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int code = GET_PARAM(1);
+
+ cv::Mat src(size, CV_8UC1);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::demosaicing(d_src, dst, code);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ if (code >= cv::COLOR_COLORCVT_MAX)
+ {
+ FAIL_NO_CPU();
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::cvtColor(src, dst, code);
+
+ CPU_SANITY_CHECK(dst);
+ }
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// SwapChannels
+
+PERF_TEST_P(Sz, SwapChannels,
+ GPU_TYPICAL_MAT_SIZES)
+{
+ const cv::Size size = GetParam();
+
+ cv::Mat src(size, CV_8UC4);
+ declare.in(src, WARMUP_RNG);
+
+ const int dstOrder[] = {2, 1, 0, 3};
+
+ if (PERF_RUN_GPU())
+ {
+ cv::gpu::GpuMat dst(src);
+
+ TEST_CYCLE() cv::gpu::swapChannels(dst, dstOrder);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// AlphaComp
+
+CV_ENUM(AlphaOp, cv::gpu::ALPHA_OVER, cv::gpu::ALPHA_IN, cv::gpu::ALPHA_OUT, cv::gpu::ALPHA_ATOP, cv::gpu::ALPHA_XOR, cv::gpu::ALPHA_PLUS, cv::gpu::ALPHA_OVER_PREMUL, cv::gpu::ALPHA_IN_PREMUL, cv::gpu::ALPHA_OUT_PREMUL, cv::gpu::ALPHA_ATOP_PREMUL, cv::gpu::ALPHA_XOR_PREMUL, cv::gpu::ALPHA_PLUS_PREMUL, cv::gpu::ALPHA_PREMUL)
+
+DEF_PARAM_TEST(Sz_Type_Op, cv::Size, MatType, AlphaOp);
+
+PERF_TEST_P(Sz_Type_Op, AlphaComp,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8UC4, CV_16UC4, CV_32SC4, CV_32FC4),
+ AlphaOp::all()))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int type = GET_PARAM(1);
+ const int alpha_op = GET_PARAM(2);
+
+ cv::Mat img1(size, type);
+ cv::Mat img2(size, type);
+ declare.in(img1, img2, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_img1(img1);
+ const cv::gpu::GpuMat d_img2(img2);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::alphaComp(d_img1, d_img2, dst, alpha_op);
+
+ GPU_SANITY_CHECK(dst, 1e-3, ERROR_RELATIVE);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace std;
+using namespace testing;
+using namespace perf;
+
+//////////////////////////////////////////////////////////////////////
+// CornerHarris
+
+DEF_PARAM_TEST(Image_Type_Border_BlockSz_ApertureSz, string, MatType, BorderMode, int, int);
+
+PERF_TEST_P(Image_Type_Border_BlockSz_ApertureSz, CornerHarris,
+ Combine(Values<string>("gpu/stereobm/aloe-L.png"),
+ Values(CV_8UC1, CV_32FC1),
+ Values(BorderMode(cv::BORDER_REFLECT101), BorderMode(cv::BORDER_REPLICATE), BorderMode(cv::BORDER_REFLECT)),
+ Values(3, 5, 7),
+ Values(0, 3, 5, 7)))
+{
+ const string fileName = GET_PARAM(0);
+ const int type = GET_PARAM(1);
+ const int borderMode = GET_PARAM(2);
+ const int blockSize = GET_PARAM(3);
+ const int apertureSize = GET_PARAM(4);
+
+ cv::Mat img = readImage(fileName, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(img.empty());
+
+ img.convertTo(img, type, type == CV_32F ? 1.0 / 255.0 : 1.0);
+
+ const double k = 0.5;
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_img(img);
+ cv::gpu::GpuMat dst;
+ cv::gpu::GpuMat d_Dx;
+ cv::gpu::GpuMat d_Dy;
+ cv::gpu::GpuMat d_buf;
+
+ TEST_CYCLE() cv::gpu::cornerHarris(d_img, dst, d_Dx, d_Dy, d_buf, blockSize, apertureSize, k, borderMode);
+
+ GPU_SANITY_CHECK(dst, 1e-4);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::cornerHarris(img, dst, blockSize, apertureSize, k, borderMode);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// CornerMinEigenVal
+
+PERF_TEST_P(Image_Type_Border_BlockSz_ApertureSz, CornerMinEigenVal,
+ Combine(Values<string>("gpu/stereobm/aloe-L.png"),
+ Values(CV_8UC1, CV_32FC1),
+ Values(BorderMode(cv::BORDER_REFLECT101), BorderMode(cv::BORDER_REPLICATE), BorderMode(cv::BORDER_REFLECT)),
+ Values(3, 5, 7),
+ Values(0, 3, 5, 7)))
+{
+ const string fileName = GET_PARAM(0);
+ const int type = GET_PARAM(1);
+ const int borderMode = GET_PARAM(2);
+ const int blockSize = GET_PARAM(3);
+ const int apertureSize = GET_PARAM(4);
+
+ cv::Mat img = readImage(fileName, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(img.empty());
+
+ img.convertTo(img, type, type == CV_32F ? 1.0 / 255.0 : 1.0);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_img(img);
+ cv::gpu::GpuMat dst;
+ cv::gpu::GpuMat d_Dx;
+ cv::gpu::GpuMat d_Dy;
+ cv::gpu::GpuMat d_buf;
+
+ TEST_CYCLE() cv::gpu::cornerMinEigenVal(d_img, dst, d_Dx, d_Dy, d_buf, blockSize, apertureSize, borderMode);
+
+ GPU_SANITY_CHECK(dst, 1e-4);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::cornerMinEigenVal(img, dst, blockSize, apertureSize, borderMode);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
//
//M*/
-#ifndef __FFMPEG_VIDEO_SOURCE_H__
-#define __FFMPEG_VIDEO_SOURCE_H__
+#include "perf_precomp.hpp"
-#include "precomp.hpp"
-#include "thread_wrappers.h"
+using namespace std;
+using namespace testing;
+using namespace perf;
-#if defined(HAVE_CUDA) && defined(HAVE_NVCUVID)
+//////////////////////////////////////////////////////
+// GoodFeaturesToTrack
-struct InputMediaStream_FFMPEG;
+DEF_PARAM_TEST(Image_MinDistance, string, double);
-namespace cv { namespace gpu
+PERF_TEST_P(Image_MinDistance, GoodFeaturesToTrack,
+ Combine(Values<string>("gpu/perf/aloe.png"),
+ Values(0.0, 3.0)))
{
- namespace detail
- {
- class FFmpegVideoSource : public VideoReader_GPU::VideoSource
- {
- public:
- FFmpegVideoSource(const String& fname);
- ~FFmpegVideoSource();
+ const string fileName = GET_PARAM(0);
+ const double minDistance = GET_PARAM(1);
- VideoReader_GPU::FormatInfo format() const;
- void start();
- void stop();
- bool isStarted() const;
- bool hasError() const;
+ const cv::Mat image = readImage(fileName, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(image.empty());
- private:
- FFmpegVideoSource(const FFmpegVideoSource&);
- FFmpegVideoSource& operator =(const FFmpegVideoSource&);
+ const int maxCorners = 8000;
+ const double qualityLevel = 0.01;
- VideoReader_GPU::FormatInfo format_;
+ if (PERF_RUN_GPU())
+ {
+ cv::gpu::GoodFeaturesToTrackDetector_GPU d_detector(maxCorners, qualityLevel, minDistance);
- InputMediaStream_FFMPEG* stream_;
+ const cv::gpu::GpuMat d_image(image);
+ cv::gpu::GpuMat pts;
- std::auto_ptr<Thread> thread_;
- volatile bool stop_;
- volatile bool hasError_;
+ TEST_CYCLE() d_detector(d_image, pts);
- static void readLoop(void* userData);
- };
+ GPU_SANITY_CHECK(pts);
}
-}}
+ else
+ {
+ cv::Mat pts;
-#endif // HAVE_CUDA
+ TEST_CYCLE() cv::goodFeaturesToTrack(image, pts, maxCorners, qualityLevel, minDistance);
-#endif // __CUVUD_VIDEO_SOURCE_H__
+ CPU_SANITY_CHECK(pts);
+ }
+}
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace std;
+using namespace testing;
+using namespace perf;
+
+//////////////////////////////////////////////////////////////////////
+// HistEvenC1
+
+PERF_TEST_P(Sz_Depth, HistEvenC1,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_16S)))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+
+ cv::Mat src(size, depth);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+ cv::gpu::GpuMat d_buf;
+
+ TEST_CYCLE() cv::gpu::histEven(d_src, dst, d_buf, 30, 0, 180);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ const int hbins = 30;
+ const float hranges[] = {0.0f, 180.0f};
+ const int histSize[] = {hbins};
+ const float* ranges[] = {hranges};
+ const int channels[] = {0};
+
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::calcHist(&src, 1, channels, cv::Mat(), dst, 1, histSize, ranges);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// HistEvenC4
+
+PERF_TEST_P(Sz_Depth, HistEvenC4,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_16S)))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+
+ cv::Mat src(size, CV_MAKE_TYPE(depth, 4));
+ declare.in(src, WARMUP_RNG);
+
+ int histSize[] = {30, 30, 30, 30};
+ int lowerLevel[] = {0, 0, 0, 0};
+ int upperLevel[] = {180, 180, 180, 180};
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat d_hist[4];
+ cv::gpu::GpuMat d_buf;
+
+ TEST_CYCLE() cv::gpu::histEven(d_src, d_hist, d_buf, histSize, lowerLevel, upperLevel);
+
+ cv::Mat cpu_hist0, cpu_hist1, cpu_hist2, cpu_hist3;
+ d_hist[0].download(cpu_hist0);
+ d_hist[1].download(cpu_hist1);
+ d_hist[2].download(cpu_hist2);
+ d_hist[3].download(cpu_hist3);
+ SANITY_CHECK(cpu_hist0);
+ SANITY_CHECK(cpu_hist1);
+ SANITY_CHECK(cpu_hist2);
+ SANITY_CHECK(cpu_hist3);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// CalcHist
+
+PERF_TEST_P(Sz, CalcHist,
+ GPU_TYPICAL_MAT_SIZES)
+{
+ const cv::Size size = GetParam();
+
+ cv::Mat src(size, CV_8UC1);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::calcHist(d_src, dst);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// EqualizeHist
+
+PERF_TEST_P(Sz, EqualizeHist,
+ GPU_TYPICAL_MAT_SIZES)
+{
+ const cv::Size size = GetParam();
+
+ cv::Mat src(size, CV_8UC1);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+ cv::gpu::GpuMat d_hist;
+ cv::gpu::GpuMat d_buf;
+
+ TEST_CYCLE() cv::gpu::equalizeHist(d_src, dst, d_hist, d_buf);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::equalizeHist(src, dst);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// CLAHE
+
+DEF_PARAM_TEST(Sz_ClipLimit, cv::Size, double);
+
+PERF_TEST_P(Sz_ClipLimit, CLAHE,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(0.0, 40.0)))
+{
+ const cv::Size size = GET_PARAM(0);
+ const double clipLimit = GET_PARAM(1);
+
+ cv::Mat src(size, CV_8UC1);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ cv::Ptr<cv::gpu::CLAHE> clahe = cv::gpu::createCLAHE(clipLimit);
+ cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() clahe->apply(d_src, dst);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ cv::Ptr<cv::CLAHE> clahe = cv::createCLAHE(clipLimit);
+ cv::Mat dst;
+
+ TEST_CYCLE() clahe->apply(src, dst);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace std;
+using namespace testing;
+using namespace perf;
+
+//////////////////////////////////////////////////////////////////////
+// HoughLines
+
+namespace
+{
+ struct Vec4iComparator
+ {
+ bool operator()(const cv::Vec4i& a, const cv::Vec4i b) const
+ {
+ if (a[0] != b[0]) return a[0] < b[0];
+ else if(a[1] != b[1]) return a[1] < b[1];
+ else if(a[2] != b[2]) return a[2] < b[2];
+ else return a[3] < b[3];
+ }
+ };
+ struct Vec3fComparator
+ {
+ bool operator()(const cv::Vec3f& a, const cv::Vec3f b) const
+ {
+ if(a[0] != b[0]) return a[0] < b[0];
+ else if(a[1] != b[1]) return a[1] < b[1];
+ else return a[2] < b[2];
+ }
+ };
+ struct Vec2fComparator
+ {
+ bool operator()(const cv::Vec2f& a, const cv::Vec2f b) const
+ {
+ if(a[0] != b[0]) return a[0] < b[0];
+ else return a[1] < b[1];
+ }
+ };
+}
+
+PERF_TEST_P(Sz, HoughLines,
+ GPU_TYPICAL_MAT_SIZES)
+{
+ declare.time(30.0);
+
+ const cv::Size size = GetParam();
+
+ const float rho = 1.0f;
+ const float theta = static_cast<float>(CV_PI / 180.0);
+ const int threshold = 300;
+
+ cv::Mat src(size, CV_8UC1, cv::Scalar::all(0));
+ cv::line(src, cv::Point(0, 100), cv::Point(src.cols, 100), cv::Scalar::all(255), 1);
+ cv::line(src, cv::Point(0, 200), cv::Point(src.cols, 200), cv::Scalar::all(255), 1);
+ cv::line(src, cv::Point(0, 400), cv::Point(src.cols, 400), cv::Scalar::all(255), 1);
+ cv::line(src, cv::Point(100, 0), cv::Point(100, src.rows), cv::Scalar::all(255), 1);
+ cv::line(src, cv::Point(200, 0), cv::Point(200, src.rows), cv::Scalar::all(255), 1);
+ cv::line(src, cv::Point(400, 0), cv::Point(400, src.rows), cv::Scalar::all(255), 1);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat d_lines;
+ cv::gpu::HoughLinesBuf d_buf;
+
+ TEST_CYCLE() cv::gpu::HoughLines(d_src, d_lines, d_buf, rho, theta, threshold);
+
+ cv::Mat gpu_lines(d_lines.row(0));
+ cv::Vec2f* begin = gpu_lines.ptr<cv::Vec2f>(0);
+ cv::Vec2f* end = begin + gpu_lines.cols;
+ std::sort(begin, end, Vec2fComparator());
+ SANITY_CHECK(gpu_lines);
+ }
+ else
+ {
+ std::vector<cv::Vec2f> cpu_lines;
+
+ TEST_CYCLE() cv::HoughLines(src, cpu_lines, rho, theta, threshold);
+
+ SANITY_CHECK(cpu_lines);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// HoughLinesP
+
+DEF_PARAM_TEST_1(Image, std::string);
+
+PERF_TEST_P(Image, HoughLinesP,
+ testing::Values("cv/shared/pic5.png", "stitching/a1.png"))
+{
+ declare.time(30.0);
+
+ const std::string fileName = getDataPath(GetParam());
+
+ const float rho = 1.0f;
+ const float theta = static_cast<float>(CV_PI / 180.0);
+ const int threshold = 100;
+ const int minLineLenght = 50;
+ const int maxLineGap = 5;
+
+ const cv::Mat image = cv::imread(fileName, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(image.empty());
+
+ cv::Mat mask;
+ cv::Canny(image, mask, 50, 100);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_mask(mask);
+ cv::gpu::GpuMat d_lines;
+ cv::gpu::HoughLinesBuf d_buf;
+
+ TEST_CYCLE() cv::gpu::HoughLinesP(d_mask, d_lines, d_buf, rho, theta, minLineLenght, maxLineGap);
+
+ cv::Mat gpu_lines(d_lines);
+ cv::Vec4i* begin = gpu_lines.ptr<cv::Vec4i>();
+ cv::Vec4i* end = begin + gpu_lines.cols;
+ std::sort(begin, end, Vec4iComparator());
+ SANITY_CHECK(gpu_lines);
+ }
+ else
+ {
+ std::vector<cv::Vec4i> cpu_lines;
+
+ TEST_CYCLE() cv::HoughLinesP(mask, cpu_lines, rho, theta, threshold, minLineLenght, maxLineGap);
+
+ SANITY_CHECK(cpu_lines);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// HoughCircles
+
+DEF_PARAM_TEST(Sz_Dp_MinDist, cv::Size, float, float);
+
+PERF_TEST_P(Sz_Dp_MinDist, HoughCircles,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(1.0f, 2.0f, 4.0f),
+ Values(1.0f)))
+{
+ declare.time(30.0);
+
+ const cv::Size size = GET_PARAM(0);
+ const float dp = GET_PARAM(1);
+ const float minDist = GET_PARAM(2);
+
+ const int minRadius = 10;
+ const int maxRadius = 30;
+ const int cannyThreshold = 100;
+ const int votesThreshold = 15;
+
+ cv::Mat src(size, CV_8UC1, cv::Scalar::all(0));
+ cv::circle(src, cv::Point(100, 100), 20, cv::Scalar::all(255), -1);
+ cv::circle(src, cv::Point(200, 200), 25, cv::Scalar::all(255), -1);
+ cv::circle(src, cv::Point(200, 100), 25, cv::Scalar::all(255), -1);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat d_circles;
+ cv::gpu::HoughCirclesBuf d_buf;
+
+ TEST_CYCLE() cv::gpu::HoughCircles(d_src, d_circles, d_buf, cv::HOUGH_GRADIENT, dp, minDist, cannyThreshold, votesThreshold, minRadius, maxRadius);
+
+ cv::Mat gpu_circles(d_circles);
+ cv::Vec3f* begin = gpu_circles.ptr<cv::Vec3f>(0);
+ cv::Vec3f* end = begin + gpu_circles.cols;
+ std::sort(begin, end, Vec3fComparator());
+ SANITY_CHECK(gpu_circles);
+ }
+ else
+ {
+ std::vector<cv::Vec3f> cpu_circles;
+
+ TEST_CYCLE() cv::HoughCircles(src, cpu_circles, cv::HOUGH_GRADIENT, dp, minDist, cannyThreshold, votesThreshold, minRadius, maxRadius);
+
+ SANITY_CHECK(cpu_circles);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// GeneralizedHough
+
+enum { GHT_POSITION = cv::GeneralizedHough::GHT_POSITION,
+ GHT_SCALE = cv::GeneralizedHough::GHT_SCALE,
+ GHT_ROTATION = cv::GeneralizedHough::GHT_ROTATION
+ };
+
+CV_FLAGS(GHMethod, GHT_POSITION, GHT_SCALE, GHT_ROTATION);
+
+DEF_PARAM_TEST(Method_Sz, GHMethod, cv::Size);
+
+PERF_TEST_P(Method_Sz, GeneralizedHough,
+ Combine(Values(GHMethod(GHT_POSITION), GHMethod(GHT_POSITION | GHT_SCALE), GHMethod(GHT_POSITION | GHT_ROTATION), GHMethod(GHT_POSITION | GHT_SCALE | GHT_ROTATION)),
+ GPU_TYPICAL_MAT_SIZES))
+{
+ declare.time(10);
+
+ const int method = GET_PARAM(0);
+ const cv::Size imageSize = GET_PARAM(1);
+
+ const cv::Mat templ = readImage("cv/shared/templ.png", cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(templ.empty());
+
+ cv::Mat image(imageSize, CV_8UC1, cv::Scalar::all(0));
+ templ.copyTo(image(cv::Rect(50, 50, templ.cols, templ.rows)));
+
+ cv::RNG rng(123456789);
+ const int objCount = rng.uniform(5, 15);
+ for (int i = 0; i < objCount; ++i)
+ {
+ double scale = rng.uniform(0.7, 1.3);
+ bool rotate = 1 == rng.uniform(0, 2);
+
+ cv::Mat obj;
+ cv::resize(templ, obj, cv::Size(), scale, scale);
+ if (rotate)
+ obj = obj.t();
+
+ cv::Point pos;
+
+ pos.x = rng.uniform(0, image.cols - obj.cols);
+ pos.y = rng.uniform(0, image.rows - obj.rows);
+
+ cv::Mat roi = image(cv::Rect(pos, obj.size()));
+ cv::add(roi, obj, roi);
+ }
+
+ cv::Mat edges;
+ cv::Canny(image, edges, 50, 100);
+
+ cv::Mat dx, dy;
+ cv::Sobel(image, dx, CV_32F, 1, 0);
+ cv::Sobel(image, dy, CV_32F, 0, 1);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_edges(edges);
+ const cv::gpu::GpuMat d_dx(dx);
+ const cv::gpu::GpuMat d_dy(dy);
+ cv::gpu::GpuMat posAndVotes;
+
+ cv::Ptr<cv::gpu::GeneralizedHough_GPU> d_hough = cv::gpu::GeneralizedHough_GPU::create(method);
+ if (method & GHT_ROTATION)
+ {
+ d_hough->set("maxAngle", 90.0);
+ d_hough->set("angleStep", 2.0);
+ }
+
+ d_hough->setTemplate(cv::gpu::GpuMat(templ));
+
+ TEST_CYCLE() d_hough->detect(d_edges, d_dx, d_dy, posAndVotes);
+
+ const cv::gpu::GpuMat positions(1, posAndVotes.cols, CV_32FC4, posAndVotes.data);
+ GPU_SANITY_CHECK(positions);
+ }
+ else
+ {
+ cv::Mat positions;
+
+ cv::Ptr<cv::GeneralizedHough> hough = cv::GeneralizedHough::create(method);
+ if (method & GHT_ROTATION)
+ {
+ hough->set("maxAngle", 90.0);
+ hough->set("angleStep", 2.0);
+ }
+
+ hough->setTemplate(templ);
+
+ TEST_CYCLE() hough->detect(edges, dx, dy, positions);
+
+ CPU_SANITY_CHECK(positions);
+ }
+}
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace perf;
+
+CV_PERF_TEST_MAIN(gpuimgproc, printCudaInfo())
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace std;
+using namespace testing;
+using namespace perf;
+
+////////////////////////////////////////////////////////////////////////////////
+// MatchTemplate8U
+
+CV_ENUM(TemplateMethod, TM_SQDIFF, TM_SQDIFF_NORMED, TM_CCORR, TM_CCORR_NORMED, TM_CCOEFF, TM_CCOEFF_NORMED)
+
+DEF_PARAM_TEST(Sz_TemplateSz_Cn_Method, cv::Size, cv::Size, MatCn, TemplateMethod);
+
+PERF_TEST_P(Sz_TemplateSz_Cn_Method, MatchTemplate8U,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(cv::Size(5, 5), cv::Size(16, 16), cv::Size(30, 30)),
+ GPU_CHANNELS_1_3_4,
+ TemplateMethod::all()))
+{
+ declare.time(300.0);
+
+ const cv::Size size = GET_PARAM(0);
+ const cv::Size templ_size = GET_PARAM(1);
+ const int cn = GET_PARAM(2);
+ const int method = GET_PARAM(3);
+
+ cv::Mat image(size, CV_MAKE_TYPE(CV_8U, cn));
+ cv::Mat templ(templ_size, CV_MAKE_TYPE(CV_8U, cn));
+ declare.in(image, templ, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_image(image);
+ const cv::gpu::GpuMat d_templ(templ);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::matchTemplate(d_image, d_templ, dst, method);
+
+ GPU_SANITY_CHECK(dst, 1e-5, ERROR_RELATIVE);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::matchTemplate(image, templ, dst, method);
+
+ CPU_SANITY_CHECK(dst);
+ }
+};
+
+////////////////////////////////////////////////////////////////////////////////
+// MatchTemplate32F
+
+PERF_TEST_P(Sz_TemplateSz_Cn_Method, MatchTemplate32F,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(cv::Size(5, 5), cv::Size(16, 16), cv::Size(30, 30)),
+ GPU_CHANNELS_1_3_4,
+ Values(TemplateMethod(cv::TM_SQDIFF), TemplateMethod(cv::TM_CCORR))))
+{
+ declare.time(300.0);
+
+ const cv::Size size = GET_PARAM(0);
+ const cv::Size templ_size = GET_PARAM(1);
+ const int cn = GET_PARAM(2);
+ int method = GET_PARAM(3);
+
+ cv::Mat image(size, CV_MAKE_TYPE(CV_32F, cn));
+ cv::Mat templ(templ_size, CV_MAKE_TYPE(CV_32F, cn));
+ declare.in(image, templ, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_image(image);
+ const cv::gpu::GpuMat d_templ(templ);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::matchTemplate(d_image, d_templ, dst, method);
+
+ GPU_SANITY_CHECK(dst, 1e-6, ERROR_RELATIVE);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::matchTemplate(image, templ, dst, method);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace std;
+using namespace testing;
+using namespace perf;
+
+//////////////////////////////////////////////////////////////////////
+// MeanShiftFiltering
+
+DEF_PARAM_TEST_1(Image, string);
+
+PERF_TEST_P(Image, MeanShiftFiltering,
+ Values<string>("gpu/meanshift/cones.png"))
+{
+ declare.time(300.0);
+
+ const cv::Mat img = readImage(GetParam());
+ ASSERT_FALSE(img.empty());
+
+ cv::Mat rgba;
+ cv::cvtColor(img, rgba, cv::COLOR_BGR2BGRA);
+
+ const int sp = 50;
+ const int sr = 50;
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(rgba);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::meanShiftFiltering(d_src, dst, sp, sr);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::pyrMeanShiftFiltering(img, dst, sp, sr);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// MeanShiftProc
+
+PERF_TEST_P(Image, MeanShiftProc,
+ Values<string>("gpu/meanshift/cones.png"))
+{
+ declare.time(300.0);
+
+ const cv::Mat img = readImage(GetParam());
+ ASSERT_FALSE(img.empty());
+
+ cv::Mat rgba;
+ cv::cvtColor(img, rgba, cv::COLOR_BGR2BGRA);
+
+ const int sp = 50;
+ const int sr = 50;
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(rgba);
+ cv::gpu::GpuMat dstr;
+ cv::gpu::GpuMat dstsp;
+
+ TEST_CYCLE() cv::gpu::meanShiftProc(d_src, dstr, dstsp, sp, sr);
+
+ GPU_SANITY_CHECK(dstr);
+ GPU_SANITY_CHECK(dstsp);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// MeanShiftSegmentation
+
+PERF_TEST_P(Image, MeanShiftSegmentation,
+ Values<string>("gpu/meanshift/cones.png"))
+{
+ declare.time(300.0);
+
+ const cv::Mat img = readImage(GetParam());
+ ASSERT_FALSE(img.empty());
+
+ cv::Mat rgba;
+ cv::cvtColor(img, rgba, cv::COLOR_BGR2BGRA);
+
+ const int sp = 10;
+ const int sr = 10;
+ const int minsize = 20;
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(rgba);
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::gpu::meanShiftSegmentation(d_src, dst, sp, sr, minsize);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifdef __GNUC__
+# pragma GCC diagnostic ignored "-Wmissing-declarations"
+# if defined __clang__ || defined __APPLE__
+# pragma GCC diagnostic ignored "-Wmissing-prototypes"
+# pragma GCC diagnostic ignored "-Wextra"
+# endif
+#endif
+
+#ifndef __OPENCV_PERF_PRECOMP_HPP__
+#define __OPENCV_PERF_PRECOMP_HPP__
+
+#include "opencv2/ts.hpp"
+#include "opencv2/ts/gpu_perf.hpp"
+
+#include "opencv2/gpuimgproc.hpp"
+#include "opencv2/imgproc.hpp"
+
+#ifdef GTEST_CREATE_SHARED_LIBRARY
+#error no modules except ts should have GTEST_CREATE_SHARED_LIBRARY defined
+#endif
+
+#endif
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "precomp.hpp"
+
+using namespace cv;
+using namespace cv::gpu;
+
+#if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
+
+void cv::gpu::bilateralFilter(const GpuMat&, GpuMat&, int, float, float, int, Stream&) { throw_no_cuda(); }
+
+#else
+
+namespace cv { namespace gpu { namespace cudev
+{
+ namespace imgproc
+ {
+ template<typename T>
+ void bilateral_filter_gpu(const PtrStepSzb& src, PtrStepSzb dst, int kernel_size, float sigma_spatial, float sigma_color, int borderMode, cudaStream_t stream);
+ }
+}}}
+
+void cv::gpu::bilateralFilter(const GpuMat& src, GpuMat& dst, int kernel_size, float sigma_color, float sigma_spatial, int borderMode, Stream& s)
+{
+ using cv::gpu::cudev::imgproc::bilateral_filter_gpu;
+
+ typedef void (*func_t)(const PtrStepSzb& src, PtrStepSzb dst, int kernel_size, float sigma_spatial, float sigma_color, int borderMode, cudaStream_t s);
+
+ static const func_t funcs[6][4] =
+ {
+ {bilateral_filter_gpu<uchar> , 0 /*bilateral_filter_gpu<uchar2>*/ , bilateral_filter_gpu<uchar3> , bilateral_filter_gpu<uchar4> },
+ {0 /*bilateral_filter_gpu<schar>*/, 0 /*bilateral_filter_gpu<schar2>*/ , 0 /*bilateral_filter_gpu<schar3>*/, 0 /*bilateral_filter_gpu<schar4>*/},
+ {bilateral_filter_gpu<ushort> , 0 /*bilateral_filter_gpu<ushort2>*/, bilateral_filter_gpu<ushort3> , bilateral_filter_gpu<ushort4> },
+ {bilateral_filter_gpu<short> , 0 /*bilateral_filter_gpu<short2>*/ , bilateral_filter_gpu<short3> , bilateral_filter_gpu<short4> },
+ {0 /*bilateral_filter_gpu<int>*/ , 0 /*bilateral_filter_gpu<int2>*/ , 0 /*bilateral_filter_gpu<int3>*/ , 0 /*bilateral_filter_gpu<int4>*/ },
+ {bilateral_filter_gpu<float> , 0 /*bilateral_filter_gpu<float2>*/ , bilateral_filter_gpu<float3> , bilateral_filter_gpu<float4> }
+ };
+
+ sigma_color = (sigma_color <= 0 ) ? 1 : sigma_color;
+ sigma_spatial = (sigma_spatial <= 0 ) ? 1 : sigma_spatial;
+
+
+ int radius = (kernel_size <= 0) ? cvRound(sigma_spatial*1.5) : kernel_size/2;
+ kernel_size = std::max(radius, 1)*2 + 1;
+
+ CV_Assert(src.depth() <= CV_32F && src.channels() <= 4);
+ const func_t func = funcs[src.depth()][src.channels() - 1];
+ CV_Assert(func != 0);
+
+ CV_Assert(borderMode == BORDER_REFLECT101 || borderMode == BORDER_REPLICATE || borderMode == BORDER_CONSTANT || borderMode == BORDER_REFLECT || borderMode == BORDER_WRAP);
+
+ dst.create(src.size(), src.type());
+ func(src, dst, kernel_size, sigma_spatial, sigma_color, borderMode, StreamAccessor::getStream(s));
+}
+
+#endif
#else
+////////////////////////////////////////////////////////////////////////
+// blendLinear
+
namespace cv { namespace gpu { namespace cudev
{
namespace blend
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "precomp.hpp"
+
+using namespace cv;
+using namespace cv::gpu;
+
+#if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
+
+void cv::gpu::Canny(const GpuMat&, GpuMat&, double, double, int, bool) { throw_no_cuda(); }
+void cv::gpu::Canny(const GpuMat&, CannyBuf&, GpuMat&, double, double, int, bool) { throw_no_cuda(); }
+void cv::gpu::Canny(const GpuMat&, const GpuMat&, GpuMat&, double, double, bool) { throw_no_cuda(); }
+void cv::gpu::Canny(const GpuMat&, const GpuMat&, CannyBuf&, GpuMat&, double, double, bool) { throw_no_cuda(); }
+void cv::gpu::CannyBuf::create(const Size&, int) { throw_no_cuda(); }
+void cv::gpu::CannyBuf::release() { throw_no_cuda(); }
+
+#else /* !defined (HAVE_CUDA) */
+
+void cv::gpu::CannyBuf::create(const Size& image_size, int apperture_size)
+{
+ if (apperture_size > 0)
+ {
+ ensureSizeIsEnough(image_size, CV_32SC1, dx);
+ ensureSizeIsEnough(image_size, CV_32SC1, dy);
+
+ if (apperture_size != 3)
+ {
+ filterDX = createDerivFilter_GPU(CV_8UC1, CV_32S, 1, 0, apperture_size, BORDER_REPLICATE);
+ filterDY = createDerivFilter_GPU(CV_8UC1, CV_32S, 0, 1, apperture_size, BORDER_REPLICATE);
+ }
+ }
+
+ ensureSizeIsEnough(image_size, CV_32FC1, mag);
+ ensureSizeIsEnough(image_size, CV_32SC1, map);
+
+ ensureSizeIsEnough(1, image_size.area(), CV_16UC2, st1);
+ ensureSizeIsEnough(1, image_size.area(), CV_16UC2, st2);
+}
+
+void cv::gpu::CannyBuf::release()
+{
+ dx.release();
+ dy.release();
+ mag.release();
+ map.release();
+ st1.release();
+ st2.release();
+}
+
+namespace canny
+{
+ void calcMagnitude(PtrStepSzb srcWhole, int xoff, int yoff, PtrStepSzi dx, PtrStepSzi dy, PtrStepSzf mag, bool L2Grad);
+ void calcMagnitude(PtrStepSzi dx, PtrStepSzi dy, PtrStepSzf mag, bool L2Grad);
+
+ void calcMap(PtrStepSzi dx, PtrStepSzi dy, PtrStepSzf mag, PtrStepSzi map, float low_thresh, float high_thresh);
+
+ void edgesHysteresisLocal(PtrStepSzi map, ushort2* st1);
+
+ void edgesHysteresisGlobal(PtrStepSzi map, ushort2* st1, ushort2* st2);
+
+ void getEdges(PtrStepSzi map, PtrStepSzb dst);
+}
+
+namespace
+{
+ void CannyCaller(const GpuMat& dx, const GpuMat& dy, CannyBuf& buf, GpuMat& dst, float low_thresh, float high_thresh)
+ {
+ using namespace canny;
+
+ buf.map.setTo(Scalar::all(0));
+ calcMap(dx, dy, buf.mag, buf.map, low_thresh, high_thresh);
+
+ edgesHysteresisLocal(buf.map, buf.st1.ptr<ushort2>());
+
+ edgesHysteresisGlobal(buf.map, buf.st1.ptr<ushort2>(), buf.st2.ptr<ushort2>());
+
+ getEdges(buf.map, dst);
+ }
+}
+
+void cv::gpu::Canny(const GpuMat& src, GpuMat& dst, double low_thresh, double high_thresh, int apperture_size, bool L2gradient)
+{
+ CannyBuf buf;
+ Canny(src, buf, dst, low_thresh, high_thresh, apperture_size, L2gradient);
+}
+
+void cv::gpu::Canny(const GpuMat& src, CannyBuf& buf, GpuMat& dst, double low_thresh, double high_thresh, int apperture_size, bool L2gradient)
+{
+ using namespace canny;
+
+ CV_Assert(src.type() == CV_8UC1);
+
+ if (!deviceSupports(SHARED_ATOMICS))
+ CV_Error(cv::Error::StsNotImplemented, "The device doesn't support shared atomics");
+
+ if( low_thresh > high_thresh )
+ std::swap( low_thresh, high_thresh);
+
+ dst.create(src.size(), CV_8U);
+ buf.create(src.size(), apperture_size);
+
+ if (apperture_size == 3)
+ {
+ Size wholeSize;
+ Point ofs;
+ src.locateROI(wholeSize, ofs);
+ GpuMat srcWhole(wholeSize, src.type(), src.datastart, src.step);
+
+ calcMagnitude(srcWhole, ofs.x, ofs.y, buf.dx, buf.dy, buf.mag, L2gradient);
+ }
+ else
+ {
+ buf.filterDX->apply(src, buf.dx, Rect(0, 0, src.cols, src.rows));
+ buf.filterDY->apply(src, buf.dy, Rect(0, 0, src.cols, src.rows));
+
+ calcMagnitude(buf.dx, buf.dy, buf.mag, L2gradient);
+ }
+
+ CannyCaller(buf.dx, buf.dy, buf, dst, static_cast<float>(low_thresh), static_cast<float>(high_thresh));
+}
+
+void cv::gpu::Canny(const GpuMat& dx, const GpuMat& dy, GpuMat& dst, double low_thresh, double high_thresh, bool L2gradient)
+{
+ CannyBuf buf;
+ Canny(dx, dy, buf, dst, low_thresh, high_thresh, L2gradient);
+}
+
+void cv::gpu::Canny(const GpuMat& dx, const GpuMat& dy, CannyBuf& buf, GpuMat& dst, double low_thresh, double high_thresh, bool L2gradient)
+{
+ using namespace canny;
+
+ CV_Assert(TargetArchs::builtWith(SHARED_ATOMICS) && DeviceInfo().supports(SHARED_ATOMICS));
+ CV_Assert(dx.type() == CV_32SC1 && dy.type() == CV_32SC1 && dx.size() == dy.size());
+
+ if( low_thresh > high_thresh )
+ std::swap( low_thresh, high_thresh);
+
+ dst.create(dx.size(), CV_8U);
+ buf.create(dx.size(), -1);
+
+ calcMagnitude(dx, dy, buf.mag, L2gradient);
+
+ CannyCaller(dx, dy, buf, dst, static_cast<float>(low_thresh), static_cast<float>(high_thresh));
+}
+
+#endif /* !defined (HAVE_CUDA) */
#if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
void cv::gpu::cvtColor(const GpuMat&, GpuMat&, int, int, Stream&) { throw_no_cuda(); }
+
void cv::gpu::demosaicing(const GpuMat&, GpuMat&, int, int, Stream&) { throw_no_cuda(); }
+
void cv::gpu::swapChannels(GpuMat&, const int[], Stream&) { throw_no_cuda(); }
+
void cv::gpu::gammaCorrection(const GpuMat&, GpuMat&, bool, Stream&) { throw_no_cuda(); }
+void cv::gpu::alphaComp(const GpuMat&, const GpuMat&, GpuMat&, int, Stream&) { throw_no_cuda(); }
+
#else /* !defined (HAVE_CUDA) */
#include "cvt_color_internal.h"
}
}
+////////////////////////////////////////////////////////////////////////
+// cvtColor
+
void cv::gpu::cvtColor(const GpuMat& src, GpuMat& dst, int code, int dcn, Stream& stream)
{
typedef void (*func_t)(const GpuMat& src, GpuMat& dst, int dcn, Stream& stream);
func(src, dst, dcn, stream);
}
+////////////////////////////////////////////////////////////////////////
+// demosaicing
+
void cv::gpu::demosaicing(const GpuMat& src, GpuMat& dst, int code, int dcn, Stream& stream)
{
const int depth = src.depth();
}
}
+////////////////////////////////////////////////////////////////////////
+// swapChannels
+
void cv::gpu::swapChannels(GpuMat& image, const int dstOrder[4], Stream& s)
{
CV_Assert(image.type() == CV_8UC4);
cudaSafeCall( cudaDeviceSynchronize() );
}
+////////////////////////////////////////////////////////////////////////
+// gammaCorrection
+
void cv::gpu::gammaCorrection(const GpuMat& src, GpuMat& dst, bool forward, Stream& stream)
{
#if (CUDA_VERSION < 5000)
#endif
}
+////////////////////////////////////////////////////////////////////////
+// alphaComp
+
+namespace
+{
+ template <int DEPTH> struct NppAlphaCompFunc
+ {
+ typedef typename NPPTypeTraits<DEPTH>::npp_type npp_t;
+
+ typedef NppStatus (*func_t)(const npp_t* pSrc1, int nSrc1Step, const npp_t* pSrc2, int nSrc2Step, npp_t* pDst, int nDstStep, NppiSize oSizeROI, NppiAlphaOp eAlphaOp);
+ };
+
+ template <int DEPTH, typename NppAlphaCompFunc<DEPTH>::func_t func> struct NppAlphaComp
+ {
+ typedef typename NPPTypeTraits<DEPTH>::npp_type npp_t;
+
+ static void call(const GpuMat& img1, const GpuMat& img2, GpuMat& dst, NppiAlphaOp eAlphaOp, cudaStream_t stream)
+ {
+ NppStreamHandler h(stream);
+
+ NppiSize oSizeROI;
+ oSizeROI.width = img1.cols;
+ oSizeROI.height = img2.rows;
+
+ nppSafeCall( func(img1.ptr<npp_t>(), static_cast<int>(img1.step), img2.ptr<npp_t>(), static_cast<int>(img2.step),
+ dst.ptr<npp_t>(), static_cast<int>(dst.step), oSizeROI, eAlphaOp) );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+ };
+}
+
+void cv::gpu::alphaComp(const GpuMat& img1, const GpuMat& img2, GpuMat& dst, int alpha_op, Stream& stream)
+{
+ static const NppiAlphaOp npp_alpha_ops[] = {
+ NPPI_OP_ALPHA_OVER,
+ NPPI_OP_ALPHA_IN,
+ NPPI_OP_ALPHA_OUT,
+ NPPI_OP_ALPHA_ATOP,
+ NPPI_OP_ALPHA_XOR,
+ NPPI_OP_ALPHA_PLUS,
+ NPPI_OP_ALPHA_OVER_PREMUL,
+ NPPI_OP_ALPHA_IN_PREMUL,
+ NPPI_OP_ALPHA_OUT_PREMUL,
+ NPPI_OP_ALPHA_ATOP_PREMUL,
+ NPPI_OP_ALPHA_XOR_PREMUL,
+ NPPI_OP_ALPHA_PLUS_PREMUL,
+ NPPI_OP_ALPHA_PREMUL
+ };
+
+ typedef void (*func_t)(const GpuMat& img1, const GpuMat& img2, GpuMat& dst, NppiAlphaOp eAlphaOp, cudaStream_t stream);
+
+ static const func_t funcs[] =
+ {
+ NppAlphaComp<CV_8U, nppiAlphaComp_8u_AC4R>::call,
+ 0,
+ NppAlphaComp<CV_16U, nppiAlphaComp_16u_AC4R>::call,
+ 0,
+ NppAlphaComp<CV_32S, nppiAlphaComp_32s_AC4R>::call,
+ NppAlphaComp<CV_32F, nppiAlphaComp_32f_AC4R>::call
+ };
+
+ CV_Assert( img1.type() == CV_8UC4 || img1.type() == CV_16UC4 || img1.type() == CV_32SC4 || img1.type() == CV_32FC4 );
+ CV_Assert( img1.size() == img2.size() && img1.type() == img2.type() );
+
+ dst.create(img1.size(), img1.type());
+
+ const func_t func = funcs[img1.depth()];
+
+ func(img1, img2, dst, npp_alpha_ops[alpha_op], StreamAccessor::getStream(stream));
+}
+
#endif /* !defined (HAVE_CUDA) */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "precomp.hpp"
+
+using namespace cv;
+using namespace cv::gpu;
+
+#if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
+
+void cv::gpu::cornerHarris(const GpuMat&, GpuMat&, int, int, double, int) { throw_no_cuda(); }
+void cv::gpu::cornerHarris(const GpuMat&, GpuMat&, GpuMat&, GpuMat&, int, int, double, int) { throw_no_cuda(); }
+void cv::gpu::cornerHarris(const GpuMat&, GpuMat&, GpuMat&, GpuMat&, GpuMat&, int, int, double, int, Stream&) { throw_no_cuda(); }
+
+void cv::gpu::cornerMinEigenVal(const GpuMat&, GpuMat&, int, int, int) { throw_no_cuda(); }
+void cv::gpu::cornerMinEigenVal(const GpuMat&, GpuMat&, GpuMat&, GpuMat&, int, int, int) { throw_no_cuda(); }
+void cv::gpu::cornerMinEigenVal(const GpuMat&, GpuMat&, GpuMat&, GpuMat&, GpuMat&, int, int, int, Stream&) { throw_no_cuda(); }
+
+#else /* !defined (HAVE_CUDA) */
+
+namespace cv { namespace gpu { namespace cudev
+{
+ namespace imgproc
+ {
+ void cornerHarris_gpu(int block_size, float k, PtrStepSzf Dx, PtrStepSzf Dy, PtrStepSzf dst, int border_type, cudaStream_t stream);
+ void cornerMinEigenVal_gpu(int block_size, PtrStepSzf Dx, PtrStepSzf Dy, PtrStepSzf dst, int border_type, cudaStream_t stream);
+ }
+}}}
+
+namespace
+{
+ void extractCovData(const GpuMat& src, GpuMat& Dx, GpuMat& Dy, GpuMat& buf, int blockSize, int ksize, int borderType, Stream& stream)
+ {
+ double scale = static_cast<double>(1 << ((ksize > 0 ? ksize : 3) - 1)) * blockSize;
+
+ if (ksize < 0)
+ scale *= 2.;
+
+ if (src.depth() == CV_8U)
+ scale *= 255.;
+
+ scale = 1./scale;
+
+ Dx.create(src.size(), CV_32F);
+ Dy.create(src.size(), CV_32F);
+
+ if (ksize > 0)
+ {
+ Sobel(src, Dx, CV_32F, 1, 0, buf, ksize, scale, borderType, -1, stream);
+ Sobel(src, Dy, CV_32F, 0, 1, buf, ksize, scale, borderType, -1, stream);
+ }
+ else
+ {
+ Scharr(src, Dx, CV_32F, 1, 0, buf, scale, borderType, -1, stream);
+ Scharr(src, Dy, CV_32F, 0, 1, buf, scale, borderType, -1, stream);
+ }
+ }
+}
+
+void cv::gpu::cornerHarris(const GpuMat& src, GpuMat& dst, int blockSize, int ksize, double k, int borderType)
+{
+ GpuMat Dx, Dy;
+ cornerHarris(src, dst, Dx, Dy, blockSize, ksize, k, borderType);
+}
+
+void cv::gpu::cornerHarris(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, int blockSize, int ksize, double k, int borderType)
+{
+ GpuMat buf;
+ cornerHarris(src, dst, Dx, Dy, buf, blockSize, ksize, k, borderType);
+}
+
+void cv::gpu::cornerHarris(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, GpuMat& buf, int blockSize, int ksize, double k, int borderType, Stream& stream)
+{
+ using namespace cv::gpu::cudev::imgproc;
+
+ CV_Assert(borderType == cv::BORDER_REFLECT101 || borderType == cv::BORDER_REPLICATE || borderType == cv::BORDER_REFLECT);
+
+ extractCovData(src, Dx, Dy, buf, blockSize, ksize, borderType, stream);
+
+ dst.create(src.size(), CV_32F);
+
+ cornerHarris_gpu(blockSize, static_cast<float>(k), Dx, Dy, dst, borderType, StreamAccessor::getStream(stream));
+}
+
+void cv::gpu::cornerMinEigenVal(const GpuMat& src, GpuMat& dst, int blockSize, int ksize, int borderType)
+{
+ GpuMat Dx, Dy;
+ cornerMinEigenVal(src, dst, Dx, Dy, blockSize, ksize, borderType);
+}
+
+void cv::gpu::cornerMinEigenVal(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, int blockSize, int ksize, int borderType)
+{
+ GpuMat buf;
+ cornerMinEigenVal(src, dst, Dx, Dy, buf, blockSize, ksize, borderType);
+}
+
+void cv::gpu::cornerMinEigenVal(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, GpuMat& buf, int blockSize, int ksize, int borderType, Stream& stream)
+{
+ using namespace ::cv::gpu::cudev::imgproc;
+
+ CV_Assert(borderType == cv::BORDER_REFLECT101 || borderType == cv::BORDER_REPLICATE || borderType == cv::BORDER_REFLECT);
+
+ extractCovData(src, Dx, Dy, buf, blockSize, ksize, borderType, stream);
+
+ dst.create(src.size(), CV_32F);
+
+ cornerMinEigenVal_gpu(blockSize, Dx, Dy, dst, borderType, StreamAccessor::getStream(stream));
+}
+
+#endif /* !defined (HAVE_CUDA) */
static caller_t funcs[] =
{
- bilateral_caller<T, BrdReflect101>,
- bilateral_caller<T, BrdReplicate>,
bilateral_caller<T, BrdConstant>,
+ bilateral_caller<T, BrdReplicate>,
bilateral_caller<T, BrdReflect>,
bilateral_caller<T, BrdWrap>,
+ bilateral_caller<T, BrdReflect101>
};
funcs[borderMode](src, dst, kernel_size, gauss_spatial_coeff, gauss_color_coeff, stream);
}
texture<uchar, cudaTextureType2D, cudaReadModeElementType> tex_src(false, cudaFilterModePoint, cudaAddressModeClamp);
struct SrcTex
{
- const int xoff;
- const int yoff;
+ int xoff;
+ int yoff;
__host__ SrcTex(int _xoff, int _yoff) : xoff(_xoff), yoff(_yoff) {}
__device__ __forceinline__ int operator ()(int y, int x) const
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/vec_traits.hpp"
+#include "opencv2/core/cuda/vec_math.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/border_interpolate.hpp"
+
+namespace cv { namespace gpu { namespace cudev
+{
+ namespace imgproc
+ {
+ /////////////////////////////////////////// Corner Harris /////////////////////////////////////////////////
+
+ texture<float, cudaTextureType2D, cudaReadModeElementType> harrisDxTex(0, cudaFilterModePoint, cudaAddressModeClamp);
+ texture<float, cudaTextureType2D, cudaReadModeElementType> harrisDyTex(0, cudaFilterModePoint, cudaAddressModeClamp);
+
+ __global__ void cornerHarris_kernel(const int block_size, const float k, PtrStepSzf dst)
+ {
+ const int x = blockIdx.x * blockDim.x + threadIdx.x;
+ const int y = blockIdx.y * blockDim.y + threadIdx.y;
+
+ if (x < dst.cols && y < dst.rows)
+ {
+ float a = 0.f;
+ float b = 0.f;
+ float c = 0.f;
+
+ const int ibegin = y - (block_size / 2);
+ const int jbegin = x - (block_size / 2);
+ const int iend = ibegin + block_size;
+ const int jend = jbegin + block_size;
+
+ for (int i = ibegin; i < iend; ++i)
+ {
+ for (int j = jbegin; j < jend; ++j)
+ {
+ float dx = tex2D(harrisDxTex, j, i);
+ float dy = tex2D(harrisDyTex, j, i);
+
+ a += dx * dx;
+ b += dx * dy;
+ c += dy * dy;
+ }
+ }
+
+ dst(y, x) = a * c - b * b - k * (a + c) * (a + c);
+ }
+ }
+
+ template <typename BR, typename BC>
+ __global__ void cornerHarris_kernel(const int block_size, const float k, PtrStepSzf dst, const BR border_row, const BC border_col)
+ {
+ const int x = blockIdx.x * blockDim.x + threadIdx.x;
+ const int y = blockIdx.y * blockDim.y + threadIdx.y;
+
+ if (x < dst.cols && y < dst.rows)
+ {
+ float a = 0.f;
+ float b = 0.f;
+ float c = 0.f;
+
+ const int ibegin = y - (block_size / 2);
+ const int jbegin = x - (block_size / 2);
+ const int iend = ibegin + block_size;
+ const int jend = jbegin + block_size;
+
+ for (int i = ibegin; i < iend; ++i)
+ {
+ const int y = border_col.idx_row(i);
+
+ for (int j = jbegin; j < jend; ++j)
+ {
+ const int x = border_row.idx_col(j);
+
+ float dx = tex2D(harrisDxTex, x, y);
+ float dy = tex2D(harrisDyTex, x, y);
+
+ a += dx * dx;
+ b += dx * dy;
+ c += dy * dy;
+ }
+ }
+
+ dst(y, x) = a * c - b * b - k * (a + c) * (a + c);
+ }
+ }
+
+ void cornerHarris_gpu(int block_size, float k, PtrStepSzf Dx, PtrStepSzf Dy, PtrStepSzf dst, int border_type, cudaStream_t stream)
+ {
+ dim3 block(32, 8);
+ dim3 grid(divUp(Dx.cols, block.x), divUp(Dx.rows, block.y));
+
+ bindTexture(&harrisDxTex, Dx);
+ bindTexture(&harrisDyTex, Dy);
+
+ switch (border_type)
+ {
+ case BORDER_REFLECT101:
+ cornerHarris_kernel<<<grid, block, 0, stream>>>(block_size, k, dst, BrdRowReflect101<void>(Dx.cols), BrdColReflect101<void>(Dx.rows));
+ break;
+
+ case BORDER_REFLECT:
+ cornerHarris_kernel<<<grid, block, 0, stream>>>(block_size, k, dst, BrdRowReflect<void>(Dx.cols), BrdColReflect<void>(Dx.rows));
+ break;
+
+ case BORDER_REPLICATE:
+ cornerHarris_kernel<<<grid, block, 0, stream>>>(block_size, k, dst);
+ break;
+ }
+
+ cudaSafeCall( cudaGetLastError() );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+
+ /////////////////////////////////////////// Corner Min Eigen Val /////////////////////////////////////////////////
+
+ texture<float, cudaTextureType2D, cudaReadModeElementType> minEigenValDxTex(0, cudaFilterModePoint, cudaAddressModeClamp);
+ texture<float, cudaTextureType2D, cudaReadModeElementType> minEigenValDyTex(0, cudaFilterModePoint, cudaAddressModeClamp);
+
+ __global__ void cornerMinEigenVal_kernel(const int block_size, PtrStepSzf dst)
+ {
+ const int x = blockIdx.x * blockDim.x + threadIdx.x;
+ const int y = blockIdx.y * blockDim.y + threadIdx.y;
+
+ if (x < dst.cols && y < dst.rows)
+ {
+ float a = 0.f;
+ float b = 0.f;
+ float c = 0.f;
+
+ const int ibegin = y - (block_size / 2);
+ const int jbegin = x - (block_size / 2);
+ const int iend = ibegin + block_size;
+ const int jend = jbegin + block_size;
+
+ for (int i = ibegin; i < iend; ++i)
+ {
+ for (int j = jbegin; j < jend; ++j)
+ {
+ float dx = tex2D(minEigenValDxTex, j, i);
+ float dy = tex2D(minEigenValDyTex, j, i);
+
+ a += dx * dx;
+ b += dx * dy;
+ c += dy * dy;
+ }
+ }
+
+ a *= 0.5f;
+ c *= 0.5f;
+
+ dst(y, x) = (a + c) - sqrtf((a - c) * (a - c) + b * b);
+ }
+ }
+
+
+ template <typename BR, typename BC>
+ __global__ void cornerMinEigenVal_kernel(const int block_size, PtrStepSzf dst, const BR border_row, const BC border_col)
+ {
+ const int x = blockIdx.x * blockDim.x + threadIdx.x;
+ const int y = blockIdx.y * blockDim.y + threadIdx.y;
+
+ if (x < dst.cols && y < dst.rows)
+ {
+ float a = 0.f;
+ float b = 0.f;
+ float c = 0.f;
+
+ const int ibegin = y - (block_size / 2);
+ const int jbegin = x - (block_size / 2);
+ const int iend = ibegin + block_size;
+ const int jend = jbegin + block_size;
+
+ for (int i = ibegin; i < iend; ++i)
+ {
+ int y = border_col.idx_row(i);
+
+ for (int j = jbegin; j < jend; ++j)
+ {
+ int x = border_row.idx_col(j);
+
+ float dx = tex2D(minEigenValDxTex, x, y);
+ float dy = tex2D(minEigenValDyTex, x, y);
+
+ a += dx * dx;
+ b += dx * dy;
+ c += dy * dy;
+ }
+ }
+
+ a *= 0.5f;
+ c *= 0.5f;
+
+ dst(y, x) = (a + c) - sqrtf((a - c) * (a - c) + b * b);
+ }
+ }
+
+ void cornerMinEigenVal_gpu(int block_size, PtrStepSzf Dx, PtrStepSzf Dy, PtrStepSzf dst, int border_type, cudaStream_t stream)
+ {
+ dim3 block(32, 8);
+ dim3 grid(divUp(Dx.cols, block.x), divUp(Dx.rows, block.y));
+
+ bindTexture(&minEigenValDxTex, Dx);
+ bindTexture(&minEigenValDyTex, Dy);
+
+ switch (border_type)
+ {
+ case BORDER_REFLECT101:
+ cornerMinEigenVal_kernel<<<grid, block, 0, stream>>>(block_size, dst, BrdRowReflect101<void>(Dx.cols), BrdColReflect101<void>(Dx.rows));
+ break;
+
+ case BORDER_REFLECT:
+ cornerMinEigenVal_kernel<<<grid, block, 0, stream>>>(block_size, dst, BrdRowReflect<void>(Dx.cols), BrdColReflect<void>(Dx.rows));
+ break;
+
+ case BORDER_REPLICATE:
+ cornerMinEigenVal_kernel<<<grid, block, 0, stream>>>(block_size, dst);
+ break;
+ }
+
+ cudaSafeCall( cudaGetLastError() );
+
+ if (stream == 0)
+ cudaSafeCall(cudaDeviceSynchronize());
+ }
+ }
+}}}
+
+#endif
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/vec_traits.hpp"
+#include "opencv2/core/cuda/vec_math.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/border_interpolate.hpp"
+
+namespace cv { namespace gpu { namespace cudev
+{
+ namespace imgproc
+ {
+ texture<uchar4, 2> tex_meanshift;
+
+ __device__ short2 do_mean_shift(int x0, int y0, unsigned char* out,
+ size_t out_step, int cols, int rows,
+ int sp, int sr, int maxIter, float eps)
+ {
+ int isr2 = sr*sr;
+ uchar4 c = tex2D(tex_meanshift, x0, y0 );
+
+ // iterate meanshift procedure
+ for( int iter = 0; iter < maxIter; iter++ )
+ {
+ int count = 0;
+ int s0 = 0, s1 = 0, s2 = 0, sx = 0, sy = 0;
+ float icount;
+
+ //mean shift: process pixels in window (p-sigmaSp)x(p+sigmaSp)
+ int minx = x0-sp;
+ int miny = y0-sp;
+ int maxx = x0+sp;
+ int maxy = y0+sp;
+
+ for( int y = miny; y <= maxy; y++)
+ {
+ int rowCount = 0;
+ for( int x = minx; x <= maxx; x++ )
+ {
+ uchar4 t = tex2D( tex_meanshift, x, y );
+
+ int norm2 = (t.x - c.x) * (t.x - c.x) + (t.y - c.y) * (t.y - c.y) + (t.z - c.z) * (t.z - c.z);
+ if( norm2 <= isr2 )
+ {
+ s0 += t.x; s1 += t.y; s2 += t.z;
+ sx += x; rowCount++;
+ }
+ }
+ count += rowCount;
+ sy += y*rowCount;
+ }
+
+ if( count == 0 )
+ break;
+
+ icount = 1.f/count;
+ int x1 = __float2int_rz(sx*icount);
+ int y1 = __float2int_rz(sy*icount);
+ s0 = __float2int_rz(s0*icount);
+ s1 = __float2int_rz(s1*icount);
+ s2 = __float2int_rz(s2*icount);
+
+ int norm2 = (s0 - c.x) * (s0 - c.x) + (s1 - c.y) * (s1 - c.y) + (s2 - c.z) * (s2 - c.z);
+
+ bool stopFlag = (x0 == x1 && y0 == y1) || (::abs(x1-x0) + ::abs(y1-y0) + norm2 <= eps);
+
+ x0 = x1; y0 = y1;
+ c.x = s0; c.y = s1; c.z = s2;
+
+ if( stopFlag )
+ break;
+ }
+
+ int base = (blockIdx.y * blockDim.y + threadIdx.y) * out_step + (blockIdx.x * blockDim.x + threadIdx.x) * 4 * sizeof(uchar);
+ *(uchar4*)(out + base) = c;
+
+ return make_short2((short)x0, (short)y0);
+ }
+
+ __global__ void meanshift_kernel(unsigned char* out, size_t out_step, int cols, int rows, int sp, int sr, int maxIter, float eps )
+ {
+ int x0 = blockIdx.x * blockDim.x + threadIdx.x;
+ int y0 = blockIdx.y * blockDim.y + threadIdx.y;
+
+ if( x0 < cols && y0 < rows )
+ do_mean_shift(x0, y0, out, out_step, cols, rows, sp, sr, maxIter, eps);
+ }
+
+ void meanShiftFiltering_gpu(const PtrStepSzb& src, PtrStepSzb dst, int sp, int sr, int maxIter, float eps, cudaStream_t stream)
+ {
+ dim3 grid(1, 1, 1);
+ dim3 threads(32, 8, 1);
+ grid.x = divUp(src.cols, threads.x);
+ grid.y = divUp(src.rows, threads.y);
+
+ cudaChannelFormatDesc desc = cudaCreateChannelDesc<uchar4>();
+ cudaSafeCall( cudaBindTexture2D( 0, tex_meanshift, src.data, desc, src.cols, src.rows, src.step ) );
+
+ meanshift_kernel<<< grid, threads, 0, stream >>>( dst.data, dst.step, dst.cols, dst.rows, sp, sr, maxIter, eps );
+ cudaSafeCall( cudaGetLastError() );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+
+ __global__ void meanshiftproc_kernel(unsigned char* outr, size_t outrstep,
+ unsigned char* outsp, size_t outspstep,
+ int cols, int rows,
+ int sp, int sr, int maxIter, float eps)
+ {
+ int x0 = blockIdx.x * blockDim.x + threadIdx.x;
+ int y0 = blockIdx.y * blockDim.y + threadIdx.y;
+
+ if( x0 < cols && y0 < rows )
+ {
+ int basesp = (blockIdx.y * blockDim.y + threadIdx.y) * outspstep + (blockIdx.x * blockDim.x + threadIdx.x) * 2 * sizeof(short);
+ *(short2*)(outsp + basesp) = do_mean_shift(x0, y0, outr, outrstep, cols, rows, sp, sr, maxIter, eps);
+ }
+ }
+
+ void meanShiftProc_gpu(const PtrStepSzb& src, PtrStepSzb dstr, PtrStepSzb dstsp, int sp, int sr, int maxIter, float eps, cudaStream_t stream)
+ {
+ dim3 grid(1, 1, 1);
+ dim3 threads(32, 8, 1);
+ grid.x = divUp(src.cols, threads.x);
+ grid.y = divUp(src.rows, threads.y);
+
+ cudaChannelFormatDesc desc = cudaCreateChannelDesc<uchar4>();
+ cudaSafeCall( cudaBindTexture2D( 0, tex_meanshift, src.data, desc, src.cols, src.rows, src.step ) );
+
+ meanshiftproc_kernel<<< grid, threads, 0, stream >>>( dstr.data, dstr.step, dstsp.data, dstsp.step, dstr.cols, dstr.rows, sp, sr, maxIter, eps );
+ cudaSafeCall( cudaGetLastError() );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+ }
+}}}
+
+#endif
void cv::gpu::GoodFeaturesToTrackDetector_GPU::operator ()(const GpuMat& image, GpuMat& corners, const GpuMat& mask)
{
+#ifndef HAVE_OPENCV_GPUARITHM
+ (void) image;
+ (void) corners;
+ (void) mask;
+ throw_no_cuda();
+#else
using namespace cv::gpu::cudev::gfft;
CV_Assert(qualityLevel > 0 && minDistance >= 0 && maxCorners >= 0);
cornerMinEigenVal(image, eig_, Dx_, Dy_, buf_, blockSize, 3);
double maxVal = 0;
- minMax(eig_, 0, &maxVal, GpuMat(), minMaxbuf_);
+ gpu::minMax(eig_, 0, &maxVal, GpuMat(), minMaxbuf_);
ensureSizeIsEnough(1, std::max(1000, static_cast<int>(image.size().area() * 0.05)), CV_32FC2, tmpCorners_);
corners.upload(Mat(1, static_cast<int>(tmp2.size()), CV_32FC2, &tmp2[0]));
}
+#endif
}
#endif /* !defined (HAVE_CUDA) */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "precomp.hpp"
+
+using namespace cv;
+using namespace cv::gpu;
+
+#if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
+
+void cv::gpu::evenLevels(GpuMat&, int, int, int) { throw_no_cuda(); }
+
+void cv::gpu::histEven(const GpuMat&, GpuMat&, int, int, int, Stream&) { throw_no_cuda(); }
+void cv::gpu::histEven(const GpuMat&, GpuMat&, GpuMat&, int, int, int, Stream&) { throw_no_cuda(); }
+void cv::gpu::histEven(const GpuMat&, GpuMat*, int*, int*, int*, Stream&) { throw_no_cuda(); }
+void cv::gpu::histEven(const GpuMat&, GpuMat*, GpuMat&, int*, int*, int*, Stream&) { throw_no_cuda(); }
+
+void cv::gpu::histRange(const GpuMat&, GpuMat&, const GpuMat&, Stream&) { throw_no_cuda(); }
+void cv::gpu::histRange(const GpuMat&, GpuMat&, const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+void cv::gpu::histRange(const GpuMat&, GpuMat*, const GpuMat*, Stream&) { throw_no_cuda(); }
+void cv::gpu::histRange(const GpuMat&, GpuMat*, const GpuMat*, GpuMat&, Stream&) { throw_no_cuda(); }
+
+void cv::gpu::calcHist(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+
+void cv::gpu::equalizeHist(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+void cv::gpu::equalizeHist(const GpuMat&, GpuMat&, GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+
+cv::Ptr<cv::gpu::CLAHE> cv::gpu::createCLAHE(double, cv::Size) { throw_no_cuda(); return cv::Ptr<cv::gpu::CLAHE>(); }
+
+#else /* !defined (HAVE_CUDA) */
+
+////////////////////////////////////////////////////////////////////////
+// NPP Histogram
+
+namespace
+{
+ typedef NppStatus (*get_buf_size_c1_t)(NppiSize oSizeROI, int nLevels, int* hpBufferSize);
+ typedef NppStatus (*get_buf_size_c4_t)(NppiSize oSizeROI, int nLevels[], int* hpBufferSize);
+
+ template<int SDEPTH> struct NppHistogramEvenFuncC1
+ {
+ typedef typename NPPTypeTraits<SDEPTH>::npp_type src_t;
+
+ typedef NppStatus (*func_ptr)(const src_t* pSrc, int nSrcStep, NppiSize oSizeROI, Npp32s * pHist,
+ int nLevels, Npp32s nLowerLevel, Npp32s nUpperLevel, Npp8u * pBuffer);
+ };
+ template<int SDEPTH> struct NppHistogramEvenFuncC4
+ {
+ typedef typename NPPTypeTraits<SDEPTH>::npp_type src_t;
+
+ typedef NppStatus (*func_ptr)(const src_t* pSrc, int nSrcStep, NppiSize oSizeROI,
+ Npp32s * pHist[4], int nLevels[4], Npp32s nLowerLevel[4], Npp32s nUpperLevel[4], Npp8u * pBuffer);
+ };
+
+ template<int SDEPTH, typename NppHistogramEvenFuncC1<SDEPTH>::func_ptr func, get_buf_size_c1_t get_buf_size>
+ struct NppHistogramEvenC1
+ {
+ typedef typename NppHistogramEvenFuncC1<SDEPTH>::src_t src_t;
+
+ static void hist(const GpuMat& src, GpuMat& hist, GpuMat& buffer, int histSize, int lowerLevel, int upperLevel, cudaStream_t stream)
+ {
+ int levels = histSize + 1;
+ hist.create(1, histSize, CV_32S);
+
+ NppiSize sz;
+ sz.width = src.cols;
+ sz.height = src.rows;
+
+ int buf_size;
+ get_buf_size(sz, levels, &buf_size);
+
+ ensureSizeIsEnough(1, buf_size, CV_8U, buffer);
+
+ NppStreamHandler h(stream);
+
+ nppSafeCall( func(src.ptr<src_t>(), static_cast<int>(src.step), sz, hist.ptr<Npp32s>(), levels,
+ lowerLevel, upperLevel, buffer.ptr<Npp8u>()) );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+ };
+ template<int SDEPTH, typename NppHistogramEvenFuncC4<SDEPTH>::func_ptr func, get_buf_size_c4_t get_buf_size>
+ struct NppHistogramEvenC4
+ {
+ typedef typename NppHistogramEvenFuncC4<SDEPTH>::src_t src_t;
+
+ static void hist(const GpuMat& src, GpuMat hist[4], GpuMat& buffer, int histSize[4], int lowerLevel[4], int upperLevel[4], cudaStream_t stream)
+ {
+ int levels[] = {histSize[0] + 1, histSize[1] + 1, histSize[2] + 1, histSize[3] + 1};
+ hist[0].create(1, histSize[0], CV_32S);
+ hist[1].create(1, histSize[1], CV_32S);
+ hist[2].create(1, histSize[2], CV_32S);
+ hist[3].create(1, histSize[3], CV_32S);
+
+ NppiSize sz;
+ sz.width = src.cols;
+ sz.height = src.rows;
+
+ Npp32s* pHist[] = {hist[0].ptr<Npp32s>(), hist[1].ptr<Npp32s>(), hist[2].ptr<Npp32s>(), hist[3].ptr<Npp32s>()};
+
+ int buf_size;
+ get_buf_size(sz, levels, &buf_size);
+
+ ensureSizeIsEnough(1, buf_size, CV_8U, buffer);
+
+ NppStreamHandler h(stream);
+
+ nppSafeCall( func(src.ptr<src_t>(), static_cast<int>(src.step), sz, pHist, levels, lowerLevel, upperLevel, buffer.ptr<Npp8u>()) );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+ };
+
+ template<int SDEPTH> struct NppHistogramRangeFuncC1
+ {
+ typedef typename NPPTypeTraits<SDEPTH>::npp_type src_t;
+ typedef Npp32s level_t;
+ enum {LEVEL_TYPE_CODE=CV_32SC1};
+
+ typedef NppStatus (*func_ptr)(const src_t* pSrc, int nSrcStep, NppiSize oSizeROI, Npp32s* pHist,
+ const Npp32s* pLevels, int nLevels, Npp8u* pBuffer);
+ };
+ template<> struct NppHistogramRangeFuncC1<CV_32F>
+ {
+ typedef Npp32f src_t;
+ typedef Npp32f level_t;
+ enum {LEVEL_TYPE_CODE=CV_32FC1};
+
+ typedef NppStatus (*func_ptr)(const Npp32f* pSrc, int nSrcStep, NppiSize oSizeROI, Npp32s* pHist,
+ const Npp32f* pLevels, int nLevels, Npp8u* pBuffer);
+ };
+ template<int SDEPTH> struct NppHistogramRangeFuncC4
+ {
+ typedef typename NPPTypeTraits<SDEPTH>::npp_type src_t;
+ typedef Npp32s level_t;
+ enum {LEVEL_TYPE_CODE=CV_32SC1};
+
+ typedef NppStatus (*func_ptr)(const src_t* pSrc, int nSrcStep, NppiSize oSizeROI, Npp32s* pHist[4],
+ const Npp32s* pLevels[4], int nLevels[4], Npp8u* pBuffer);
+ };
+ template<> struct NppHistogramRangeFuncC4<CV_32F>
+ {
+ typedef Npp32f src_t;
+ typedef Npp32f level_t;
+ enum {LEVEL_TYPE_CODE=CV_32FC1};
+
+ typedef NppStatus (*func_ptr)(const Npp32f* pSrc, int nSrcStep, NppiSize oSizeROI, Npp32s* pHist[4],
+ const Npp32f* pLevels[4], int nLevels[4], Npp8u* pBuffer);
+ };
+
+ template<int SDEPTH, typename NppHistogramRangeFuncC1<SDEPTH>::func_ptr func, get_buf_size_c1_t get_buf_size>
+ struct NppHistogramRangeC1
+ {
+ typedef typename NppHistogramRangeFuncC1<SDEPTH>::src_t src_t;
+ typedef typename NppHistogramRangeFuncC1<SDEPTH>::level_t level_t;
+ enum {LEVEL_TYPE_CODE=NppHistogramRangeFuncC1<SDEPTH>::LEVEL_TYPE_CODE};
+
+ static void hist(const GpuMat& src, GpuMat& hist, const GpuMat& levels, GpuMat& buffer, cudaStream_t stream)
+ {
+ CV_Assert(levels.type() == LEVEL_TYPE_CODE && levels.rows == 1);
+
+ hist.create(1, levels.cols - 1, CV_32S);
+
+ NppiSize sz;
+ sz.width = src.cols;
+ sz.height = src.rows;
+
+ int buf_size;
+ get_buf_size(sz, levels.cols, &buf_size);
+
+ ensureSizeIsEnough(1, buf_size, CV_8U, buffer);
+
+ NppStreamHandler h(stream);
+
+ nppSafeCall( func(src.ptr<src_t>(), static_cast<int>(src.step), sz, hist.ptr<Npp32s>(), levels.ptr<level_t>(), levels.cols, buffer.ptr<Npp8u>()) );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+ };
+ template<int SDEPTH, typename NppHistogramRangeFuncC4<SDEPTH>::func_ptr func, get_buf_size_c4_t get_buf_size>
+ struct NppHistogramRangeC4
+ {
+ typedef typename NppHistogramRangeFuncC4<SDEPTH>::src_t src_t;
+ typedef typename NppHistogramRangeFuncC1<SDEPTH>::level_t level_t;
+ enum {LEVEL_TYPE_CODE=NppHistogramRangeFuncC1<SDEPTH>::LEVEL_TYPE_CODE};
+
+ static void hist(const GpuMat& src, GpuMat hist[4], const GpuMat levels[4], GpuMat& buffer, cudaStream_t stream)
+ {
+ CV_Assert(levels[0].type() == LEVEL_TYPE_CODE && levels[0].rows == 1);
+ CV_Assert(levels[1].type() == LEVEL_TYPE_CODE && levels[1].rows == 1);
+ CV_Assert(levels[2].type() == LEVEL_TYPE_CODE && levels[2].rows == 1);
+ CV_Assert(levels[3].type() == LEVEL_TYPE_CODE && levels[3].rows == 1);
+
+ hist[0].create(1, levels[0].cols - 1, CV_32S);
+ hist[1].create(1, levels[1].cols - 1, CV_32S);
+ hist[2].create(1, levels[2].cols - 1, CV_32S);
+ hist[3].create(1, levels[3].cols - 1, CV_32S);
+
+ Npp32s* pHist[] = {hist[0].ptr<Npp32s>(), hist[1].ptr<Npp32s>(), hist[2].ptr<Npp32s>(), hist[3].ptr<Npp32s>()};
+ int nLevels[] = {levels[0].cols, levels[1].cols, levels[2].cols, levels[3].cols};
+ const level_t* pLevels[] = {levels[0].ptr<level_t>(), levels[1].ptr<level_t>(), levels[2].ptr<level_t>(), levels[3].ptr<level_t>()};
+
+ NppiSize sz;
+ sz.width = src.cols;
+ sz.height = src.rows;
+
+ int buf_size;
+ get_buf_size(sz, nLevels, &buf_size);
+
+ ensureSizeIsEnough(1, buf_size, CV_8U, buffer);
+
+ NppStreamHandler h(stream);
+
+ nppSafeCall( func(src.ptr<src_t>(), static_cast<int>(src.step), sz, pHist, pLevels, nLevels, buffer.ptr<Npp8u>()) );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+ };
+}
+
+void cv::gpu::evenLevels(GpuMat& levels, int nLevels, int lowerLevel, int upperLevel)
+{
+ Mat host_levels(1, nLevels, CV_32SC1);
+ nppSafeCall( nppiEvenLevelsHost_32s(host_levels.ptr<Npp32s>(), nLevels, lowerLevel, upperLevel) );
+ levels.upload(host_levels);
+}
+
+void cv::gpu::histEven(const GpuMat& src, GpuMat& hist, int histSize, int lowerLevel, int upperLevel, Stream& stream)
+{
+ GpuMat buf;
+ histEven(src, hist, buf, histSize, lowerLevel, upperLevel, stream);
+}
+
+void cv::gpu::histEven(const GpuMat& src, GpuMat& hist, GpuMat& buf, int histSize, int lowerLevel, int upperLevel, Stream& stream)
+{
+ CV_Assert(src.type() == CV_8UC1 || src.type() == CV_16UC1 || src.type() == CV_16SC1 );
+
+ typedef void (*hist_t)(const GpuMat& src, GpuMat& hist, GpuMat& buf, int levels, int lowerLevel, int upperLevel, cudaStream_t stream);
+ static const hist_t hist_callers[] =
+ {
+ NppHistogramEvenC1<CV_8U , nppiHistogramEven_8u_C1R , nppiHistogramEvenGetBufferSize_8u_C1R >::hist,
+ 0,
+ NppHistogramEvenC1<CV_16U, nppiHistogramEven_16u_C1R, nppiHistogramEvenGetBufferSize_16u_C1R>::hist,
+ NppHistogramEvenC1<CV_16S, nppiHistogramEven_16s_C1R, nppiHistogramEvenGetBufferSize_16s_C1R>::hist
+ };
+
+ hist_callers[src.depth()](src, hist, buf, histSize, lowerLevel, upperLevel, StreamAccessor::getStream(stream));
+}
+
+void cv::gpu::histEven(const GpuMat& src, GpuMat hist[4], int histSize[4], int lowerLevel[4], int upperLevel[4], Stream& stream)
+{
+ GpuMat buf;
+ histEven(src, hist, buf, histSize, lowerLevel, upperLevel, stream);
+}
+
+void cv::gpu::histEven(const GpuMat& src, GpuMat hist[4], GpuMat& buf, int histSize[4], int lowerLevel[4], int upperLevel[4], Stream& stream)
+{
+ CV_Assert(src.type() == CV_8UC4 || src.type() == CV_16UC4 || src.type() == CV_16SC4 );
+
+ typedef void (*hist_t)(const GpuMat& src, GpuMat hist[4], GpuMat& buf, int levels[4], int lowerLevel[4], int upperLevel[4], cudaStream_t stream);
+ static const hist_t hist_callers[] =
+ {
+ NppHistogramEvenC4<CV_8U , nppiHistogramEven_8u_C4R , nppiHistogramEvenGetBufferSize_8u_C4R >::hist,
+ 0,
+ NppHistogramEvenC4<CV_16U, nppiHistogramEven_16u_C4R, nppiHistogramEvenGetBufferSize_16u_C4R>::hist,
+ NppHistogramEvenC4<CV_16S, nppiHistogramEven_16s_C4R, nppiHistogramEvenGetBufferSize_16s_C4R>::hist
+ };
+
+ hist_callers[src.depth()](src, hist, buf, histSize, lowerLevel, upperLevel, StreamAccessor::getStream(stream));
+}
+
+void cv::gpu::histRange(const GpuMat& src, GpuMat& hist, const GpuMat& levels, Stream& stream)
+{
+ GpuMat buf;
+ histRange(src, hist, levels, buf, stream);
+}
+
+void cv::gpu::histRange(const GpuMat& src, GpuMat& hist, const GpuMat& levels, GpuMat& buf, Stream& stream)
+{
+ CV_Assert(src.type() == CV_8UC1 || src.type() == CV_16UC1 || src.type() == CV_16SC1 || src.type() == CV_32FC1);
+
+ typedef void (*hist_t)(const GpuMat& src, GpuMat& hist, const GpuMat& levels, GpuMat& buf, cudaStream_t stream);
+ static const hist_t hist_callers[] =
+ {
+ NppHistogramRangeC1<CV_8U , nppiHistogramRange_8u_C1R , nppiHistogramRangeGetBufferSize_8u_C1R >::hist,
+ 0,
+ NppHistogramRangeC1<CV_16U, nppiHistogramRange_16u_C1R, nppiHistogramRangeGetBufferSize_16u_C1R>::hist,
+ NppHistogramRangeC1<CV_16S, nppiHistogramRange_16s_C1R, nppiHistogramRangeGetBufferSize_16s_C1R>::hist,
+ 0,
+ NppHistogramRangeC1<CV_32F, nppiHistogramRange_32f_C1R, nppiHistogramRangeGetBufferSize_32f_C1R>::hist
+ };
+
+ hist_callers[src.depth()](src, hist, levels, buf, StreamAccessor::getStream(stream));
+}
+
+void cv::gpu::histRange(const GpuMat& src, GpuMat hist[4], const GpuMat levels[4], Stream& stream)
+{
+ GpuMat buf;
+ histRange(src, hist, levels, buf, stream);
+}
+
+void cv::gpu::histRange(const GpuMat& src, GpuMat hist[4], const GpuMat levels[4], GpuMat& buf, Stream& stream)
+{
+ CV_Assert(src.type() == CV_8UC4 || src.type() == CV_16UC4 || src.type() == CV_16SC4 || src.type() == CV_32FC4);
+
+ typedef void (*hist_t)(const GpuMat& src, GpuMat hist[4], const GpuMat levels[4], GpuMat& buf, cudaStream_t stream);
+ static const hist_t hist_callers[] =
+ {
+ NppHistogramRangeC4<CV_8U , nppiHistogramRange_8u_C4R , nppiHistogramRangeGetBufferSize_8u_C4R >::hist,
+ 0,
+ NppHistogramRangeC4<CV_16U, nppiHistogramRange_16u_C4R, nppiHistogramRangeGetBufferSize_16u_C4R>::hist,
+ NppHistogramRangeC4<CV_16S, nppiHistogramRange_16s_C4R, nppiHistogramRangeGetBufferSize_16s_C4R>::hist,
+ 0,
+ NppHistogramRangeC4<CV_32F, nppiHistogramRange_32f_C4R, nppiHistogramRangeGetBufferSize_32f_C4R>::hist
+ };
+
+ hist_callers[src.depth()](src, hist, levels, buf, StreamAccessor::getStream(stream));
+}
+
+////////////////////////////////////////////////////////////////////////
+// calcHist
+
+namespace hist
+{
+ void histogram256(PtrStepSzb src, int* hist, cudaStream_t stream);
+}
+
+void cv::gpu::calcHist(const GpuMat& src, GpuMat& hist, Stream& stream)
+{
+ CV_Assert(src.type() == CV_8UC1);
+
+ hist.create(1, 256, CV_32SC1);
+ hist.setTo(Scalar::all(0));
+
+ hist::histogram256(src, hist.ptr<int>(), StreamAccessor::getStream(stream));
+}
+
+////////////////////////////////////////////////////////////////////////
+// equalizeHist
+
+namespace hist
+{
+ void equalizeHist(PtrStepSzb src, PtrStepSzb dst, const int* lut, cudaStream_t stream);
+}
+
+void cv::gpu::equalizeHist(const GpuMat& src, GpuMat& dst, Stream& stream)
+{
+ GpuMat hist;
+ GpuMat buf;
+ equalizeHist(src, dst, hist, buf, stream);
+}
+
+void cv::gpu::equalizeHist(const GpuMat& src, GpuMat& dst, GpuMat& hist, GpuMat& buf, Stream& s)
+{
+ CV_Assert(src.type() == CV_8UC1);
+
+ dst.create(src.size(), src.type());
+
+ int intBufSize;
+ nppSafeCall( nppsIntegralGetBufferSize_32s(256, &intBufSize) );
+
+ ensureSizeIsEnough(1, intBufSize + 256 * sizeof(int), CV_8UC1, buf);
+
+ GpuMat intBuf(1, intBufSize, CV_8UC1, buf.ptr());
+ GpuMat lut(1, 256, CV_32S, buf.ptr() + intBufSize);
+
+ calcHist(src, hist, s);
+
+ cudaStream_t stream = StreamAccessor::getStream(s);
+
+ NppStreamHandler h(stream);
+
+ nppSafeCall( nppsIntegral_32s(hist.ptr<Npp32s>(), lut.ptr<Npp32s>(), 256, intBuf.ptr<Npp8u>()) );
+
+ hist::equalizeHist(src, dst, lut.ptr<int>(), stream);
+}
+
+////////////////////////////////////////////////////////////////////////
+// CLAHE
+
+namespace clahe
+{
+ void calcLut(PtrStepSzb src, PtrStepb lut, int tilesX, int tilesY, int2 tileSize, int clipLimit, float lutScale, cudaStream_t stream);
+ void transform(PtrStepSzb src, PtrStepSzb dst, PtrStepb lut, int tilesX, int tilesY, int2 tileSize, cudaStream_t stream);
+}
+
+namespace
+{
+ class CLAHE_Impl : public cv::gpu::CLAHE
+ {
+ public:
+ CLAHE_Impl(double clipLimit = 40.0, int tilesX = 8, int tilesY = 8);
+
+ cv::AlgorithmInfo* info() const;
+
+ void apply(cv::InputArray src, cv::OutputArray dst);
+ void apply(InputArray src, OutputArray dst, Stream& stream);
+
+ void setClipLimit(double clipLimit);
+ double getClipLimit() const;
+
+ void setTilesGridSize(cv::Size tileGridSize);
+ cv::Size getTilesGridSize() const;
+
+ void collectGarbage();
+
+ private:
+ double clipLimit_;
+ int tilesX_;
+ int tilesY_;
+
+ GpuMat srcExt_;
+ GpuMat lut_;
+ };
+
+ CLAHE_Impl::CLAHE_Impl(double clipLimit, int tilesX, int tilesY) :
+ clipLimit_(clipLimit), tilesX_(tilesX), tilesY_(tilesY)
+ {
+ }
+
+ CV_INIT_ALGORITHM(CLAHE_Impl, "CLAHE_GPU",
+ obj.info()->addParam(obj, "clipLimit", obj.clipLimit_);
+ obj.info()->addParam(obj, "tilesX", obj.tilesX_);
+ obj.info()->addParam(obj, "tilesY", obj.tilesY_))
+
+ void CLAHE_Impl::apply(cv::InputArray _src, cv::OutputArray _dst)
+ {
+ apply(_src, _dst, Stream::Null());
+ }
+
+ void CLAHE_Impl::apply(InputArray _src, OutputArray _dst, Stream& s)
+ {
+ GpuMat src = _src.getGpuMat();
+
+ CV_Assert( src.type() == CV_8UC1 );
+
+ _dst.create( src.size(), src.type() );
+ GpuMat dst = _dst.getGpuMat();
+
+ const int histSize = 256;
+
+ ensureSizeIsEnough(tilesX_ * tilesY_, histSize, CV_8UC1, lut_);
+
+ cudaStream_t stream = StreamAccessor::getStream(s);
+
+ cv::Size tileSize;
+ GpuMat srcForLut;
+
+ if (src.cols % tilesX_ == 0 && src.rows % tilesY_ == 0)
+ {
+ tileSize = cv::Size(src.cols / tilesX_, src.rows / tilesY_);
+ srcForLut = src;
+ }
+ else
+ {
+#ifndef HAVE_OPENCV_GPUARITHM
+ throw_no_cuda();
+#else
+ cv::gpu::copyMakeBorder(src, srcExt_, 0, tilesY_ - (src.rows % tilesY_), 0, tilesX_ - (src.cols % tilesX_), cv::BORDER_REFLECT_101, cv::Scalar(), s);
+#endif
+
+ tileSize = cv::Size(srcExt_.cols / tilesX_, srcExt_.rows / tilesY_);
+ srcForLut = srcExt_;
+ }
+
+ const int tileSizeTotal = tileSize.area();
+ const float lutScale = static_cast<float>(histSize - 1) / tileSizeTotal;
+
+ int clipLimit = 0;
+ if (clipLimit_ > 0.0)
+ {
+ clipLimit = static_cast<int>(clipLimit_ * tileSizeTotal / histSize);
+ clipLimit = std::max(clipLimit, 1);
+ }
+
+ clahe::calcLut(srcForLut, lut_, tilesX_, tilesY_, make_int2(tileSize.width, tileSize.height), clipLimit, lutScale, stream);
+
+ clahe::transform(src, dst, lut_, tilesX_, tilesY_, make_int2(tileSize.width, tileSize.height), stream);
+ }
+
+ void CLAHE_Impl::setClipLimit(double clipLimit)
+ {
+ clipLimit_ = clipLimit;
+ }
+
+ double CLAHE_Impl::getClipLimit() const
+ {
+ return clipLimit_;
+ }
+
+ void CLAHE_Impl::setTilesGridSize(cv::Size tileGridSize)
+ {
+ tilesX_ = tileGridSize.width;
+ tilesY_ = tileGridSize.height;
+ }
+
+ cv::Size CLAHE_Impl::getTilesGridSize() const
+ {
+ return cv::Size(tilesX_, tilesY_);
+ }
+
+ void CLAHE_Impl::collectGarbage()
+ {
+ srcExt_.release();
+ lut_.release();
+ }
+}
+
+cv::Ptr<cv::gpu::CLAHE> cv::gpu::createCLAHE(double clipLimit, cv::Size tileGridSize)
+{
+ return new CLAHE_Impl(clipLimit, tileGridSize.width, tileGridSize.height);
+}
+
+#endif /* !defined (HAVE_CUDA) */
using namespace cv;
using namespace cv::gpu;
-#if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
+#if !defined (HAVE_CUDA) || !defined (HAVE_OPENCV_GPUARITHM) || defined (CUDA_DISABLER)
void cv::gpu::matchTemplate(const GpuMat&, const GpuMat&, GpuMat&, int, Stream&) { throw_no_cuda(); }
return;
}
- ConvolveBuf convolve_buf;
+ gpu::ConvolveBuf convolve_buf;
convolve_buf.user_block_size = buf.user_block_size;
if (image.channels() == 1)
- convolve(image.reshape(1), templ.reshape(1), result, true, convolve_buf, stream);
+ gpu::convolve(image.reshape(1), templ.reshape(1), result, true, convolve_buf, stream);
else
{
GpuMat result_;
- convolve(image.reshape(1), templ.reshape(1), result_, true, convolve_buf, stream);
+ gpu::convolve(image.reshape(1), templ.reshape(1), result_, true, convolve_buf, stream);
extractFirstChannel_32F(result_, result, image.channels(), StreamAccessor::getStream(stream));
}
}
matchTemplate_CCORR_8U(image, templ, result, buf, stream);
buf.image_sqsums.resize(1);
- sqrIntegral(image.reshape(1), buf.image_sqsums[0], stream);
+ gpu::sqrIntegral(image.reshape(1), buf.image_sqsums[0], stream);
- unsigned long long templ_sqsum = (unsigned long long)sqrSum(templ.reshape(1))[0];
+ unsigned long long templ_sqsum = (unsigned long long)gpu::sqrSum(templ.reshape(1))[0];
normalize_8U(templ.cols, templ.rows, buf.image_sqsums[0], templ_sqsum, result, image.channels(), StreamAccessor::getStream(stream));
}
}
buf.image_sqsums.resize(1);
- sqrIntegral(image.reshape(1), buf.image_sqsums[0], stream);
+ gpu::sqrIntegral(image.reshape(1), buf.image_sqsums[0], stream);
- unsigned long long templ_sqsum = (unsigned long long)sqrSum(templ.reshape(1))[0];
+ unsigned long long templ_sqsum = (unsigned long long)gpu::sqrSum(templ.reshape(1))[0];
matchTemplate_CCORR_8U(image, templ, result, buf, stream);
matchTemplatePrepared_SQDIFF_8U(templ.cols, templ.rows, buf.image_sqsums[0], templ_sqsum, result, image.channels(), StreamAccessor::getStream(stream));
const GpuMat& image, const GpuMat& templ, GpuMat& result, MatchTemplateBuf &buf, Stream& stream)
{
buf.image_sqsums.resize(1);
- sqrIntegral(image.reshape(1), buf.image_sqsums[0], stream);
+ gpu::sqrIntegral(image.reshape(1), buf.image_sqsums[0], stream);
- unsigned long long templ_sqsum = (unsigned long long)sqrSum(templ.reshape(1))[0];
+ unsigned long long templ_sqsum = (unsigned long long)gpu::sqrSum(templ.reshape(1))[0];
matchTemplate_CCORR_8U(image, templ, result, buf, stream);
matchTemplatePrepared_SQDIFF_NORMED_8U(templ.cols, templ.rows, buf.image_sqsums[0], templ_sqsum, result, image.channels(), StreamAccessor::getStream(stream));
if (image.channels() == 1)
{
buf.image_sums.resize(1);
- integral(image, buf.image_sums[0], stream);
+ gpu::integral(image, buf.image_sums[0], stream);
unsigned int templ_sum = (unsigned int)sum(templ)[0];
matchTemplatePrepared_CCOFF_8U(templ.cols, templ.rows, buf.image_sums[0], templ_sum, result, StreamAccessor::getStream(stream));
}
else
{
- split(image, buf.images);
+ gpu::split(image, buf.images);
buf.image_sums.resize(buf.images.size());
for (int i = 0; i < image.channels(); ++i)
- integral(buf.images[i], buf.image_sums[i], stream);
+ gpu::integral(buf.images[i], buf.image_sums[i], stream);
- Scalar templ_sum = sum(templ);
+ Scalar templ_sum = gpu::sum(templ);
switch (image.channels())
{
if (image.channels() == 1)
{
buf.image_sums.resize(1);
- integral(image, buf.image_sums[0], stream);
+ gpu::integral(image, buf.image_sums[0], stream);
buf.image_sqsums.resize(1);
- sqrIntegral(image, buf.image_sqsums[0], stream);
+ gpu::sqrIntegral(image, buf.image_sqsums[0], stream);
- unsigned int templ_sum = (unsigned int)sum(templ)[0];
- unsigned long long templ_sqsum = (unsigned long long)sqrSum(templ)[0];
+ unsigned int templ_sum = (unsigned int)gpu::sum(templ)[0];
+ unsigned long long templ_sqsum = (unsigned long long)gpu::sqrSum(templ)[0];
matchTemplatePrepared_CCOFF_NORMED_8U(
templ.cols, templ.rows, buf.image_sums[0], buf.image_sqsums[0],
}
else
{
- split(image, buf.images);
+ gpu::split(image, buf.images);
buf.image_sums.resize(buf.images.size());
buf.image_sqsums.resize(buf.images.size());
for (int i = 0; i < image.channels(); ++i)
{
- integral(buf.images[i], buf.image_sums[i], stream);
- sqrIntegral(buf.images[i], buf.image_sqsums[i], stream);
+ gpu::integral(buf.images[i], buf.image_sums[i], stream);
+ gpu::sqrIntegral(buf.images[i], buf.image_sqsums[i], stream);
}
- Scalar templ_sum = sum(templ);
- Scalar templ_sqsum = sqrSum(templ);
+ Scalar templ_sum = gpu::sum(templ);
+ Scalar templ_sqsum = gpu::sqrSum(templ);
switch (image.channels())
{
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "precomp.hpp"
+
+using namespace cv;
+using namespace cv::gpu;
+
+#if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
+
+void cv::gpu::meanShiftFiltering(const GpuMat&, GpuMat&, int, int, TermCriteria, Stream&) { throw_no_cuda(); }
+void cv::gpu::meanShiftProc(const GpuMat&, GpuMat&, GpuMat&, int, int, TermCriteria, Stream&) { throw_no_cuda(); }
+
+#else /* !defined (HAVE_CUDA) */
+
+////////////////////////////////////////////////////////////////////////
+// meanShiftFiltering_GPU
+
+namespace cv { namespace gpu { namespace cudev
+{
+ namespace imgproc
+ {
+ void meanShiftFiltering_gpu(const PtrStepSzb& src, PtrStepSzb dst, int sp, int sr, int maxIter, float eps, cudaStream_t stream);
+ }
+}}}
+
+void cv::gpu::meanShiftFiltering(const GpuMat& src, GpuMat& dst, int sp, int sr, TermCriteria criteria, Stream& stream)
+{
+ using namespace ::cv::gpu::cudev::imgproc;
+
+ if( src.empty() )
+ CV_Error( cv::Error::StsBadArg, "The input image is empty" );
+
+ if( src.depth() != CV_8U || src.channels() != 4 )
+ CV_Error( cv::Error::StsUnsupportedFormat, "Only 8-bit, 4-channel images are supported" );
+
+ dst.create( src.size(), CV_8UC4 );
+
+ if( !(criteria.type & TermCriteria::MAX_ITER) )
+ criteria.maxCount = 5;
+
+ int maxIter = std::min(std::max(criteria.maxCount, 1), 100);
+
+ float eps;
+ if( !(criteria.type & TermCriteria::EPS) )
+ eps = 1.f;
+ eps = (float)std::max(criteria.epsilon, 0.0);
+
+ meanShiftFiltering_gpu(src, dst, sp, sr, maxIter, eps, StreamAccessor::getStream(stream));
+}
+
+////////////////////////////////////////////////////////////////////////
+// meanShiftProc_GPU
+
+namespace cv { namespace gpu { namespace cudev
+{
+ namespace imgproc
+ {
+ void meanShiftProc_gpu(const PtrStepSzb& src, PtrStepSzb dstr, PtrStepSzb dstsp, int sp, int sr, int maxIter, float eps, cudaStream_t stream);
+ }
+}}}
+
+void cv::gpu::meanShiftProc(const GpuMat& src, GpuMat& dstr, GpuMat& dstsp, int sp, int sr, TermCriteria criteria, Stream& stream)
+{
+ using namespace ::cv::gpu::cudev::imgproc;
+
+ if( src.empty() )
+ CV_Error( cv::Error::StsBadArg, "The input image is empty" );
+
+ if( src.depth() != CV_8U || src.channels() != 4 )
+ CV_Error( cv::Error::StsUnsupportedFormat, "Only 8-bit, 4-channel images are supported" );
+
+ dstr.create( src.size(), CV_8UC4 );
+ dstsp.create( src.size(), CV_16SC2 );
+
+ if( !(criteria.type & TermCriteria::MAX_ITER) )
+ criteria.maxCount = 5;
+
+ int maxIter = std::min(std::max(criteria.maxCount, 1), 100);
+
+ float eps;
+ if( !(criteria.type & TermCriteria::EPS) )
+ eps = 1.f;
+ eps = (float)std::max(criteria.epsilon, 0.0);
+
+ meanShiftProc_gpu(src, dstr, dstsp, sp, sr, maxIter, eps, StreamAccessor::getStream(stream));
+}
+
+#endif /* !defined (HAVE_CUDA) */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_PRECOMP_H__
+#define __OPENCV_PRECOMP_H__
+
+#include "opencv2/gpuimgproc.hpp"
+#include "opencv2/gpufilters.hpp"
+
+#include "opencv2/core/private.hpp"
+#include "opencv2/core/gpu_private.hpp"
+
+#include "opencv2/opencv_modules.hpp"
+
+#ifdef HAVE_OPENCV_GPUARITHM
+# include "opencv2/gpuarithm.hpp"
+#endif
+
+#endif /* __OPENCV_PRECOMP_H__ */
using namespace cvtest;
-CV_ENUM(ThreshOp, THRESH_BINARY, THRESH_BINARY_INV, THRESH_TRUNC, THRESH_TOZERO, THRESH_TOZERO_INV)
+////////////////////////////////////////////////////////
+// BilateralFilter
-PARAM_TEST_CASE(Threshold, cv::gpu::DeviceInfo, cv::Size, MatType, ThreshOp, UseRoi)
+PARAM_TEST_CASE(BilateralFilter, cv::gpu::DeviceInfo, cv::Size, MatType)
{
cv::gpu::DeviceInfo devInfo;
cv::Size size;
int type;
- int threshOp;
- bool useRoi;
+ int kernel_size;
+ float sigma_color;
+ float sigma_spatial;
virtual void SetUp()
{
devInfo = GET_PARAM(0);
size = GET_PARAM(1);
type = GET_PARAM(2);
- threshOp = GET_PARAM(3);
- useRoi = GET_PARAM(4);
+
+ kernel_size = 5;
+ sigma_color = 10.f;
+ sigma_spatial = 3.5f;
cv::gpu::setDevice(devInfo.deviceID());
}
};
-GPU_TEST_P(Threshold, Accuracy)
+GPU_TEST_P(BilateralFilter, Accuracy)
{
cv::Mat src = randomMat(size, type);
- double maxVal = randomDouble(20.0, 127.0);
- double thresh = randomDouble(0.0, maxVal);
- cv::gpu::GpuMat dst = createMat(src.size(), src.type(), useRoi);
- cv::gpu::threshold(loadMat(src, useRoi), dst, thresh, maxVal, threshOp);
+ src.convertTo(src, type);
+ cv::gpu::GpuMat dst;
+
+ cv::gpu::bilateralFilter(loadMat(src), dst, kernel_size, sigma_color, sigma_spatial);
cv::Mat dst_gold;
- cv::threshold(src, dst_gold, thresh, maxVal, threshOp);
+ cv::bilateralFilter(src, dst_gold, kernel_size, sigma_color, sigma_spatial);
- EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
+ EXPECT_MAT_NEAR(dst_gold, dst, src.depth() == CV_32F ? 1e-3 : 1.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, Threshold, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_ImgProc, BilateralFilter, testing::Combine(
ALL_DEVICES,
- DIFFERENT_SIZES,
- testing::Values(MatType(CV_8UC1), MatType(CV_16SC1), MatType(CV_32FC1)),
- ThreshOp::all(),
- WHOLE_SUBMAT));
+ testing::Values(cv::Size(128, 128), cv::Size(113, 113), cv::Size(639, 481)),
+ testing::Values(MatType(CV_8UC1), MatType(CV_8UC3), MatType(CV_32FC1), MatType(CV_32FC3))
+ ));
+
#endif // HAVE_CUDA
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+#ifdef HAVE_CUDA
+
+using namespace cvtest;
+
+////////////////////////////////////////////////////////////////////////////
+// Blend
+
+namespace
+{
+ template <typename T>
+ void blendLinearGold(const cv::Mat& img1, const cv::Mat& img2, const cv::Mat& weights1, const cv::Mat& weights2, cv::Mat& result_gold)
+ {
+ result_gold.create(img1.size(), img1.type());
+
+ int cn = img1.channels();
+
+ for (int y = 0; y < img1.rows; ++y)
+ {
+ const float* weights1_row = weights1.ptr<float>(y);
+ const float* weights2_row = weights2.ptr<float>(y);
+ const T* img1_row = img1.ptr<T>(y);
+ const T* img2_row = img2.ptr<T>(y);
+ T* result_gold_row = result_gold.ptr<T>(y);
+
+ for (int x = 0; x < img1.cols * cn; ++x)
+ {
+ float w1 = weights1_row[x / cn];
+ float w2 = weights2_row[x / cn];
+ result_gold_row[x] = static_cast<T>((img1_row[x] * w1 + img2_row[x] * w2) / (w1 + w2 + 1e-5f));
+ }
+ }
+ }
+}
+
+PARAM_TEST_CASE(Blend, cv::gpu::DeviceInfo, cv::Size, MatType, UseRoi)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ int type;
+ bool useRoi;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ type = GET_PARAM(2);
+ useRoi = GET_PARAM(3);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(Blend, Accuracy)
+{
+ int depth = CV_MAT_DEPTH(type);
+
+ cv::Mat img1 = randomMat(size, type, 0.0, depth == CV_8U ? 255.0 : 1.0);
+ cv::Mat img2 = randomMat(size, type, 0.0, depth == CV_8U ? 255.0 : 1.0);
+ cv::Mat weights1 = randomMat(size, CV_32F, 0, 1);
+ cv::Mat weights2 = randomMat(size, CV_32F, 0, 1);
+
+ cv::gpu::GpuMat result;
+ cv::gpu::blendLinear(loadMat(img1, useRoi), loadMat(img2, useRoi), loadMat(weights1, useRoi), loadMat(weights2, useRoi), result);
+
+ cv::Mat result_gold;
+ if (depth == CV_8U)
+ blendLinearGold<uchar>(img1, img2, weights1, weights2, result_gold);
+ else
+ blendLinearGold<float>(img1, img2, weights1, weights2, result_gold);
+
+ EXPECT_MAT_NEAR(result_gold, result, CV_MAT_DEPTH(type) == CV_8U ? 1.0 : 1e-5);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_ImgProc, Blend, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ testing::Values(MatType(CV_8UC1), MatType(CV_8UC3), MatType(CV_8UC4), MatType(CV_32FC1), MatType(CV_32FC3), MatType(CV_32FC4)),
+ WHOLE_SUBMAT));
+
+#endif // HAVE_CUDA
using namespace cvtest;
+////////////////////////////////////////////////////////
+// Canny
+
namespace
{
- IMPLEMENT_PARAM_CLASS(Border, int)
+ IMPLEMENT_PARAM_CLASS(AppertureSize, int)
+ IMPLEMENT_PARAM_CLASS(L2gradient, bool)
}
-PARAM_TEST_CASE(CopyMakeBorder, cv::gpu::DeviceInfo, cv::Size, MatType, Border, BorderType, UseRoi)
+PARAM_TEST_CASE(Canny, cv::gpu::DeviceInfo, AppertureSize, L2gradient, UseRoi)
{
cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- int type;
- int border;
- int borderType;
+ int apperture_size;
+ bool useL2gradient;
bool useRoi;
virtual void SetUp()
{
devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- type = GET_PARAM(2);
- border = GET_PARAM(3);
- borderType = GET_PARAM(4);
- useRoi = GET_PARAM(5);
+ apperture_size = GET_PARAM(1);
+ useL2gradient = GET_PARAM(2);
+ useRoi = GET_PARAM(3);
cv::gpu::setDevice(devInfo.deviceID());
}
};
-GPU_TEST_P(CopyMakeBorder, Accuracy)
+GPU_TEST_P(Canny, Accuracy)
{
- cv::Mat src = randomMat(size, type);
- cv::Scalar val = randomScalar(0, 255);
+ cv::Mat img = readImage("stereobm/aloe-L.png", cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(img.empty());
+
+ double low_thresh = 50.0;
+ double high_thresh = 100.0;
- cv::gpu::GpuMat dst = createMat(cv::Size(size.width + 2 * border, size.height + 2 * border), type, useRoi);
- cv::gpu::copyMakeBorder(loadMat(src, useRoi), dst, border, border, border, border, borderType, val);
+ if (!supportFeature(devInfo, cv::gpu::SHARED_ATOMICS))
+ {
+ try
+ {
+ cv::gpu::GpuMat edges;
+ cv::gpu::Canny(loadMat(img), edges, low_thresh, high_thresh, apperture_size, useL2gradient);
+ }
+ catch (const cv::Exception& e)
+ {
+ ASSERT_EQ(cv::Error::StsNotImplemented, e.code);
+ }
+ }
+ else
+ {
+ cv::gpu::GpuMat edges;
+ cv::gpu::Canny(loadMat(img, useRoi), edges, low_thresh, high_thresh, apperture_size, useL2gradient);
- cv::Mat dst_gold;
- cv::copyMakeBorder(src, dst_gold, border, border, border, border, borderType, val);
+ cv::Mat edges_gold;
+ cv::Canny(img, edges_gold, low_thresh, high_thresh, apperture_size, useL2gradient);
- EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
+ EXPECT_MAT_SIMILAR(edges_gold, edges, 2e-2);
+ }
}
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, CopyMakeBorder, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_ImgProc, Canny, testing::Combine(
ALL_DEVICES,
- DIFFERENT_SIZES,
- testing::Values(MatType(CV_8UC1),
- MatType(CV_8UC3),
- MatType(CV_8UC4),
- MatType(CV_16UC1),
- MatType(CV_16UC3),
- MatType(CV_16UC4),
- MatType(CV_32FC1),
- MatType(CV_32FC3),
- MatType(CV_32FC4)),
- testing::Values(Border(1), Border(10), Border(50)),
- ALL_BORDER_TYPES,
+ testing::Values(AppertureSize(3), AppertureSize(5)),
+ testing::Values(L2gradient(false), L2gradient(true)),
WHOLE_SUBMAT));
#endif // HAVE_CUDA
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+#ifdef HAVE_CUDA
+
+using namespace cvtest;
+
+///////////////////////////////////////////////////////////////////////////////////////////////////////
+// CornerHarris
+
+namespace
+{
+ IMPLEMENT_PARAM_CLASS(BlockSize, int);
+ IMPLEMENT_PARAM_CLASS(ApertureSize, int);
+}
+
+PARAM_TEST_CASE(CornerHarris, cv::gpu::DeviceInfo, MatType, BorderType, BlockSize, ApertureSize)
+{
+ cv::gpu::DeviceInfo devInfo;
+ int type;
+ int borderType;
+ int blockSize;
+ int apertureSize;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ type = GET_PARAM(1);
+ borderType = GET_PARAM(2);
+ blockSize = GET_PARAM(3);
+ apertureSize = GET_PARAM(4);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(CornerHarris, Accuracy)
+{
+ cv::Mat src = readImageType("stereobm/aloe-L.png", type);
+ ASSERT_FALSE(src.empty());
+
+ double k = randomDouble(0.1, 0.9);
+
+ cv::gpu::GpuMat dst;
+ cv::gpu::cornerHarris(loadMat(src), dst, blockSize, apertureSize, k, borderType);
+
+ cv::Mat dst_gold;
+ cv::cornerHarris(src, dst_gold, blockSize, apertureSize, k, borderType);
+
+ EXPECT_MAT_NEAR(dst_gold, dst, 0.02);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_ImgProc, CornerHarris, testing::Combine(
+ ALL_DEVICES,
+ testing::Values(MatType(CV_8UC1), MatType(CV_32FC1)),
+ testing::Values(BorderType(cv::BORDER_REFLECT101), BorderType(cv::BORDER_REPLICATE), BorderType(cv::BORDER_REFLECT)),
+ testing::Values(BlockSize(3), BlockSize(5), BlockSize(7)),
+ testing::Values(ApertureSize(0), ApertureSize(3), ApertureSize(5), ApertureSize(7))));
+
+///////////////////////////////////////////////////////////////////////////////////////////////////////
+// cornerMinEigen
+
+PARAM_TEST_CASE(CornerMinEigen, cv::gpu::DeviceInfo, MatType, BorderType, BlockSize, ApertureSize)
+{
+ cv::gpu::DeviceInfo devInfo;
+ int type;
+ int borderType;
+ int blockSize;
+ int apertureSize;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ type = GET_PARAM(1);
+ borderType = GET_PARAM(2);
+ blockSize = GET_PARAM(3);
+ apertureSize = GET_PARAM(4);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(CornerMinEigen, Accuracy)
+{
+ cv::Mat src = readImageType("stereobm/aloe-L.png", type);
+ ASSERT_FALSE(src.empty());
+
+ cv::gpu::GpuMat dst;
+ cv::gpu::cornerMinEigenVal(loadMat(src), dst, blockSize, apertureSize, borderType);
+
+ cv::Mat dst_gold;
+ cv::cornerMinEigenVal(src, dst_gold, blockSize, apertureSize, borderType);
+
+ EXPECT_MAT_NEAR(dst_gold, dst, 0.02);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_ImgProc, CornerMinEigen, testing::Combine(
+ ALL_DEVICES,
+ testing::Values(MatType(CV_8UC1), MatType(CV_32FC1)),
+ testing::Values(BorderType(cv::BORDER_REFLECT101), BorderType(cv::BORDER_REPLICATE), BorderType(cv::BORDER_REFLECT)),
+ testing::Values(BlockSize(3), BlockSize(5), BlockSize(7)),
+ testing::Values(ApertureSize(0), ApertureSize(3), ApertureSize(5), ApertureSize(7))));
+
+#endif // HAVE_CUDA
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+#ifdef HAVE_CUDA
+
+using namespace cvtest;
+
+//////////////////////////////////////////////////////
+// GoodFeaturesToTrack
+
+namespace
+{
+ IMPLEMENT_PARAM_CLASS(MinDistance, double)
+}
+
+PARAM_TEST_CASE(GoodFeaturesToTrack, cv::gpu::DeviceInfo, MinDistance)
+{
+ cv::gpu::DeviceInfo devInfo;
+ double minDistance;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ minDistance = GET_PARAM(1);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(GoodFeaturesToTrack, Accuracy)
+{
+ cv::Mat image = readImage("opticalflow/frame0.png", cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(image.empty());
+
+ int maxCorners = 1000;
+ double qualityLevel = 0.01;
+
+ cv::gpu::GoodFeaturesToTrackDetector_GPU detector(maxCorners, qualityLevel, minDistance);
+
+ cv::gpu::GpuMat d_pts;
+ detector(loadMat(image), d_pts);
+
+ ASSERT_FALSE(d_pts.empty());
+
+ std::vector<cv::Point2f> pts(d_pts.cols);
+ cv::Mat pts_mat(1, d_pts.cols, CV_32FC2, (void*) &pts[0]);
+ d_pts.download(pts_mat);
+
+ std::vector<cv::Point2f> pts_gold;
+ cv::goodFeaturesToTrack(image, pts_gold, maxCorners, qualityLevel, minDistance);
+
+ ASSERT_EQ(pts_gold.size(), pts.size());
+
+ size_t mistmatch = 0;
+ for (size_t i = 0; i < pts.size(); ++i)
+ {
+ cv::Point2i a = pts_gold[i];
+ cv::Point2i b = pts[i];
+
+ bool eq = std::abs(a.x - b.x) < 1 && std::abs(a.y - b.y) < 1;
+
+ if (!eq)
+ ++mistmatch;
+ }
+
+ double bad_ratio = static_cast<double>(mistmatch) / pts.size();
+
+ ASSERT_LE(bad_ratio, 0.01);
+}
+
+GPU_TEST_P(GoodFeaturesToTrack, EmptyCorners)
+{
+ int maxCorners = 1000;
+ double qualityLevel = 0.01;
+
+ cv::gpu::GoodFeaturesToTrackDetector_GPU detector(maxCorners, qualityLevel, minDistance);
+
+ cv::gpu::GpuMat src(100, 100, CV_8UC1, cv::Scalar::all(0));
+ cv::gpu::GpuMat corners(1, maxCorners, CV_32FC2);
+
+ detector(src, corners);
+
+ ASSERT_TRUE(corners.empty());
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_ImgProc, GoodFeaturesToTrack, testing::Combine(
+ ALL_DEVICES,
+ testing::Values(MinDistance(0.0), MinDistance(3.0))));
+
+#endif // HAVE_CUDA
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+#ifdef HAVE_CUDA
+
+using namespace cvtest;
+
+///////////////////////////////////////////////////////////////////////////////////////////////////////
+// HistEven
+
+struct HistEven : testing::TestWithParam<cv::gpu::DeviceInfo>
+{
+ cv::gpu::DeviceInfo devInfo;
+
+ virtual void SetUp()
+ {
+ devInfo = GetParam();
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(HistEven, Accuracy)
+{
+ cv::Mat img = readImage("stereobm/aloe-L.png");
+ ASSERT_FALSE(img.empty());
+
+ cv::Mat hsv;
+ cv::cvtColor(img, hsv, cv::COLOR_BGR2HSV);
+
+ int hbins = 30;
+ float hranges[] = {0.0f, 180.0f};
+
+ std::vector<cv::gpu::GpuMat> srcs;
+ cv::gpu::split(loadMat(hsv), srcs);
+
+ cv::gpu::GpuMat hist;
+ cv::gpu::histEven(srcs[0], hist, hbins, (int)hranges[0], (int)hranges[1]);
+
+ cv::MatND histnd;
+ int histSize[] = {hbins};
+ const float* ranges[] = {hranges};
+ int channels[] = {0};
+ cv::calcHist(&hsv, 1, channels, cv::Mat(), histnd, 1, histSize, ranges);
+
+ cv::Mat hist_gold = histnd;
+ hist_gold = hist_gold.t();
+ hist_gold.convertTo(hist_gold, CV_32S);
+
+ EXPECT_MAT_NEAR(hist_gold, hist, 0.0);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_ImgProc, HistEven, ALL_DEVICES);
+
+///////////////////////////////////////////////////////////////////////////////////////////////////////
+// CalcHist
+
+namespace
+{
+ void calcHistGold(const cv::Mat& src, cv::Mat& hist)
+ {
+ hist.create(1, 256, CV_32SC1);
+ hist.setTo(cv::Scalar::all(0));
+
+ int* hist_row = hist.ptr<int>();
+ for (int y = 0; y < src.rows; ++y)
+ {
+ const uchar* src_row = src.ptr(y);
+
+ for (int x = 0; x < src.cols; ++x)
+ ++hist_row[src_row[x]];
+ }
+ }
+}
+
+PARAM_TEST_CASE(CalcHist, cv::gpu::DeviceInfo, cv::Size)
+{
+ cv::gpu::DeviceInfo devInfo;
+
+ cv::Size size;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(CalcHist, Accuracy)
+{
+ cv::Mat src = randomMat(size, CV_8UC1);
+
+ cv::gpu::GpuMat hist;
+ cv::gpu::calcHist(loadMat(src), hist);
+
+ cv::Mat hist_gold;
+ calcHistGold(src, hist_gold);
+
+ EXPECT_MAT_NEAR(hist_gold, hist, 0.0);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_ImgProc, CalcHist, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES));
+
+///////////////////////////////////////////////////////////////////////////////////////////////////////
+// EqualizeHist
+
+PARAM_TEST_CASE(EqualizeHist, cv::gpu::DeviceInfo, cv::Size)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(EqualizeHist, Accuracy)
+{
+ cv::Mat src = randomMat(size, CV_8UC1);
+
+ cv::gpu::GpuMat dst;
+ cv::gpu::equalizeHist(loadMat(src), dst);
+
+ cv::Mat dst_gold;
+ cv::equalizeHist(src, dst_gold);
+
+ EXPECT_MAT_NEAR(dst_gold, dst, 3.0);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_ImgProc, EqualizeHist, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES));
+
+///////////////////////////////////////////////////////////////////////////////////////////////////////
+// CLAHE
+
+namespace
+{
+ IMPLEMENT_PARAM_CLASS(ClipLimit, double)
+}
+
+PARAM_TEST_CASE(CLAHE, cv::gpu::DeviceInfo, cv::Size, ClipLimit)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ double clipLimit;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ clipLimit = GET_PARAM(2);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(CLAHE, Accuracy)
+{
+ cv::Mat src = randomMat(size, CV_8UC1);
+
+ cv::Ptr<cv::gpu::CLAHE> clahe = cv::gpu::createCLAHE(clipLimit);
+ cv::gpu::GpuMat dst;
+ clahe->apply(loadMat(src), dst);
+
+ cv::Ptr<cv::CLAHE> clahe_gold = cv::createCLAHE(clipLimit);
+ cv::Mat dst_gold;
+ clahe_gold->apply(src, dst_gold);
+
+ ASSERT_MAT_NEAR(dst_gold, dst, 1.0);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_ImgProc, CLAHE, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ testing::Values(0.0, 40.0)));
+
+#endif // HAVE_CUDA
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+CV_GPU_TEST_MAIN("gpu")
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+#ifdef HAVE_CUDA
+
+using namespace cvtest;
+
+////////////////////////////////////////////////////////////////////////////////
+// MatchTemplate8U
+
+CV_ENUM(TemplateMethod, cv::TM_SQDIFF, cv::TM_SQDIFF_NORMED, cv::TM_CCORR, cv::TM_CCORR_NORMED, cv::TM_CCOEFF, cv::TM_CCOEFF_NORMED)
+#define ALL_TEMPLATE_METHODS testing::Values(TemplateMethod(cv::TM_SQDIFF), TemplateMethod(cv::TM_SQDIFF_NORMED), TemplateMethod(cv::TM_CCORR), TemplateMethod(cv::TM_CCORR_NORMED), TemplateMethod(cv::TM_CCOEFF), TemplateMethod(cv::TM_CCOEFF_NORMED))
+
+namespace
+{
+ IMPLEMENT_PARAM_CLASS(TemplateSize, cv::Size);
+}
+
+PARAM_TEST_CASE(MatchTemplate8U, cv::gpu::DeviceInfo, cv::Size, TemplateSize, Channels, TemplateMethod)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ cv::Size templ_size;
+ int cn;
+ int method;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ templ_size = GET_PARAM(2);
+ cn = GET_PARAM(3);
+ method = GET_PARAM(4);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(MatchTemplate8U, Accuracy)
+{
+ cv::Mat image = randomMat(size, CV_MAKETYPE(CV_8U, cn));
+ cv::Mat templ = randomMat(templ_size, CV_MAKETYPE(CV_8U, cn));
+
+ cv::gpu::GpuMat dst;
+ cv::gpu::matchTemplate(loadMat(image), loadMat(templ), dst, method);
+
+ cv::Mat dst_gold;
+ cv::matchTemplate(image, templ, dst_gold, method);
+
+ EXPECT_MAT_NEAR(dst_gold, dst, templ_size.area() * 1e-1);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_ImgProc, MatchTemplate8U, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ testing::Values(TemplateSize(cv::Size(5, 5)), TemplateSize(cv::Size(16, 16)), TemplateSize(cv::Size(30, 30))),
+ testing::Values(Channels(1), Channels(3), Channels(4)),
+ ALL_TEMPLATE_METHODS));
+
+////////////////////////////////////////////////////////////////////////////////
+// MatchTemplate32F
+
+PARAM_TEST_CASE(MatchTemplate32F, cv::gpu::DeviceInfo, cv::Size, TemplateSize, Channels, TemplateMethod)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ cv::Size templ_size;
+ int cn;
+ int method;
+
+ int n, m, h, w;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ templ_size = GET_PARAM(2);
+ cn = GET_PARAM(3);
+ method = GET_PARAM(4);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(MatchTemplate32F, Regression)
+{
+ cv::Mat image = randomMat(size, CV_MAKETYPE(CV_32F, cn));
+ cv::Mat templ = randomMat(templ_size, CV_MAKETYPE(CV_32F, cn));
+
+ cv::gpu::GpuMat dst;
+ cv::gpu::matchTemplate(loadMat(image), loadMat(templ), dst, method);
+
+ cv::Mat dst_gold;
+ cv::matchTemplate(image, templ, dst_gold, method);
+
+ EXPECT_MAT_NEAR(dst_gold, dst, templ_size.area() * 1e-1);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_ImgProc, MatchTemplate32F, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ testing::Values(TemplateSize(cv::Size(5, 5)), TemplateSize(cv::Size(16, 16)), TemplateSize(cv::Size(30, 30))),
+ testing::Values(Channels(1), Channels(3), Channels(4)),
+ testing::Values(TemplateMethod(cv::TM_SQDIFF), TemplateMethod(cv::TM_CCORR))));
+
+////////////////////////////////////////////////////////////////////////////////
+// MatchTemplateBlackSource
+
+PARAM_TEST_CASE(MatchTemplateBlackSource, cv::gpu::DeviceInfo, TemplateMethod)
+{
+ cv::gpu::DeviceInfo devInfo;
+ int method;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ method = GET_PARAM(1);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(MatchTemplateBlackSource, Accuracy)
+{
+ cv::Mat image = readImage("matchtemplate/black.png");
+ ASSERT_FALSE(image.empty());
+
+ cv::Mat pattern = readImage("matchtemplate/cat.png");
+ ASSERT_FALSE(pattern.empty());
+
+ cv::gpu::GpuMat d_dst;
+ cv::gpu::matchTemplate(loadMat(image), loadMat(pattern), d_dst, method);
+
+ cv::Mat dst(d_dst);
+
+ double maxValue;
+ cv::Point maxLoc;
+ cv::minMaxLoc(dst, NULL, &maxValue, NULL, &maxLoc);
+
+ cv::Point maxLocGold = cv::Point(284, 12);
+
+ ASSERT_EQ(maxLocGold, maxLoc);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_ImgProc, MatchTemplateBlackSource, testing::Combine(
+ ALL_DEVICES,
+ testing::Values(TemplateMethod(cv::TM_CCOEFF_NORMED), TemplateMethod(cv::TM_CCORR_NORMED))));
+
+////////////////////////////////////////////////////////////////////////////////
+// MatchTemplate_CCOEF_NORMED
+
+PARAM_TEST_CASE(MatchTemplate_CCOEF_NORMED, cv::gpu::DeviceInfo, std::pair<std::string, std::string>)
+{
+ cv::gpu::DeviceInfo devInfo;
+ std::string imageName;
+ std::string patternName;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ imageName = GET_PARAM(1).first;
+ patternName = GET_PARAM(1).second;
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(MatchTemplate_CCOEF_NORMED, Accuracy)
+{
+ cv::Mat image = readImage(imageName);
+ ASSERT_FALSE(image.empty());
+
+ cv::Mat pattern = readImage(patternName);
+ ASSERT_FALSE(pattern.empty());
+
+ cv::gpu::GpuMat d_dst;
+ cv::gpu::matchTemplate(loadMat(image), loadMat(pattern), d_dst, cv::TM_CCOEFF_NORMED);
+
+ cv::Mat dst(d_dst);
+
+ cv::Point minLoc, maxLoc;
+ double minVal, maxVal;
+ cv::minMaxLoc(dst, &minVal, &maxVal, &minLoc, &maxLoc);
+
+ cv::Mat dstGold;
+ cv::matchTemplate(image, pattern, dstGold, cv::TM_CCOEFF_NORMED);
+
+ double minValGold, maxValGold;
+ cv::Point minLocGold, maxLocGold;
+ cv::minMaxLoc(dstGold, &minValGold, &maxValGold, &minLocGold, &maxLocGold);
+
+ ASSERT_EQ(minLocGold, minLoc);
+ ASSERT_EQ(maxLocGold, maxLoc);
+ ASSERT_LE(maxVal, 1.0);
+ ASSERT_GE(minVal, -1.0);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_ImgProc, MatchTemplate_CCOEF_NORMED, testing::Combine(
+ ALL_DEVICES,
+ testing::Values(std::make_pair(std::string("matchtemplate/source-0.png"), std::string("matchtemplate/target-0.png")))));
+
+////////////////////////////////////////////////////////////////////////////////
+// MatchTemplate_CanFindBigTemplate
+
+struct MatchTemplate_CanFindBigTemplate : testing::TestWithParam<cv::gpu::DeviceInfo>
+{
+ cv::gpu::DeviceInfo devInfo;
+
+ virtual void SetUp()
+ {
+ devInfo = GetParam();
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(MatchTemplate_CanFindBigTemplate, SQDIFF_NORMED)
+{
+ cv::Mat scene = readImage("matchtemplate/scene.png");
+ ASSERT_FALSE(scene.empty());
+
+ cv::Mat templ = readImage("matchtemplate/template.png");
+ ASSERT_FALSE(templ.empty());
+
+ cv::gpu::GpuMat d_result;
+ cv::gpu::matchTemplate(loadMat(scene), loadMat(templ), d_result, cv::TM_SQDIFF_NORMED);
+
+ cv::Mat result(d_result);
+
+ double minVal;
+ cv::Point minLoc;
+ cv::minMaxLoc(result, &minVal, 0, &minLoc, 0);
+
+ ASSERT_GE(minVal, 0);
+ ASSERT_LT(minVal, 1e-3);
+ ASSERT_EQ(344, minLoc.x);
+ ASSERT_EQ(0, minLoc.y);
+}
+
+GPU_TEST_P(MatchTemplate_CanFindBigTemplate, SQDIFF)
+{
+ cv::Mat scene = readImage("matchtemplate/scene.png");
+ ASSERT_FALSE(scene.empty());
+
+ cv::Mat templ = readImage("matchtemplate/template.png");
+ ASSERT_FALSE(templ.empty());
+
+ cv::gpu::GpuMat d_result;
+ cv::gpu::matchTemplate(loadMat(scene), loadMat(templ), d_result, cv::TM_SQDIFF);
+
+ cv::Mat result(d_result);
+
+ double minVal;
+ cv::Point minLoc;
+ cv::minMaxLoc(result, &minVal, 0, &minLoc, 0);
+
+ ASSERT_GE(minVal, 0);
+ ASSERT_EQ(344, minLoc.x);
+ ASSERT_EQ(0, minLoc.y);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_ImgProc, MatchTemplate_CanFindBigTemplate, ALL_DEVICES);
+
+#endif // HAVE_CUDA
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+#ifdef HAVE_CUDA
+
+using namespace cvtest;
+
+////////////////////////////////////////////////////////////////////////////////
+// MeanShift
+
+struct MeanShift : testing::TestWithParam<cv::gpu::DeviceInfo>
+{
+ cv::gpu::DeviceInfo devInfo;
+
+ cv::Mat img;
+
+ int spatialRad;
+ int colorRad;
+
+ virtual void SetUp()
+ {
+ devInfo = GetParam();
+
+ cv::gpu::setDevice(devInfo.deviceID());
+
+ img = readImageType("meanshift/cones.png", CV_8UC4);
+ ASSERT_FALSE(img.empty());
+
+ spatialRad = 30;
+ colorRad = 30;
+ }
+};
+
+GPU_TEST_P(MeanShift, Filtering)
+{
+ cv::Mat img_template;
+ if (supportFeature(devInfo, cv::gpu::FEATURE_SET_COMPUTE_20))
+ img_template = readImage("meanshift/con_result.png");
+ else
+ img_template = readImage("meanshift/con_result_CC1X.png");
+ ASSERT_FALSE(img_template.empty());
+
+ cv::gpu::GpuMat d_dst;
+ cv::gpu::meanShiftFiltering(loadMat(img), d_dst, spatialRad, colorRad);
+
+ ASSERT_EQ(CV_8UC4, d_dst.type());
+
+ cv::Mat dst(d_dst);
+
+ cv::Mat result;
+ cv::cvtColor(dst, result, cv::COLOR_BGRA2BGR);
+
+ EXPECT_MAT_NEAR(img_template, result, 0.0);
+}
+
+GPU_TEST_P(MeanShift, Proc)
+{
+ cv::FileStorage fs;
+ if (supportFeature(devInfo, cv::gpu::FEATURE_SET_COMPUTE_20))
+ fs.open(std::string(cvtest::TS::ptr()->get_data_path()) + "meanshift/spmap.yaml", cv::FileStorage::READ);
+ else
+ fs.open(std::string(cvtest::TS::ptr()->get_data_path()) + "meanshift/spmap_CC1X.yaml", cv::FileStorage::READ);
+ ASSERT_TRUE(fs.isOpened());
+
+ cv::Mat spmap_template;
+ fs["spmap"] >> spmap_template;
+ ASSERT_FALSE(spmap_template.empty());
+
+ cv::gpu::GpuMat rmap_filtered;
+ cv::gpu::meanShiftFiltering(loadMat(img), rmap_filtered, spatialRad, colorRad);
+
+ cv::gpu::GpuMat rmap;
+ cv::gpu::GpuMat spmap;
+ cv::gpu::meanShiftProc(loadMat(img), rmap, spmap, spatialRad, colorRad);
+
+ ASSERT_EQ(CV_8UC4, rmap.type());
+
+ EXPECT_MAT_NEAR(rmap_filtered, rmap, 0.0);
+ EXPECT_MAT_NEAR(spmap_template, spmap, 0.0);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_ImgProc, MeanShift, ALL_DEVICES);
+
+////////////////////////////////////////////////////////////////////////////////
+// MeanShiftSegmentation
+
+namespace
+{
+ IMPLEMENT_PARAM_CLASS(MinSize, int);
+}
+
+PARAM_TEST_CASE(MeanShiftSegmentation, cv::gpu::DeviceInfo, MinSize)
+{
+ cv::gpu::DeviceInfo devInfo;
+ int minsize;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ minsize = GET_PARAM(1);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(MeanShiftSegmentation, Regression)
+{
+ cv::Mat img = readImageType("meanshift/cones.png", CV_8UC4);
+ ASSERT_FALSE(img.empty());
+
+ std::ostringstream path;
+ path << "meanshift/cones_segmented_sp10_sr10_minsize" << minsize;
+ if (supportFeature(devInfo, cv::gpu::FEATURE_SET_COMPUTE_20))
+ path << ".png";
+ else
+ path << "_CC1X.png";
+ cv::Mat dst_gold = readImage(path.str());
+ ASSERT_FALSE(dst_gold.empty());
+
+ cv::Mat dst;
+ cv::gpu::meanShiftSegmentation(loadMat(img), dst, 10, 10, minsize);
+
+ cv::Mat dst_rgb;
+ cv::cvtColor(dst, dst_rgb, cv::COLOR_BGRA2BGR);
+
+ EXPECT_MAT_SIMILAR(dst_gold, dst_rgb, 1e-3);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_ImgProc, MeanShiftSegmentation, testing::Combine(
+ ALL_DEVICES,
+ testing::Values(MinSize(0), MinSize(4), MinSize(20), MinSize(84), MinSize(340), MinSize(1364))));
+
+#endif // HAVE_CUDA
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifdef __GNUC__
+# pragma GCC diagnostic ignored "-Wmissing-declarations"
+# if defined __clang__ || defined __APPLE__
+# pragma GCC diagnostic ignored "-Wmissing-prototypes"
+# pragma GCC diagnostic ignored "-Wextra"
+# endif
+#endif
+
+#ifndef __OPENCV_TEST_PRECOMP_HPP__
+#define __OPENCV_TEST_PRECOMP_HPP__
+
+#include "opencv2/ts.hpp"
+#include "opencv2/ts/gpu_test.hpp"
+
+#include "opencv2/gpuimgproc.hpp"
+#include "opencv2/gpuarithm.hpp"
+#include "opencv2/imgproc.hpp"
+
+#endif
--- /dev/null
+if(NOT HAVE_CUDA)
+ ocv_module_disable(gpulegacy)
+endif()
+
+set(the_description "GPU-accelerated Computer Vision (legacy)")
+
+ocv_warnings_disable(CMAKE_CXX_FLAGS /wd4127 /wd4130 /wd4324 /wd4512 /wd4310 -Wundef -Wmissing-declarations -Wuninitialized)
+
+ocv_define_module(gpulegacy opencv_core OPTIONAL opencv_objdetect)
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_GPULEGACY_HPP__
+#define __OPENCV_GPULEGACY_HPP__
+
+#include "opencv2/gpulegacy/NCV.hpp"
+#include "opencv2/gpulegacy/NPP_staging.hpp"
+#include "opencv2/gpulegacy/NCVPyramid.hpp"
+#include "opencv2/gpulegacy/NCVHaarObjectDetection.hpp"
+#include "opencv2/gpulegacy/NCVBroxOpticalFlow.hpp"
+
+#endif /* __OPENCV_GPULEGACY_HPP__ */
#ifndef _ncv_hpp_
#define _ncv_hpp_
-#if (defined WIN32 || defined _WIN32 || defined WINCE) && defined CVAPI_EXPORTS
- #define NCV_EXPORTS __declspec(dllexport)
-#else
- #define NCV_EXPORTS
-#endif
+#include "opencv2/core/cvdef.h"
#ifdef _WIN32
#define WIN32_LEAN_AND_MEAN
//==============================================================================
-NCV_EXPORTS void ncvDebugOutput(const cv::String &msg);
+CV_EXPORTS void ncvDebugOutput(const cv::String &msg);
typedef void NCVDebugOutputHandler(const cv::String &msg);
-NCV_EXPORTS void ncvSetDebugOutputHandler(NCVDebugOutputHandler* func);
+CV_EXPORTS void ncvSetDebugOutputHandler(NCVDebugOutputHandler* func);
#define ncvAssertPrintCheck(pred, msg) \
typedef struct _NcvTimer *NcvTimer;
-NCV_EXPORTS NcvTimer ncvStartTimer(void);
+CV_EXPORTS NcvTimer ncvStartTimer(void);
-NCV_EXPORTS double ncvEndQueryTimerUs(NcvTimer t);
+CV_EXPORTS double ncvEndQueryTimerUs(NcvTimer t);
-NCV_EXPORTS double ncvEndQueryTimerMs(NcvTimer t);
+CV_EXPORTS double ncvEndQueryTimerMs(NcvTimer t);
//==============================================================================
/**
* Calculates the aligned top bound value
*/
-NCV_EXPORTS Ncv32u alignUp(Ncv32u what, Ncv32u alignment);
+CV_EXPORTS Ncv32u alignUp(Ncv32u what, Ncv32u alignment);
/**
/**
* NCVMemPtr
*/
-struct NCV_EXPORTS NCVMemPtr
+struct CV_EXPORTS NCVMemPtr
{
void *ptr;
NCVMemoryType memtype;
/**
* NCVMemSegment
*/
-struct NCV_EXPORTS NCVMemSegment
+struct CV_EXPORTS NCVMemSegment
{
NCVMemPtr begin;
size_t size;
/**
* INCVMemAllocator (Interface)
*/
-class NCV_EXPORTS INCVMemAllocator
+class CV_EXPORTS INCVMemAllocator
{
public:
virtual ~INCVMemAllocator() = 0;
/**
* NCVMemStackAllocator
*/
-class NCV_EXPORTS NCVMemStackAllocator : public INCVMemAllocator
+class CV_EXPORTS NCVMemStackAllocator : public INCVMemAllocator
{
NCVMemStackAllocator();
NCVMemStackAllocator(const NCVMemStackAllocator &);
/**
* NCVMemNativeAllocator
*/
-class NCV_EXPORTS NCVMemNativeAllocator : public INCVMemAllocator
+class CV_EXPORTS NCVMemNativeAllocator : public INCVMemAllocator
{
public:
/**
* Copy dispatchers
*/
-NCV_EXPORTS NCVStatus memSegCopyHelper(void *dst, NCVMemoryType dstType,
+CV_EXPORTS NCVStatus memSegCopyHelper(void *dst, NCVMemoryType dstType,
const void *src, NCVMemoryType srcType,
size_t sz, cudaStream_t cuStream);
-NCV_EXPORTS NCVStatus memSegCopyHelper2D(void *dst, Ncv32u dstPitch, NCVMemoryType dstType,
+CV_EXPORTS NCVStatus memSegCopyHelper2D(void *dst, Ncv32u dstPitch, NCVMemoryType dstType,
const void *src, Ncv32u srcPitch, NCVMemoryType srcType,
Ncv32u widthbytes, Ncv32u height, cudaStream_t cuStream);
/**
* Operations with rectangles
*/
-NCV_EXPORTS NCVStatus ncvGroupRectangles_host(NCVVector<NcvRect32u> &hypotheses, Ncv32u &numHypotheses,
+CV_EXPORTS NCVStatus ncvGroupRectangles_host(NCVVector<NcvRect32u> &hypotheses, Ncv32u &numHypotheses,
Ncv32u minNeighbors, Ncv32f intersectEps, NCVVector<Ncv32u> *hypothesesWeights);
-NCV_EXPORTS NCVStatus ncvDrawRects_8u_host(Ncv8u *h_dst, Ncv32u dstStride, Ncv32u dstWidth, Ncv32u dstHeight,
+CV_EXPORTS NCVStatus ncvDrawRects_8u_host(Ncv8u *h_dst, Ncv32u dstStride, Ncv32u dstWidth, Ncv32u dstHeight,
NcvRect32u *h_rects, Ncv32u numRects, Ncv8u color);
-NCV_EXPORTS NCVStatus ncvDrawRects_32u_host(Ncv32u *h_dst, Ncv32u dstStride, Ncv32u dstWidth, Ncv32u dstHeight,
+CV_EXPORTS NCVStatus ncvDrawRects_32u_host(Ncv32u *h_dst, Ncv32u dstStride, Ncv32u dstWidth, Ncv32u dstHeight,
NcvRect32u *h_rects, Ncv32u numRects, Ncv32u color);
-NCV_EXPORTS NCVStatus ncvDrawRects_8u_device(Ncv8u *d_dst, Ncv32u dstStride, Ncv32u dstWidth, Ncv32u dstHeight,
+CV_EXPORTS NCVStatus ncvDrawRects_8u_device(Ncv8u *d_dst, Ncv32u dstStride, Ncv32u dstWidth, Ncv32u dstHeight,
NcvRect32u *d_rects, Ncv32u numRects, Ncv8u color, cudaStream_t cuStream);
-NCV_EXPORTS NCVStatus ncvDrawRects_32u_device(Ncv32u *d_dst, Ncv32u dstStride, Ncv32u dstWidth, Ncv32u dstHeight,
+CV_EXPORTS NCVStatus ncvDrawRects_32u_device(Ncv32u *d_dst, Ncv32u dstStride, Ncv32u dstWidth, Ncv32u dstHeight,
NcvRect32u *d_rects, Ncv32u numRects, Ncv32u color, cudaStream_t cuStream);
#ifndef _ncv_optical_flow_h_
#define _ncv_optical_flow_h_
-#include "NCV.hpp"
+#include "opencv2/gpulegacy/NCV.hpp"
/// \brief Model and solver parameters
struct NCVBroxOpticalFlowDescriptor
/// \return computation status
/////////////////////////////////////////////////////////////////////////////////////////
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus NCVBroxOpticalFlow(const NCVBroxOpticalFlowDescriptor desc,
INCVMemAllocator &gpu_mem_allocator,
const NCVMatrix<Ncv32f> &frame0,
#ifndef _ncvhaarobjectdetection_hpp_
#define _ncvhaarobjectdetection_hpp_
-#include "NCV.hpp"
+#include "opencv2/gpulegacy/NCV.hpp"
//==============================================================================
};
-NCV_EXPORTS NCVStatus ncvDetectObjectsMultiScale_device(NCVMatrix<Ncv8u> &d_srcImg,
+CV_EXPORTS NCVStatus ncvDetectObjectsMultiScale_device(NCVMatrix<Ncv8u> &d_srcImg,
NcvSize32u srcRoi,
NCVVector<NcvRect32u> &d_dstRects,
Ncv32u &dstNumRects,
#define HAAR_STDDEV_BORDER 1
-NCV_EXPORTS NCVStatus ncvApplyHaarClassifierCascade_device(NCVMatrix<Ncv32u> &d_integralImage,
+CV_EXPORTS NCVStatus ncvApplyHaarClassifierCascade_device(NCVMatrix<Ncv32u> &d_integralImage,
NCVMatrix<Ncv32f> &d_weights,
NCVMatrixAlloc<Ncv32u> &d_pixelMask,
Ncv32u &numDetections,
cudaStream_t cuStream);
-NCV_EXPORTS NCVStatus ncvApplyHaarClassifierCascade_host(NCVMatrix<Ncv32u> &h_integralImage,
+CV_EXPORTS NCVStatus ncvApplyHaarClassifierCascade_host(NCVMatrix<Ncv32u> &h_integralImage,
NCVMatrix<Ncv32f> &h_weights,
NCVMatrixAlloc<Ncv32u> &h_pixelMask,
Ncv32u &numDetections,
#define RECT_SIMILARITY_PROPORTION 0.2f
-NCV_EXPORTS NCVStatus ncvGrowDetectionsVector_device(NCVVector<Ncv32u> &pixelMask,
+CV_EXPORTS NCVStatus ncvGrowDetectionsVector_device(NCVVector<Ncv32u> &pixelMask,
Ncv32u numPixelMaskDetections,
NCVVector<NcvRect32u> &hypotheses,
Ncv32u &totalDetections,
cudaStream_t cuStream);
-NCV_EXPORTS NCVStatus ncvGrowDetectionsVector_host(NCVVector<Ncv32u> &pixelMask,
+CV_EXPORTS NCVStatus ncvGrowDetectionsVector_host(NCVVector<Ncv32u> &pixelMask,
Ncv32u numPixelMaskDetections,
NCVVector<NcvRect32u> &hypotheses,
Ncv32u &totalDetections,
Ncv32f curScale);
-NCV_EXPORTS NCVStatus ncvHaarGetClassifierSize(const cv::String &filename, Ncv32u &numStages,
+CV_EXPORTS NCVStatus ncvHaarGetClassifierSize(const cv::String &filename, Ncv32u &numStages,
Ncv32u &numNodes, Ncv32u &numFeatures);
-NCV_EXPORTS NCVStatus ncvHaarLoadFromFile_host(const cv::String &filename,
+CV_EXPORTS NCVStatus ncvHaarLoadFromFile_host(const cv::String &filename,
HaarClassifierCascadeDescriptor &haar,
NCVVector<HaarStage64> &h_HaarStages,
NCVVector<HaarClassifierNode128> &h_HaarNodes,
NCVVector<HaarFeature64> &h_HaarFeatures);
-NCV_EXPORTS NCVStatus ncvHaarStoreNVBIN_host(const cv::String &filename,
+CV_EXPORTS NCVStatus ncvHaarStoreNVBIN_host(const cv::String &filename,
HaarClassifierCascadeDescriptor haar,
NCVVector<HaarStage64> &h_HaarStages,
NCVVector<HaarClassifierNode128> &h_HaarNodes,
#include <memory>
#include <vector>
-#include "NCV.hpp"
+#include "opencv2/gpulegacy/NCV.hpp"
+#include "opencv2/core/cuda/common.hpp"
+
+namespace cv { namespace gpu { namespace cudev
+{
+ namespace pyramid
+ {
+ CV_EXPORTS void downsampleX2(PtrStepSzb src, PtrStepSzb dst, int depth, int cn, cudaStream_t stream);
+ CV_EXPORTS void interpolateFrom1(PtrStepSzb src, PtrStepSzb dst, int depth, int cn, cudaStream_t stream);
+ }
+}}}
#if 0 //def _WIN32
template <class T>
-class NCV_EXPORTS NCVMatrixStack
+class CV_EXPORTS NCVMatrixStack
{
public:
NCVMatrixStack() {this->_arr.clear();}
template <class T>
-class NCV_EXPORTS NCVImagePyramid
+class CV_EXPORTS NCVImagePyramid
{
public:
#ifndef _npp_staging_hpp_
#define _npp_staging_hpp_
-#include "NCV.hpp"
+#include "opencv2/gpulegacy/NCV.hpp"
/**
* NOT THREAD SAFE
* \return Current CUDA stream
*/
-NCV_EXPORTS
+CV_EXPORTS
cudaStream_t nppStGetActiveCUDAstream();
* \param cudaStream [IN] cudaStream CUDA stream to become current
* \return CUDA stream used before
*/
-NCV_EXPORTS
+CV_EXPORTS
cudaStream_t nppStSetActiveCUDAstream(cudaStream_t cudaStream);
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStGetInterpolationBufferSize(NcvSize32u srcSize,
Ncv32u nStep,
Ncv32u *hpSize);
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStInterpolateFrames(const NppStInterpolationState *pState);
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStFilterRowBorder_32f_C1R(const Ncv32f *pSrc,
NcvSize32u srcSize,
Ncv32u nSrcStep,
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStFilterColumnBorder_32f_C1R(const Ncv32f *pSrc,
NcvSize32u srcSize,
Ncv32u nSrcStep,
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStVectorWarpGetBufferSize(NcvSize32u srcSize,
Ncv32u nSrcStep,
Ncv32u *hpSize);
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStVectorWarp_PSF1x1_32f_C1(const Ncv32f *pSrc,
NcvSize32u srcSize,
Ncv32u nSrcStep,
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStVectorWarp_PSF2x2_32f_C1(const Ncv32f *pSrc,
NcvSize32u srcSize,
Ncv32u nSrcStep,
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStResize_32f_C1R(const Ncv32f *pSrc,
NcvSize32u srcSize,
Ncv32u nSrcStep,
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStDecimate_32u_C1R(Ncv32u *d_src, Ncv32u srcStep,
Ncv32u *d_dst, Ncv32u dstStep,
NcvSize32u srcRoi, Ncv32u scale,
* Downsamples (decimates) an image using the nearest neighbor algorithm. 32-bit signed pixels, single channel.
* \see nppiStDecimate_32u_C1R
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStDecimate_32s_C1R(Ncv32s *d_src, Ncv32u srcStep,
Ncv32s *d_dst, Ncv32u dstStep,
NcvSize32u srcRoi, Ncv32u scale,
* Downsamples (decimates) an image using the nearest neighbor algorithm. 32-bit float pixels, single channel.
* \see nppiStDecimate_32u_C1R
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStDecimate_32f_C1R(Ncv32f *d_src, Ncv32u srcStep,
Ncv32f *d_dst, Ncv32u dstStep,
NcvSize32u srcRoi, Ncv32u scale,
* Downsamples (decimates) an image using the nearest neighbor algorithm. 64-bit unsigned pixels, single channel.
* \see nppiStDecimate_32u_C1R
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStDecimate_64u_C1R(Ncv64u *d_src, Ncv32u srcStep,
Ncv64u *d_dst, Ncv32u dstStep,
NcvSize32u srcRoi, Ncv32u scale,
* Downsamples (decimates) an image using the nearest neighbor algorithm. 64-bit signed pixels, single channel.
* \see nppiStDecimate_32u_C1R
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStDecimate_64s_C1R(Ncv64s *d_src, Ncv32u srcStep,
Ncv64s *d_dst, Ncv32u dstStep,
NcvSize32u srcRoi, Ncv32u scale,
* Downsamples (decimates) an image using the nearest neighbor algorithm. 64-bit float pixels, single channel.
* \see nppiStDecimate_32u_C1R
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStDecimate_64f_C1R(Ncv64f *d_src, Ncv32u srcStep,
Ncv64f *d_dst, Ncv32u dstStep,
NcvSize32u srcRoi, Ncv32u scale,
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStDecimate_32u_C1R_host(Ncv32u *h_src, Ncv32u srcStep,
Ncv32u *h_dst, Ncv32u dstStep,
NcvSize32u srcRoi, Ncv32u scale);
* Downsamples (decimates) an image using the nearest neighbor algorithm. 32-bit signed pixels, single channel. Host implementation.
* \see nppiStDecimate_32u_C1R_host
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStDecimate_32s_C1R_host(Ncv32s *h_src, Ncv32u srcStep,
Ncv32s *h_dst, Ncv32u dstStep,
NcvSize32u srcRoi, Ncv32u scale);
* Downsamples (decimates) an image using the nearest neighbor algorithm. 32-bit float pixels, single channel. Host implementation.
* \see nppiStDecimate_32u_C1R_host
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStDecimate_32f_C1R_host(Ncv32f *h_src, Ncv32u srcStep,
Ncv32f *h_dst, Ncv32u dstStep,
NcvSize32u srcRoi, Ncv32u scale);
* Downsamples (decimates) an image using the nearest neighbor algorithm. 64-bit unsigned pixels, single channel. Host implementation.
* \see nppiStDecimate_32u_C1R_host
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStDecimate_64u_C1R_host(Ncv64u *h_src, Ncv32u srcStep,
Ncv64u *h_dst, Ncv32u dstStep,
NcvSize32u srcRoi, Ncv32u scale);
* Downsamples (decimates) an image using the nearest neighbor algorithm. 64-bit signed pixels, single channel. Host implementation.
* \see nppiStDecimate_32u_C1R_host
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStDecimate_64s_C1R_host(Ncv64s *h_src, Ncv32u srcStep,
Ncv64s *h_dst, Ncv32u dstStep,
NcvSize32u srcRoi, Ncv32u scale);
* Downsamples (decimates) an image using the nearest neighbor algorithm. 64-bit float pixels, single channel. Host implementation.
* \see nppiStDecimate_32u_C1R_host
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStDecimate_64f_C1R_host(Ncv64f *h_src, Ncv32u srcStep,
Ncv64f *h_dst, Ncv32u dstStep,
NcvSize32u srcRoi, Ncv32u scale);
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStRectStdDev_32f_C1R(Ncv32u *d_sum, Ncv32u sumStep,
Ncv64u *d_sqsum, Ncv32u sqsumStep,
Ncv32f *d_norm, Ncv32u normStep,
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStRectStdDev_32f_C1R_host(Ncv32u *h_sum, Ncv32u sumStep,
Ncv64u *h_sqsum, Ncv32u sqsumStep,
Ncv32f *h_norm, Ncv32u normStep,
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStTranspose_32u_C1R(Ncv32u *d_src, Ncv32u srcStride,
Ncv32u *d_dst, Ncv32u dstStride, NcvSize32u srcRoi);
* Transposes an image. 32-bit signed pixels, single channel
* \see nppiStTranspose_32u_C1R
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStTranspose_32s_C1R(Ncv32s *d_src, Ncv32u srcStride,
Ncv32s *d_dst, Ncv32u dstStride, NcvSize32u srcRoi);
* Transposes an image. 32-bit float pixels, single channel
* \see nppiStTranspose_32u_C1R
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStTranspose_32f_C1R(Ncv32f *d_src, Ncv32u srcStride,
Ncv32f *d_dst, Ncv32u dstStride, NcvSize32u srcRoi);
* Transposes an image. 64-bit unsigned pixels, single channel
* \see nppiStTranspose_32u_C1R
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStTranspose_64u_C1R(Ncv64u *d_src, Ncv32u srcStride,
Ncv64u *d_dst, Ncv32u dstStride, NcvSize32u srcRoi);
* Transposes an image. 64-bit signed pixels, single channel
* \see nppiStTranspose_32u_C1R
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStTranspose_64s_C1R(Ncv64s *d_src, Ncv32u srcStride,
Ncv64s *d_dst, Ncv32u dstStride, NcvSize32u srcRoi);
* Transposes an image. 64-bit float pixels, single channel
* \see nppiStTranspose_32u_C1R
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStTranspose_64f_C1R(Ncv64f *d_src, Ncv32u srcStride,
Ncv64f *d_dst, Ncv32u dstStride, NcvSize32u srcRoi);
* Transposes an image. 128-bit pixels of any type, single channel
* \see nppiStTranspose_32u_C1R
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStTranspose_128_C1R(void *d_src, Ncv32u srcStep,
void *d_dst, Ncv32u dstStep, NcvSize32u srcRoi);
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStTranspose_32u_C1R_host(Ncv32u *h_src, Ncv32u srcStride,
Ncv32u *h_dst, Ncv32u dstStride, NcvSize32u srcRoi);
* Transposes an image. 32-bit signed pixels, single channel. Host implementation
* \see nppiStTranspose_32u_C1R_host
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStTranspose_32s_C1R_host(Ncv32s *h_src, Ncv32u srcStride,
Ncv32s *h_dst, Ncv32u dstStride, NcvSize32u srcRoi);
* Transposes an image. 32-bit float pixels, single channel. Host implementation
* \see nppiStTranspose_32u_C1R_host
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStTranspose_32f_C1R_host(Ncv32f *h_src, Ncv32u srcStride,
Ncv32f *h_dst, Ncv32u dstStride, NcvSize32u srcRoi);
* Transposes an image. 64-bit unsigned pixels, single channel. Host implementation
* \see nppiStTranspose_32u_C1R_host
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStTranspose_64u_C1R_host(Ncv64u *h_src, Ncv32u srcStride,
Ncv64u *h_dst, Ncv32u dstStride, NcvSize32u srcRoi);
* Transposes an image. 64-bit signed pixels, single channel. Host implementation
* \see nppiStTranspose_32u_C1R_host
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStTranspose_64s_C1R_host(Ncv64s *h_src, Ncv32u srcStride,
Ncv64s *h_dst, Ncv32u dstStride, NcvSize32u srcRoi);
* Transposes an image. 64-bit float pixels, single channel. Host implementation
* \see nppiStTranspose_32u_C1R_host
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStTranspose_64f_C1R_host(Ncv64f *h_src, Ncv32u srcStride,
Ncv64f *h_dst, Ncv32u dstStride, NcvSize32u srcRoi);
* Transposes an image. 128-bit pixels of any type, single channel. Host implementation
* \see nppiStTranspose_32u_C1R_host
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStTranspose_128_C1R_host(void *d_src, Ncv32u srcStep,
void *d_dst, Ncv32u dstStep, NcvSize32u srcRoi);
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStIntegralGetSize_8u32u(NcvSize32u roiSize, Ncv32u *pBufsize, cudaDeviceProp &devProp);
* Calculates the size of the temporary buffer for integral image creation
* \see nppiStIntegralGetSize_8u32u
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStIntegralGetSize_32f32f(NcvSize32u roiSize, Ncv32u *pBufsize, cudaDeviceProp &devProp);
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStIntegral_8u32u_C1R(Ncv8u *d_src, Ncv32u srcStep,
Ncv32u *d_dst, Ncv32u dstStep, NcvSize32u roiSize,
Ncv8u *pBuffer, Ncv32u bufSize, cudaDeviceProp &devProp);
* Creates an integral image representation for the input image
* \see nppiStIntegral_8u32u_C1R
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStIntegral_32f32f_C1R(Ncv32f *d_src, Ncv32u srcStep,
Ncv32f *d_dst, Ncv32u dstStep, NcvSize32u roiSize,
Ncv8u *pBuffer, Ncv32u bufSize, cudaDeviceProp &devProp);
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStIntegral_8u32u_C1R_host(Ncv8u *h_src, Ncv32u srcStep,
Ncv32u *h_dst, Ncv32u dstStep, NcvSize32u roiSize);
* Creates an integral image representation for the input image. Host implementation
* \see nppiStIntegral_8u32u_C1R_host
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStIntegral_32f32f_C1R_host(Ncv32f *h_src, Ncv32u srcStep,
Ncv32f *h_dst, Ncv32u dstStep, NcvSize32u roiSize);
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStSqrIntegralGetSize_8u64u(NcvSize32u roiSize, Ncv32u *pBufsize, cudaDeviceProp &devProp);
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStSqrIntegral_8u64u_C1R(Ncv8u *d_src, Ncv32u srcStep,
Ncv64u *d_dst, Ncv32u dstStep, NcvSize32u roiSize,
Ncv8u *pBuffer, Ncv32u bufSize, cudaDeviceProp &devProp);
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppiStSqrIntegral_8u64u_C1R_host(Ncv8u *h_src, Ncv32u srcStep,
Ncv64u *h_dst, Ncv32u dstStep, NcvSize32u roiSize);
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppsStCompactGetSize_32u(Ncv32u srcLen, Ncv32u *pBufsize, cudaDeviceProp &devProp);
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppsStCompact_32u(Ncv32u *d_src, Ncv32u srcLen,
Ncv32u *d_dst, Ncv32u *p_dstLen,
Ncv32u elemRemove, Ncv8u *pBuffer,
* Compacts the input vector by removing elements of specified value. 32-bit signed values
* \see nppsStCompact_32u
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppsStCompact_32s(Ncv32s *d_src, Ncv32u srcLen,
Ncv32s *d_dst, Ncv32u *p_dstLen,
Ncv32s elemRemove, Ncv8u *pBuffer,
* Compacts the input vector by removing elements of specified value. 32-bit float values
* \see nppsStCompact_32u
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppsStCompact_32f(Ncv32f *d_src, Ncv32u srcLen,
Ncv32f *d_dst, Ncv32u *p_dstLen,
Ncv32f elemRemove, Ncv8u *pBuffer,
*
* \return NCV status code
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppsStCompact_32u_host(Ncv32u *h_src, Ncv32u srcLen,
Ncv32u *h_dst, Ncv32u *dstLen, Ncv32u elemRemove);
* Compacts the input vector by removing elements of specified value. 32-bit signed values. Host implementation
* \see nppsStCompact_32u_host
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppsStCompact_32s_host(Ncv32s *h_src, Ncv32u srcLen,
Ncv32s *h_dst, Ncv32u *dstLen, Ncv32s elemRemove);
* Compacts the input vector by removing elements of specified value. 32-bit float values. Host implementation
* \see nppsStCompact_32u_host
*/
-NCV_EXPORTS
+CV_EXPORTS
NCVStatus nppsStCompact_32f_host(Ncv32f *h_src, Ncv32u srcLen,
Ncv32f *h_dst, Ncv32u *dstLen, Ncv32f elemRemove);
// copy or use the software.
//
//
-// License Agreement
+// License Agreement
// For Open Source Computer Vision Library
//
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
// Third party copyrights are property of their respective owners.
//
// Redistribution and use in source and binary forms, with or without modification,
//
//M*/
-#ifndef __OPENCV_internal_shared_HPP__
-#define __OPENCV_internal_shared_HPP__
+#ifndef __OPENCV_CORE_GPULEGACY_PRIVATE_HPP__
+#define __OPENCV_CORE_GPULEGACY_PRIVATE_HPP__
-#include <cuda_runtime.h>
-#include <npp.h>
-#include "NPP_staging.hpp"
-#include "opencv2/gpu/devmem2d.hpp"
-#include "safe_call.hpp"
-#include "opencv2/core/cuda/common.hpp"
+#ifndef __OPENCV_BUILD
+# error this is a private header which should not be used from outside of the OpenCV library
+#endif
+
+#include "opencv2/core/gpu_private.hpp"
+
+#ifndef HAVE_CUDA
+# error gpulegacy module requires CUDA
+#endif
+
+#include "opencv2/gpulegacy.hpp"
namespace cv { namespace gpu
{
private:
cudaStream_t oldStream;
};
+
+ CV_EXPORTS cv::String getNcvErrorMessage(int code);
+
+ static inline void checkNcvError(int err, const char* file, const int line, const char* func)
+ {
+ if (NCV_SUCCESS != err)
+ {
+ cv::String msg = getNcvErrorMessage(err);
+ cv::error(cv::Error::GpuApiCallError, msg, func, file, line);
+ }
+ }
}}
-#endif /* __OPENCV_internal_shared_HPP__ */
+#if defined(__GNUC__)
+ #define ncvSafeCall(expr) cv::gpu::checkNcvError(expr, __FILE__, __LINE__, __func__)
+#else /* defined(__CUDACC__) || defined(__MSVC__) */
+ #define ncvSafeCall(expr) cv::gpu::checkNcvError(expr, __FILE__, __LINE__, "")
+#endif
+
+#endif // __OPENCV_CORE_GPULEGACY_PRIVATE_HPP__
//
//M*/
-#include <iostream>
-#include <vector>
-#include "NCV.hpp"
-
+#include "precomp.hpp"
//==============================================================================
//
//
//==============================================================================
+namespace
+{
+ #define error_entry(entry) { entry, #entry }
+
+ struct ErrorEntry
+ {
+ int code;
+ const char* str;
+ };
+
+ struct ErrorEntryComparer
+ {
+ int code;
+ ErrorEntryComparer(int code_) : code(code_) {}
+ bool operator()(const ErrorEntry& e) const { return e.code == code; }
+ };
+
+ //////////////////////////////////////////////////////////////////////////
+ // NCV errors
+
+ const ErrorEntry ncv_errors [] =
+ {
+ error_entry( NCV_SUCCESS ),
+ error_entry( NCV_UNKNOWN_ERROR ),
+ error_entry( NCV_CUDA_ERROR ),
+ error_entry( NCV_NPP_ERROR ),
+ error_entry( NCV_FILE_ERROR ),
+ error_entry( NCV_NULL_PTR ),
+ error_entry( NCV_INCONSISTENT_INPUT ),
+ error_entry( NCV_TEXTURE_BIND_ERROR ),
+ error_entry( NCV_DIMENSIONS_INVALID ),
+ error_entry( NCV_INVALID_ROI ),
+ error_entry( NCV_INVALID_STEP ),
+ error_entry( NCV_INVALID_SCALE ),
+ error_entry( NCV_INVALID_SCALE ),
+ error_entry( NCV_ALLOCATOR_NOT_INITIALIZED ),
+ error_entry( NCV_ALLOCATOR_BAD_ALLOC ),
+ error_entry( NCV_ALLOCATOR_BAD_DEALLOC ),
+ error_entry( NCV_ALLOCATOR_INSUFFICIENT_CAPACITY ),
+ error_entry( NCV_ALLOCATOR_DEALLOC_ORDER ),
+ error_entry( NCV_ALLOCATOR_BAD_REUSE ),
+ error_entry( NCV_MEM_COPY_ERROR ),
+ error_entry( NCV_MEM_RESIDENCE_ERROR ),
+ error_entry( NCV_MEM_INSUFFICIENT_CAPACITY ),
+ error_entry( NCV_HAAR_INVALID_PIXEL_STEP ),
+ error_entry( NCV_HAAR_TOO_MANY_FEATURES_IN_CLASSIFIER ),
+ error_entry( NCV_HAAR_TOO_MANY_FEATURES_IN_CASCADE ),
+ error_entry( NCV_HAAR_TOO_LARGE_FEATURES ),
+ error_entry( NCV_HAAR_XML_LOADING_EXCEPTION ),
+ error_entry( NCV_NOIMPL_HAAR_TILTED_FEATURES ),
+ error_entry( NCV_WARNING_HAAR_DETECTIONS_VECTOR_OVERFLOW ),
+ error_entry( NPPST_SUCCESS ),
+ error_entry( NPPST_ERROR ),
+ error_entry( NPPST_CUDA_KERNEL_EXECUTION_ERROR ),
+ error_entry( NPPST_NULL_POINTER_ERROR ),
+ error_entry( NPPST_TEXTURE_BIND_ERROR ),
+ error_entry( NPPST_MEMCPY_ERROR ),
+ error_entry( NPPST_MEM_ALLOC_ERR ),
+ error_entry( NPPST_MEMFREE_ERR ),
+ error_entry( NPPST_INVALID_ROI ),
+ error_entry( NPPST_INVALID_STEP ),
+ error_entry( NPPST_INVALID_SCALE ),
+ error_entry( NPPST_MEM_INSUFFICIENT_BUFFER ),
+ error_entry( NPPST_MEM_RESIDENCE_ERROR ),
+ error_entry( NPPST_MEM_INTERNAL_ERROR )
+ };
+
+ const size_t ncv_error_num = sizeof(ncv_errors) / sizeof(ncv_errors[0]);
+}
+
+cv::String cv::gpu::getNcvErrorMessage(int code)
+{
+ size_t idx = std::find_if(ncv_errors, ncv_errors + ncv_error_num, ErrorEntryComparer(code)) - ncv_errors;
+
+ const char* msg = (idx != ncv_error_num) ? ncv_errors[idx].str : "Unknown error code";
+ String str = cv::format("%s [Code = %d]", msg, code);
+
+ return str;
+}
+
static void stdDebugOutput(const cv::String &msg)
{
debugOutputHandler = func;
}
-#if !defined CUDA_DISABLER
-
//==============================================================================
//
//===================================================================
-NCVMemStackAllocator::NCVMemStackAllocator(Ncv32u alignment_)
- :
- currentSize(0),
- _maxSize(0),
+NCVMemStackAllocator::NCVMemStackAllocator(Ncv32u alignment_) :
+ _memType(NCVMemoryTypeNone),
+ _alignment(alignment_),
allocBegin(NULL),
begin(NULL),
end(NULL),
- _memType(NCVMemoryTypeNone),
- _alignment(alignment_),
+ currentSize(0),
+ _maxSize(0),
bReusesMemory(false)
{
NcvBool bProperAlignment = (alignment_ & (alignment_ - 1)) == 0;
}
-NCVMemStackAllocator::NCVMemStackAllocator(NCVMemoryType memT, size_t capacity, Ncv32u alignment_, void *reusePtr)
- :
- currentSize(0),
- _maxSize(0),
- allocBegin(NULL),
+NCVMemStackAllocator::NCVMemStackAllocator(NCVMemoryType memT, size_t capacity, Ncv32u alignment_, void *reusePtr) :
_memType(memT),
- _alignment(alignment_)
+ _alignment(alignment_),
+ allocBegin(NULL),
+ currentSize(0),
+ _maxSize(0)
{
NcvBool bProperAlignment = (alignment_ & (alignment_ - 1)) == 0;
ncvAssertPrintCheck(bProperAlignment, "NCVMemStackAllocator ctor:: _alignment not power of 2");
NcvBool NCVMemStackAllocator::isInitialized(void) const
{
- return ((this->_alignment & (this->_alignment-1)) == 0) && isCounting() || this->allocBegin != NULL;
+ return (((this->_alignment & (this->_alignment-1)) == 0) && isCounting()) || this->allocBegin != NULL;
}
//===================================================================
-NCVMemNativeAllocator::NCVMemNativeAllocator(NCVMemoryType memT, Ncv32u alignment_)
- :
- currentSize(0),
- _maxSize(0),
+NCVMemNativeAllocator::NCVMemNativeAllocator(NCVMemoryType memT, Ncv32u alignment_) :
_memType(memT),
- _alignment(alignment_)
+ _alignment(alignment_),
+ currentSize(0),
+ _maxSize(0)
{
ncvAssertPrintReturn(memT != NCVMemoryTypeNone, "NCVMemNativeAllocator ctor:: counting not permitted for this allocator type", );
}
//
//===================================================================
+struct RectConvert
+{
+ cv::Rect operator()(const NcvRect32u& nr) const { return cv::Rect(nr.x, nr.y, nr.width, nr.height); }
+ NcvRect32u operator()(const cv::Rect& nr) const
+ {
+ NcvRect32u rect;
+ rect.x = nr.x;
+ rect.y = nr.y;
+ rect.width = nr.width;
+ rect.height = nr.height;
+ return rect;
+ }
+};
+
+static void groupRectangles(std::vector<NcvRect32u> &hypotheses, int groupThreshold, double eps, std::vector<Ncv32u> *weights)
+{
+#ifndef HAVE_OPENCV_OBJDETECT
+ (void) hypotheses;
+ (void) groupThreshold;
+ (void) eps;
+ (void) weights;
+ CV_Error(cv::Error::StsNotImplemented, "This functionality requires objdetect module");
+#else
+ std::vector<cv::Rect> rects(hypotheses.size());
+ std::transform(hypotheses.begin(), hypotheses.end(), rects.begin(), RectConvert());
+
+ if (weights)
+ {
+ std::vector<int> weights_int;
+ weights_int.assign(weights->begin(), weights->end());
+ cv::groupRectangles(rects, weights_int, groupThreshold, eps);
+ }
+ else
+ {
+ cv::groupRectangles(rects, groupThreshold, eps);
+ }
+ std::transform(rects.begin(), rects.end(), hypotheses.begin(), RectConvert());
+ hypotheses.resize(rects.size());
+#endif
+}
-//from OpenCV
-void groupRectangles(std::vector<NcvRect32u> &hypotheses, int groupThreshold, double eps, std::vector<Ncv32u> *weights);
NCVStatus ncvGroupRectangles_host(NCVVector<NcvRect32u> &hypotheses,
{
return drawRectsWrapperHost(h_dst, dstStride, dstWidth, dstHeight, h_rects, numRects, color);
}
-
-
-const Ncv32u NUMTHREADS_DRAWRECTS = 32;
-const Ncv32u NUMTHREADS_DRAWRECTS_LOG2 = 5;
-
-
-template <class T>
-__global__ void drawRects(T *d_dst,
- Ncv32u dstStride,
- Ncv32u dstWidth,
- Ncv32u dstHeight,
- NcvRect32u *d_rects,
- Ncv32u numRects,
- T color)
-{
- Ncv32u blockId = blockIdx.y * 65535 + blockIdx.x;
- if (blockId > numRects * 4)
- {
- return;
- }
-
- NcvRect32u curRect = d_rects[blockId >> 2];
- NcvBool bVertical = blockId & 0x1;
- NcvBool bTopLeft = blockId & 0x2;
-
- Ncv32u pt0x, pt0y;
- if (bVertical)
- {
- Ncv32u numChunks = (curRect.height + NUMTHREADS_DRAWRECTS - 1) >> NUMTHREADS_DRAWRECTS_LOG2;
-
- pt0x = bTopLeft ? curRect.x : curRect.x + curRect.width - 1;
- pt0y = curRect.y;
-
- if (pt0x < dstWidth)
- {
- for (Ncv32u chunkId = 0; chunkId < numChunks; chunkId++)
- {
- Ncv32u ptY = pt0y + chunkId * NUMTHREADS_DRAWRECTS + threadIdx.x;
- if (ptY < pt0y + curRect.height && ptY < dstHeight)
- {
- d_dst[ptY * dstStride + pt0x] = color;
- }
- }
- }
- }
- else
- {
- Ncv32u numChunks = (curRect.width + NUMTHREADS_DRAWRECTS - 1) >> NUMTHREADS_DRAWRECTS_LOG2;
-
- pt0x = curRect.x;
- pt0y = bTopLeft ? curRect.y : curRect.y + curRect.height - 1;
-
- if (pt0y < dstHeight)
- {
- for (Ncv32u chunkId = 0; chunkId < numChunks; chunkId++)
- {
- Ncv32u ptX = pt0x + chunkId * NUMTHREADS_DRAWRECTS + threadIdx.x;
- if (ptX < pt0x + curRect.width && ptX < dstWidth)
- {
- d_dst[pt0y * dstStride + ptX] = color;
- }
- }
- }
- }
-}
-
-
-template <class T>
-static NCVStatus drawRectsWrapperDevice(T *d_dst,
- Ncv32u dstStride,
- Ncv32u dstWidth,
- Ncv32u dstHeight,
- NcvRect32u *d_rects,
- Ncv32u numRects,
- T color,
- cudaStream_t cuStream)
-{
- (void)cuStream;
- ncvAssertReturn(d_dst != NULL && d_rects != NULL, NCV_NULL_PTR);
- ncvAssertReturn(dstWidth > 0 && dstHeight > 0, NCV_DIMENSIONS_INVALID);
- ncvAssertReturn(dstStride >= dstWidth, NCV_INVALID_STEP);
- ncvAssertReturn(numRects <= dstWidth * dstHeight, NCV_DIMENSIONS_INVALID);
-
- if (numRects == 0)
- {
- return NCV_SUCCESS;
- }
-
- dim3 grid(numRects * 4);
- dim3 block(NUMTHREADS_DRAWRECTS);
- if (grid.x > 65535)
- {
- grid.y = (grid.x + 65534) / 65535;
- grid.x = 65535;
- }
-
- drawRects<T><<<grid, block>>>(d_dst, dstStride, dstWidth, dstHeight, d_rects, numRects, color);
-
- ncvAssertCUDALastErrorReturn(NCV_CUDA_ERROR);
-
- return NCV_SUCCESS;
-}
-
-
-NCVStatus ncvDrawRects_8u_device(Ncv8u *d_dst,
- Ncv32u dstStride,
- Ncv32u dstWidth,
- Ncv32u dstHeight,
- NcvRect32u *d_rects,
- Ncv32u numRects,
- Ncv8u color,
- cudaStream_t cuStream)
-{
- return drawRectsWrapperDevice(d_dst, dstStride, dstWidth, dstHeight, d_rects, numRects, color, cuStream);
-}
-
-
-NCVStatus ncvDrawRects_32u_device(Ncv32u *d_dst,
- Ncv32u dstStride,
- Ncv32u dstWidth,
- Ncv32u dstHeight,
- NcvRect32u *d_rects,
- Ncv32u numRects,
- Ncv32u color,
- cudaStream_t cuStream)
-{
- return drawRectsWrapperDevice(d_dst, dstStride, dstWidth, dstHeight, d_rects, numRects, color, cuStream);
-}
-
-#endif /* CUDA_DISABLER */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include <iostream>
+#include <vector>
+
+#include "opencv2/gpulegacy/NCV.hpp"
+
+//===================================================================
+//
+// Operations with rectangles
+//
+//===================================================================
+
+
+const Ncv32u NUMTHREADS_DRAWRECTS = 32;
+const Ncv32u NUMTHREADS_DRAWRECTS_LOG2 = 5;
+
+
+template <class T>
+__global__ void drawRects(T *d_dst,
+ Ncv32u dstStride,
+ Ncv32u dstWidth,
+ Ncv32u dstHeight,
+ NcvRect32u *d_rects,
+ Ncv32u numRects,
+ T color)
+{
+ Ncv32u blockId = blockIdx.y * 65535 + blockIdx.x;
+ if (blockId > numRects * 4)
+ {
+ return;
+ }
+
+ NcvRect32u curRect = d_rects[blockId >> 2];
+ NcvBool bVertical = blockId & 0x1;
+ NcvBool bTopLeft = blockId & 0x2;
+
+ Ncv32u pt0x, pt0y;
+ if (bVertical)
+ {
+ Ncv32u numChunks = (curRect.height + NUMTHREADS_DRAWRECTS - 1) >> NUMTHREADS_DRAWRECTS_LOG2;
+
+ pt0x = bTopLeft ? curRect.x : curRect.x + curRect.width - 1;
+ pt0y = curRect.y;
+
+ if (pt0x < dstWidth)
+ {
+ for (Ncv32u chunkId = 0; chunkId < numChunks; chunkId++)
+ {
+ Ncv32u ptY = pt0y + chunkId * NUMTHREADS_DRAWRECTS + threadIdx.x;
+ if (ptY < pt0y + curRect.height && ptY < dstHeight)
+ {
+ d_dst[ptY * dstStride + pt0x] = color;
+ }
+ }
+ }
+ }
+ else
+ {
+ Ncv32u numChunks = (curRect.width + NUMTHREADS_DRAWRECTS - 1) >> NUMTHREADS_DRAWRECTS_LOG2;
+
+ pt0x = curRect.x;
+ pt0y = bTopLeft ? curRect.y : curRect.y + curRect.height - 1;
+
+ if (pt0y < dstHeight)
+ {
+ for (Ncv32u chunkId = 0; chunkId < numChunks; chunkId++)
+ {
+ Ncv32u ptX = pt0x + chunkId * NUMTHREADS_DRAWRECTS + threadIdx.x;
+ if (ptX < pt0x + curRect.width && ptX < dstWidth)
+ {
+ d_dst[pt0y * dstStride + ptX] = color;
+ }
+ }
+ }
+ }
+}
+
+
+template <class T>
+static NCVStatus drawRectsWrapperDevice(T *d_dst,
+ Ncv32u dstStride,
+ Ncv32u dstWidth,
+ Ncv32u dstHeight,
+ NcvRect32u *d_rects,
+ Ncv32u numRects,
+ T color,
+ cudaStream_t cuStream)
+{
+ (void)cuStream;
+ ncvAssertReturn(d_dst != NULL && d_rects != NULL, NCV_NULL_PTR);
+ ncvAssertReturn(dstWidth > 0 && dstHeight > 0, NCV_DIMENSIONS_INVALID);
+ ncvAssertReturn(dstStride >= dstWidth, NCV_INVALID_STEP);
+ ncvAssertReturn(numRects <= dstWidth * dstHeight, NCV_DIMENSIONS_INVALID);
+
+ if (numRects == 0)
+ {
+ return NCV_SUCCESS;
+ }
+
+ dim3 grid(numRects * 4);
+ dim3 block(NUMTHREADS_DRAWRECTS);
+ if (grid.x > 65535)
+ {
+ grid.y = (grid.x + 65534) / 65535;
+ grid.x = 65535;
+ }
+
+ drawRects<T><<<grid, block>>>(d_dst, dstStride, dstWidth, dstHeight, d_rects, numRects, color);
+
+ ncvAssertCUDALastErrorReturn(NCV_CUDA_ERROR);
+
+ return NCV_SUCCESS;
+}
+
+
+NCVStatus ncvDrawRects_8u_device(Ncv8u *d_dst,
+ Ncv32u dstStride,
+ Ncv32u dstWidth,
+ Ncv32u dstHeight,
+ NcvRect32u *d_rects,
+ Ncv32u numRects,
+ Ncv8u color,
+ cudaStream_t cuStream)
+{
+ return drawRectsWrapperDevice(d_dst, dstStride, dstWidth, dstHeight, d_rects, numRects, color, cuStream);
+}
+
+
+NCVStatus ncvDrawRects_32u_device(Ncv32u *d_dst,
+ Ncv32u dstStride,
+ Ncv32u dstWidth,
+ Ncv32u dstHeight,
+ NcvRect32u *d_rects,
+ Ncv32u numRects,
+ Ncv32u color,
+ cudaStream_t cuStream)
+{
+ return drawRectsWrapperDevice(d_dst, dstStride, dstWidth, dstHeight, d_rects, numRects, color, cuStream);
+}
#ifndef _ncv_alg_hpp_
#define _ncv_alg_hpp_
-#include "NCV.hpp"
+#include "opencv2/gpulegacy/NCV.hpp"
template <class T>
//
////////////////////////////////////////////////////////////////////////////////
-#if !defined CUDA_DISABLER
-
#include <iostream>
#include <vector>
#include <memory>
-#include "NPP_staging/NPP_staging.hpp"
-#include "NCVBroxOpticalFlow.hpp"
#include "opencv2/core/cuda/utility.hpp"
+#include "opencv2/gpulegacy/NPP_staging.hpp"
+#include "opencv2/gpulegacy/NCVBroxOpticalFlow.hpp"
+
typedef NCVVectorAlloc<Ncv32f> FloatVector;
return NCV_SUCCESS;
}
-
-#endif /* CUDA_DISABLER */
//
////////////////////////////////////////////////////////////////////////////////
-#if !defined CUDA_DISABLER
-
#include <algorithm>
#include <cstdio>
-#include "NCV.hpp"
-#include "NCVAlg.hpp"
-#include "NPP_staging/NPP_staging.hpp"
-#include "NCVRuntimeTemplates.hpp"
-#include "NCVHaarObjectDetection.hpp"
#include "opencv2/core/cuda/warp.hpp"
#include "opencv2/core/cuda/warp_shuffle.hpp"
+#include "opencv2/opencv_modules.hpp"
+
+#ifdef HAVE_OPENCV_OBJDETECT
+# include "opencv2/objdetect.hpp"
+# include "opencv2/objdetect/objdetect_c.h"
+#endif
+
+#include "opencv2/gpulegacy/NCV.hpp"
+#include "opencv2/gpulegacy/NPP_staging.hpp"
+#include "opencv2/gpulegacy/NCVHaarObjectDetection.hpp"
+
+#include "NCVRuntimeTemplates.hpp"
+#include "NCVAlg.hpp"
+
//==============================================================================
//
return ncvStat;
}
-
-NCVStatus loadFromXML(const cv::String &filename,
+static NCVStatus loadFromXML(const cv::String &filename,
HaarClassifierCascadeDescriptor &haar,
std::vector<HaarStage64> &haarStages,
std::vector<HaarClassifierNode128> &haarClassifierNodes,
- std::vector<HaarFeature64> &haarFeatures);
+ std::vector<HaarFeature64> &haarFeatures)
+{
+#ifndef HAVE_OPENCV_OBJDETECT
+ (void) filename;
+ (void) haar;
+ (void) haarStages;
+ (void) haarClassifierNodes;
+ (void) haarFeatures;
+ CV_Error(cv::Error::StsNotImplemented, "This functionality requires objdetect module");
+ return NCV_HAAR_XML_LOADING_EXCEPTION;
+#else
+ NCVStatus ncvStat;
+
+ haar.NumStages = 0;
+ haar.NumClassifierRootNodes = 0;
+ haar.NumClassifierTotalNodes = 0;
+ haar.NumFeatures = 0;
+ haar.ClassifierSize.width = 0;
+ haar.ClassifierSize.height = 0;
+ haar.bHasStumpsOnly = true;
+ haar.bNeedsTiltedII = false;
+ Ncv32u curMaxTreeDepth = 0;
+
+ std::vector<HaarClassifierNode128> h_TmpClassifierNotRootNodes;
+ haarStages.resize(0);
+ haarClassifierNodes.resize(0);
+ haarFeatures.resize(0);
+
+ cv::Ptr<CvHaarClassifierCascade> oldCascade = (CvHaarClassifierCascade*)cvLoad(filename.c_str(), 0, 0, 0);
+ if (oldCascade.empty())
+ {
+ return NCV_HAAR_XML_LOADING_EXCEPTION;
+ }
+
+ haar.ClassifierSize.width = oldCascade->orig_window_size.width;
+ haar.ClassifierSize.height = oldCascade->orig_window_size.height;
+
+ int stagesCound = oldCascade->count;
+ for(int s = 0; s < stagesCound; ++s) // by stages
+ {
+ HaarStage64 curStage;
+ curStage.setStartClassifierRootNodeOffset(static_cast<Ncv32u>(haarClassifierNodes.size()));
+
+ curStage.setStageThreshold(oldCascade->stage_classifier[s].threshold);
+
+ int treesCount = oldCascade->stage_classifier[s].count;
+ for(int t = 0; t < treesCount; ++t) // by trees
+ {
+ Ncv32u nodeId = 0;
+ CvHaarClassifier* tree = &oldCascade->stage_classifier[s].classifier[t];
+
+ int nodesCount = tree->count;
+ for(int n = 0; n < nodesCount; ++n) //by features
+ {
+ CvHaarFeature* feature = &tree->haar_feature[n];
+
+ HaarClassifierNode128 curNode;
+ curNode.setThreshold(tree->threshold[n]);
+
+ NcvBool bIsLeftNodeLeaf = false;
+ NcvBool bIsRightNodeLeaf = false;
+
+ HaarClassifierNodeDescriptor32 nodeLeft;
+ if ( tree->left[n] <= 0 )
+ {
+ Ncv32f leftVal = tree->alpha[-tree->left[n]];
+ ncvStat = nodeLeft.create(leftVal);
+ ncvAssertReturn(ncvStat == NCV_SUCCESS, ncvStat);
+ bIsLeftNodeLeaf = true;
+ }
+ else
+ {
+ Ncv32u leftNodeOffset = tree->left[n];
+ nodeLeft.create((Ncv32u)(h_TmpClassifierNotRootNodes.size() + leftNodeOffset - 1));
+ haar.bHasStumpsOnly = false;
+ }
+ curNode.setLeftNodeDesc(nodeLeft);
+
+ HaarClassifierNodeDescriptor32 nodeRight;
+ if ( tree->right[n] <= 0 )
+ {
+ Ncv32f rightVal = tree->alpha[-tree->right[n]];
+ ncvStat = nodeRight.create(rightVal);
+ ncvAssertReturn(ncvStat == NCV_SUCCESS, ncvStat);
+ bIsRightNodeLeaf = true;
+ }
+ else
+ {
+ Ncv32u rightNodeOffset = tree->right[n];
+ nodeRight.create((Ncv32u)(h_TmpClassifierNotRootNodes.size() + rightNodeOffset - 1));
+ haar.bHasStumpsOnly = false;
+ }
+ curNode.setRightNodeDesc(nodeRight);
+
+ Ncv32u tiltedVal = feature->tilted;
+ haar.bNeedsTiltedII = (tiltedVal != 0);
+
+ Ncv32u featureId = 0;
+ for(int l = 0; l < CV_HAAR_FEATURE_MAX; ++l) //by rects
+ {
+ Ncv32u rectX = feature->rect[l].r.x;
+ Ncv32u rectY = feature->rect[l].r.y;
+ Ncv32u rectWidth = feature->rect[l].r.width;
+ Ncv32u rectHeight = feature->rect[l].r.height;
+
+ Ncv32f rectWeight = feature->rect[l].weight;
+
+ if (rectWeight == 0/* && rectX == 0 &&rectY == 0 && rectWidth == 0 && rectHeight == 0*/)
+ break;
+
+ HaarFeature64 curFeature;
+ ncvStat = curFeature.setRect(rectX, rectY, rectWidth, rectHeight, haar.ClassifierSize.width, haar.ClassifierSize.height);
+ curFeature.setWeight(rectWeight);
+ ncvAssertReturn(NCV_SUCCESS == ncvStat, ncvStat);
+ haarFeatures.push_back(curFeature);
+
+ featureId++;
+ }
+
+ HaarFeatureDescriptor32 tmpFeatureDesc;
+ ncvStat = tmpFeatureDesc.create(haar.bNeedsTiltedII, bIsLeftNodeLeaf, bIsRightNodeLeaf,
+ featureId, static_cast<Ncv32u>(haarFeatures.size()) - featureId);
+ ncvAssertReturn(NCV_SUCCESS == ncvStat, ncvStat);
+ curNode.setFeatureDesc(tmpFeatureDesc);
+
+ if (!nodeId)
+ {
+ //root node
+ haarClassifierNodes.push_back(curNode);
+ curMaxTreeDepth = 1;
+ }
+ else
+ {
+ //other node
+ h_TmpClassifierNotRootNodes.push_back(curNode);
+ curMaxTreeDepth++;
+ }
+
+ nodeId++;
+ }
+ }
+
+ curStage.setNumClassifierRootNodes(treesCount);
+ haarStages.push_back(curStage);
+ }
+
+ //fill in cascade stats
+ haar.NumStages = static_cast<Ncv32u>(haarStages.size());
+ haar.NumClassifierRootNodes = static_cast<Ncv32u>(haarClassifierNodes.size());
+ haar.NumClassifierTotalNodes = static_cast<Ncv32u>(haar.NumClassifierRootNodes + h_TmpClassifierNotRootNodes.size());
+ haar.NumFeatures = static_cast<Ncv32u>(haarFeatures.size());
+
+ //merge root and leaf nodes in one classifiers array
+ Ncv32u offsetRoot = static_cast<Ncv32u>(haarClassifierNodes.size());
+ for (Ncv32u i=0; i<haarClassifierNodes.size(); i++)
+ {
+ HaarFeatureDescriptor32 featureDesc = haarClassifierNodes[i].getFeatureDesc();
+
+ HaarClassifierNodeDescriptor32 nodeLeft = haarClassifierNodes[i].getLeftNodeDesc();
+ if (!featureDesc.isLeftNodeLeaf())
+ {
+ Ncv32u newOffset = nodeLeft.getNextNodeOffset() + offsetRoot;
+ nodeLeft.create(newOffset);
+ }
+ haarClassifierNodes[i].setLeftNodeDesc(nodeLeft);
+
+ HaarClassifierNodeDescriptor32 nodeRight = haarClassifierNodes[i].getRightNodeDesc();
+ if (!featureDesc.isRightNodeLeaf())
+ {
+ Ncv32u newOffset = nodeRight.getNextNodeOffset() + offsetRoot;
+ nodeRight.create(newOffset);
+ }
+ haarClassifierNodes[i].setRightNodeDesc(nodeRight);
+ }
+
+ for (Ncv32u i=0; i<h_TmpClassifierNotRootNodes.size(); i++)
+ {
+ HaarFeatureDescriptor32 featureDesc = h_TmpClassifierNotRootNodes[i].getFeatureDesc();
+
+ HaarClassifierNodeDescriptor32 nodeLeft = h_TmpClassifierNotRootNodes[i].getLeftNodeDesc();
+ if (!featureDesc.isLeftNodeLeaf())
+ {
+ Ncv32u newOffset = nodeLeft.getNextNodeOffset() + offsetRoot;
+ nodeLeft.create(newOffset);
+ }
+ h_TmpClassifierNotRootNodes[i].setLeftNodeDesc(nodeLeft);
+
+ HaarClassifierNodeDescriptor32 nodeRight = h_TmpClassifierNotRootNodes[i].getRightNodeDesc();
+ if (!featureDesc.isRightNodeLeaf())
+ {
+ Ncv32u newOffset = nodeRight.getNextNodeOffset() + offsetRoot;
+ nodeRight.create(newOffset);
+ }
+ h_TmpClassifierNotRootNodes[i].setRightNodeDesc(nodeRight);
+
+ haarClassifierNodes.push_back(h_TmpClassifierNotRootNodes[i]);
+ }
+
+ return NCV_SUCCESS;
+#endif
+}
#define NVBIN_HAAR_SIZERESERVED 16
fclose(fp);
return NCV_SUCCESS;
}
-
-#endif /* CUDA_DISABLER */
#include <limits.h>
#include <float.h>
-#include "NCV.hpp"
+#include "opencv2/gpulegacy/NCV.hpp"
template<typename TBase> inline __host__ __device__ TBase _pixMaxVal();
template<> static inline __host__ __device__ Ncv8u _pixMaxVal<Ncv8u>() {return UCHAR_MAX;}
//
//M*/
-#if !defined CUDA_DISABLER
-
-#include <cuda_runtime.h>
#include <stdio.h>
-#include "NCV.hpp"
+#include <cuda_runtime.h>
+
+#include "opencv2/core/cuda/common.hpp"
+
+#include "opencv2/gpulegacy/NCV.hpp"
+#include "opencv2/gpulegacy/NCVPyramid.hpp"
+
#include "NCVAlg.hpp"
-#include "NCVPyramid.hpp"
#include "NCVPixelOperations.hpp"
-#include "opencv2/core/cuda/common.hpp"
template<typename T, Ncv32u CN> struct __average4_CN {static __host__ __device__ T _average4_CN(const T &p00, const T &p01, const T &p10, const T &p11);};
cudaSafeCall( cudaDeviceSynchronize() );
}
- template void kernelDownsampleX2_gpu<uchar1>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void kernelDownsampleX2_gpu<uchar3>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void kernelDownsampleX2_gpu<uchar4>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
-
- template void kernelDownsampleX2_gpu<ushort1>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void kernelDownsampleX2_gpu<ushort3>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void kernelDownsampleX2_gpu<ushort4>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ void downsampleX2(PtrStepSzb src, PtrStepSzb dst, int depth, int cn, cudaStream_t stream)
+ {
+ typedef void (*func_t)(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void kernelDownsampleX2_gpu<float1>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void kernelDownsampleX2_gpu<float3>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void kernelDownsampleX2_gpu<float4>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ static const func_t funcs[6][4] =
+ {
+ {kernelDownsampleX2_gpu<uchar1> , 0 /*kernelDownsampleX2_gpu<uchar2>*/ , kernelDownsampleX2_gpu<uchar3> , kernelDownsampleX2_gpu<uchar4> },
+ {0 /*kernelDownsampleX2_gpu<char1>*/ , 0 /*kernelDownsampleX2_gpu<char2>*/ , 0 /*kernelDownsampleX2_gpu<char3>*/ , 0 /*kernelDownsampleX2_gpu<char4>*/ },
+ {kernelDownsampleX2_gpu<ushort1> , 0 /*kernelDownsampleX2_gpu<ushort2>*/, kernelDownsampleX2_gpu<ushort3> , kernelDownsampleX2_gpu<ushort4> },
+ {0 /*kernelDownsampleX2_gpu<short1>*/ , 0 /*kernelDownsampleX2_gpu<short2>*/ , 0 /*kernelDownsampleX2_gpu<short3>*/, 0 /*kernelDownsampleX2_gpu<short4>*/},
+ {0 /*kernelDownsampleX2_gpu<int1>*/ , 0 /*kernelDownsampleX2_gpu<int2>*/ , 0 /*kernelDownsampleX2_gpu<int3>*/ , 0 /*kernelDownsampleX2_gpu<int4>*/ },
+ {kernelDownsampleX2_gpu<float1> , 0 /*kernelDownsampleX2_gpu<float2>*/ , kernelDownsampleX2_gpu<float3> , kernelDownsampleX2_gpu<float4> }
+ };
+
+ const func_t func = funcs[depth][cn - 1];
+ CV_Assert(func != 0);
+
+ func(src, dst, stream);
+ }
}
}}}
cudaSafeCall( cudaDeviceSynchronize() );
}
- template void kernelInterpolateFrom1_gpu<uchar1>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void kernelInterpolateFrom1_gpu<uchar3>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void kernelInterpolateFrom1_gpu<uchar4>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
-
- template void kernelInterpolateFrom1_gpu<ushort1>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void kernelInterpolateFrom1_gpu<ushort3>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void kernelInterpolateFrom1_gpu<ushort4>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ void interpolateFrom1(PtrStepSzb src, PtrStepSzb dst, int depth, int cn, cudaStream_t stream)
+ {
+ typedef void (*func_t)(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void kernelInterpolateFrom1_gpu<float1>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void kernelInterpolateFrom1_gpu<float3>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template void kernelInterpolateFrom1_gpu<float4>(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
+ static const func_t funcs[6][4] =
+ {
+ {kernelInterpolateFrom1_gpu<uchar1> , 0 /*kernelInterpolateFrom1_gpu<uchar2>*/ , kernelInterpolateFrom1_gpu<uchar3> , kernelInterpolateFrom1_gpu<uchar4> },
+ {0 /*kernelInterpolateFrom1_gpu<char1>*/ , 0 /*kernelInterpolateFrom1_gpu<char2>*/ , 0 /*kernelInterpolateFrom1_gpu<char3>*/ , 0 /*kernelInterpolateFrom1_gpu<char4>*/ },
+ {kernelInterpolateFrom1_gpu<ushort1> , 0 /*kernelInterpolateFrom1_gpu<ushort2>*/, kernelInterpolateFrom1_gpu<ushort3> , kernelInterpolateFrom1_gpu<ushort4> },
+ {0 /*kernelInterpolateFrom1_gpu<short1>*/, 0 /*kernelInterpolateFrom1_gpu<short2>*/ , 0 /*kernelInterpolateFrom1_gpu<short3>*/, 0 /*kernelInterpolateFrom1_gpu<short4>*/},
+ {0 /*kernelInterpolateFrom1_gpu<int1>*/ , 0 /*kernelInterpolateFrom1_gpu<int2>*/ , 0 /*kernelInterpolateFrom1_gpu<int3>*/ , 0 /*kernelInterpolateFrom1_gpu<int4>*/ },
+ {kernelInterpolateFrom1_gpu<float1> , 0 /*kernelInterpolateFrom1_gpu<float2>*/ , kernelInterpolateFrom1_gpu<float3> , kernelInterpolateFrom1_gpu<float4> }
+ };
+
+ const func_t func = funcs[depth][cn - 1];
+ CV_Assert(func != 0);
+
+ func(src, dst, stream);
+ }
}
}}}
template class NCVImagePyramid<float4>;
#endif //_WIN32
-
-#endif /* CUDA_DISABLER */
#include <vector>
#include <cuda_runtime.h>
-#include "NPP_staging.hpp"
+
#include "opencv2/core/cuda/warp.hpp"
#include "opencv2/core/cuda/warp_shuffle.hpp"
+#include "opencv2/gpulegacy/NPP_staging.hpp"
+
texture<Ncv8u, 1, cudaReadModeElementType> tex8u;
texture<Ncv32u, 1, cudaReadModeElementType> tex32u;
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_PRECOMP_H__
+#define __OPENCV_PRECOMP_H__
+
+#include <limits>
+#include <iostream>
+#include <algorithm>
+
+#include "opencv2/gpulegacy.hpp"
+#include "opencv2/core/utility.hpp"
+
+#include "opencv2/opencv_modules.hpp"
+
+#ifdef HAVE_OPENCV_OBJDETECT
+# include "opencv2/objdetect.hpp"
+#endif
+
+#include "opencv2/core/gpu_private.hpp"
+#include "opencv2/gpulegacy/private.hpp"
+
+#endif /* __OPENCV_PRECOMP_H__ */
#include <vector>
#include "NCVTest.hpp"
-#include <main_test_nvidia.h>
-//enum OutputLevel
-//{
-// OutputLevelNone,
-// OutputLevelCompact,
-// OutputLevelFull
-//};
+#include "main_test_nvidia.h"
class NCVAutoTestLister
{
#include <fstream>
#include <cuda_runtime.h>
-#include "NPP_staging.hpp"
+
+#include "opencv2/gpulegacy.hpp"
struct NCVTestReport
#include <memory>
-#include "NCV.hpp"
-#include <opencv2/highgui.hpp>
+#include "opencv2/highgui.hpp"
+#include "opencv2/gpulegacy.hpp"
template <class T>
//
//M*/
-#if !defined CUDA_DISABLER
-
-#include "TestCompact.h"
-
+#include "test_precomp.hpp"
TestCompact::TestCompact(std::string testName_, NCVTestSourceProvider<Ncv32u> &src_,
Ncv32u length_, Ncv32u badElem_, Ncv32u badElemPercentage_)
{
return true;
}
-
-#endif /* CUDA_DISABLER */
//
//M*/
-#if !defined CUDA_DISABLER
-
-#include "TestDrawRects.h"
-#include "NCVHaarObjectDetection.hpp"
+#include "test_precomp.hpp"
template <class T>
template class TestDrawRects<Ncv8u>;
template class TestDrawRects<Ncv32u>;
-
-#endif /* CUDA_DISABLER */
//
//M*/
-#if !defined CUDA_DISABLER
-
-#include <float.h>
-
-#if defined(__GNUC__) && !defined(__APPLE__) && !defined(__arm__)
- #include <fpu_control.h>
-#endif
+#include "test_precomp.hpp"
namespace
{
}
}
-#include "TestHaarCascadeApplication.h"
-#include "NCVHaarObjectDetection.hpp"
-
-
TestHaarCascadeApplication::TestHaarCascadeApplication(std::string testName_, NCVTestSourceProvider<Ncv8u> &src_,
std::string cascadeName_, Ncv32u width_, Ncv32u height_)
:
{
return true;
}
-
-#endif /* CUDA_DISABLER */
//
//M*/
-#if !defined CUDA_DISABLER
-
-#include "TestHaarCascadeLoader.h"
-#include "NCVHaarObjectDetection.hpp"
+#include "test_precomp.hpp"
TestHaarCascadeLoader::TestHaarCascadeLoader(std::string testName_, std::string cascadeName_)
{
return true;
}
-
-#endif /* CUDA_DISABLER */
//
//M*/
-#if !defined CUDA_DISABLER
-
-#include "TestHypothesesFilter.h"
-#include "NCVHaarObjectDetection.hpp"
+#include "test_precomp.hpp"
TestHypothesesFilter::TestHypothesesFilter(std::string testName_, NCVTestSourceProvider<Ncv32u> &src_,
{
return true;
}
-
-#endif /* CUDA_DISABLER */
//
//M*/
-#if !defined CUDA_DISABLER
-
-#include "TestHypothesesGrow.h"
-#include "NCVHaarObjectDetection.hpp"
+#include "test_precomp.hpp"
TestHypothesesGrow::TestHypothesesGrow(std::string testName_, NCVTestSourceProvider<Ncv32u> &src_,
{
return true;
}
-
-#endif /* CUDA_DISABLER */
//
//M*/
-#if !defined CUDA_DISABLER
-
-#include <math.h>
-#include "TestIntegralImage.h"
+#include "test_precomp.hpp"
template <class T_in, class T_out>
template class TestIntegralImage<Ncv8u, Ncv32u>;
template class TestIntegralImage<Ncv32f, Ncv32f>;
-
-#endif /* CUDA_DISABLER */
//
//M*/
-#if !defined CUDA_DISABLER
-
-#include "TestIntegralImageSquared.h"
+#include "test_precomp.hpp"
TestIntegralImageSquared::TestIntegralImageSquared(std::string testName_, NCVTestSourceProvider<Ncv8u> &src_,
{
return true;
}
-
-#endif /* CUDA_DISABLER */
//
//M*/
-#if !defined CUDA_DISABLER
-
-#include <math.h>
-
-#include "TestRectStdDev.h"
+#include "test_precomp.hpp"
TestRectStdDev::TestRectStdDev(std::string testName_, NCVTestSourceProvider<Ncv8u> &src_,
{
return true;
}
-
-#endif /* CUDA_DISABLER */
//
//M*/
-#if !defined CUDA_DISABLER
-
-#include <math.h>
-
-#include "TestResize.h"
+#include "test_precomp.hpp"
template <class T>
template class TestResize<Ncv32u>;
template class TestResize<Ncv64u>;
-
-#endif /* CUDA_DISABLER */
//
//M*/
-#if !defined CUDA_DISABLER
-
-#include <math.h>
-
-#include "TestTranspose.h"
+#include "test_precomp.hpp"
template <class T>
template class TestTranspose<Ncv32u>;
template class TestTranspose<Ncv64u>;
-
-#endif /* CUDA_DISABLER */
//
//M*/
+#include "test_precomp.hpp"
+
#if defined _MSC_VER && _MSC_VER >= 1200
# pragma warning (disable : 4408 4201 4100)
#endif
-#if !defined CUDA_DISABLER
-
-#include <cstdio>
-
-#include "NCV.hpp"
-#include "NCVHaarObjectDetection.hpp"
-
-#include "TestIntegralImage.h"
-#include "TestIntegralImageSquared.h"
-#include "TestRectStdDev.h"
-#include "TestResize.h"
-#include "TestCompact.h"
-#include "TestTranspose.h"
-
-#include "TestDrawRects.h"
-#include "TestHypothesesGrow.h"
-#include "TestHypothesesFilter.h"
-#include "TestHaarCascadeLoader.h"
-#include "TestHaarCascadeApplication.h"
-
-#include "NCVAutoTestLister.hpp"
-#include "NCVTestSourceProvider.hpp"
-
-#include "main_test_nvidia.h"
-
static std::string path;
namespace {
NCVAutoTestLister testListerResize("NPPST Resize", outputLevel);
NCVTestSourceProvider<Ncv32u> testSrcRandom_32u(2010, 0, 0xFFFFFFFF, 2048, 2048);
- NCVTestSourceProvider<Ncv64u> testSrcRandom_64u(2010, 0, -1, 2048, 2048);
+ NCVTestSourceProvider<Ncv64u> testSrcRandom_64u(2010, 0, (Ncv64u) -1, 2048, 2048);
generateResizeTests(testListerResize, testSrcRandom_32u);
generateResizeTests(testListerResize, testSrcRandom_64u);
NCVAutoTestLister testListerTranspose("NPPST Transpose", outputLevel);
NCVTestSourceProvider<Ncv32u> testSrcRandom_32u(2010, 0, 0xFFFFFFFF, 2048, 2048);
- NCVTestSourceProvider<Ncv64u> testSrcRandom_64u(2010, 0, -1, 2048, 2048);
+ NCVTestSourceProvider<Ncv64u> testSrcRandom_64u(2010, 0, (Ncv64u) -1, 2048, 2048);
generateTransposeTests(testListerTranspose, testSrcRandom_32u);
generateTransposeTests(testListerTranspose, testSrcRandom_64u);
return testListerVisualize.invoke();
}
-
-#endif /* CUDA_DISABLER */
GPU_TEST_P(NCV, Visualization)
{
- // this functionality doesn't used in gpu module
bool res = nvidia_NCV_Visualization(_path, nvidiaTestOutputLevel);
ASSERT_TRUE(res);
}
-INSTANTIATE_TEST_CASE_P(GPU_NVidia, NPPST, ALL_DEVICES);
-INSTANTIATE_TEST_CASE_P(GPU_NVidia, NCV, ALL_DEVICES);
+INSTANTIATE_TEST_CASE_P(GPU_Legacy, NPPST, ALL_DEVICES);
+INSTANTIATE_TEST_CASE_P(GPU_Legacy, NCV, ALL_DEVICES);
#endif // HAVE_CUDA
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifdef __GNUC__
+# pragma GCC diagnostic ignored "-Wmissing-declarations"
+# if defined __clang__ || defined __APPLE__
+# pragma GCC diagnostic ignored "-Wmissing-prototypes"
+# pragma GCC diagnostic ignored "-Wextra"
+# endif
+#endif
+
+#ifndef __OPENCV_TEST_PRECOMP_HPP__
+#define __OPENCV_TEST_PRECOMP_HPP__
+
+#if defined(__GNUC__) && !defined(__APPLE__) && !defined(__arm__)
+ #include <fpu_control.h>
+#endif
+
+#include <cfloat>
+#include <cstdio>
+#include <cmath>
+#include <vector>
+#include <string>
+#include <map>
+#include <memory>
+#include <algorithm>
+#include <fstream>
+
+#include "opencv2/ts.hpp"
+#include "opencv2/ts/gpu_test.hpp"
+
+#include "opencv2/core/gpumat.hpp"
+#include "opencv2/gpulegacy.hpp"
+#include "opencv2/highgui.hpp"
+
+#include "opencv2/core/gpu_private.hpp"
+
+#include "NCVTest.hpp"
+#include "NCVAutoTestLister.hpp"
+#include "NCVTestSourceProvider.hpp"
+
+#include "TestIntegralImage.h"
+#include "TestIntegralImageSquared.h"
+#include "TestRectStdDev.h"
+#include "TestResize.h"
+#include "TestCompact.h"
+#include "TestTranspose.h"
+#include "TestDrawRects.h"
+#include "TestHypothesesGrow.h"
+#include "TestHypothesesFilter.h"
+#include "TestHaarCascadeLoader.h"
+#include "TestHaarCascadeApplication.h"
+
+#include "main_test_nvidia.h"
+
+#endif
--- /dev/null
+if(ANDROID OR IOS)
+ ocv_module_disable(gpuoptflow)
+endif()
+
+set(the_description "GPU-accelerated Optical Flow")
+
+ocv_warnings_disable(CMAKE_CXX_FLAGS /wd4127 /wd4324 /wd4512 -Wundef -Wmissing-declarations)
+
+ocv_define_module(gpuoptflow opencv_video opencv_legacy opencv_gpuarithm opencv_gpuwarping opencv_gpuimgproc OPTIONAL opencv_gpulegacy)
--- /dev/null
+****************************************
+gpuoptflow. GPU-accelerated Optical Flow
+****************************************
+
+.. toctree::
+ :maxdepth: 1
+
+ optflow
--- /dev/null
+Optical Flow
+============
+
+.. highlight:: cpp
+
+
+
+gpu::BroxOpticalFlow
+--------------------
+.. ocv:class:: gpu::BroxOpticalFlow
+
+Class computing the optical flow for two images using Brox et al Optical Flow algorithm ([Brox2004]_). ::
+
+ class BroxOpticalFlow
+ {
+ public:
+ BroxOpticalFlow(float alpha_, float gamma_, float scale_factor_, int inner_iterations_, int outer_iterations_, int solver_iterations_);
+
+ //! Compute optical flow
+ //! frame0 - source frame (supports only CV_32FC1 type)
+ //! frame1 - frame to track (with the same size and type as frame0)
+ //! u - flow horizontal component (along x axis)
+ //! v - flow vertical component (along y axis)
+ void operator ()(const GpuMat& frame0, const GpuMat& frame1, GpuMat& u, GpuMat& v, Stream& stream = Stream::Null());
+
+ //! flow smoothness
+ float alpha;
+
+ //! gradient constancy importance
+ float gamma;
+
+ //! pyramid scale factor
+ float scale_factor;
+
+ //! number of lagged non-linearity iterations (inner loop)
+ int inner_iterations;
+
+ //! number of warping iterations (number of pyramid levels)
+ int outer_iterations;
+
+ //! number of linear system solver iterations
+ int solver_iterations;
+
+ GpuMat buf;
+ };
+
+
+
+gpu::FarnebackOpticalFlow
+-------------------------
+.. ocv:class:: gpu::FarnebackOpticalFlow
+
+Class computing a dense optical flow using the Gunnar Farneback’s algorithm. ::
+
+ class CV_EXPORTS FarnebackOpticalFlow
+ {
+ public:
+ FarnebackOpticalFlow()
+ {
+ numLevels = 5;
+ pyrScale = 0.5;
+ fastPyramids = false;
+ winSize = 13;
+ numIters = 10;
+ polyN = 5;
+ polySigma = 1.1;
+ flags = 0;
+ }
+
+ int numLevels;
+ double pyrScale;
+ bool fastPyramids;
+ int winSize;
+ int numIters;
+ int polyN;
+ double polySigma;
+ int flags;
+
+ void operator ()(const GpuMat &frame0, const GpuMat &frame1, GpuMat &flowx, GpuMat &flowy, Stream &s = Stream::Null());
+
+ void releaseMemory();
+
+ private:
+ /* hidden */
+ };
+
+
+
+gpu::FarnebackOpticalFlow::operator ()
+--------------------------------------
+Computes a dense optical flow using the Gunnar Farneback’s algorithm.
+
+.. ocv:function:: void gpu::FarnebackOpticalFlow::operator ()(const GpuMat &frame0, const GpuMat &frame1, GpuMat &flowx, GpuMat &flowy, Stream &s = Stream::Null())
+
+ :param frame0: First 8-bit gray-scale input image
+ :param frame1: Second 8-bit gray-scale input image
+ :param flowx: Flow horizontal component
+ :param flowy: Flow vertical component
+ :param s: Stream
+
+.. seealso:: :ocv:func:`calcOpticalFlowFarneback`
+
+
+
+gpu::FarnebackOpticalFlow::releaseMemory
+----------------------------------------
+Releases unused auxiliary memory buffers.
+
+.. ocv:function:: void gpu::FarnebackOpticalFlow::releaseMemory()
+
+
+
+gpu::PyrLKOpticalFlow
+---------------------
+.. ocv:class:: gpu::PyrLKOpticalFlow
+
+Class used for calculating an optical flow. ::
+
+ class PyrLKOpticalFlow
+ {
+ public:
+ PyrLKOpticalFlow();
+
+ void sparse(const GpuMat& prevImg, const GpuMat& nextImg, const GpuMat& prevPts, GpuMat& nextPts,
+ GpuMat& status, GpuMat* err = 0);
+
+ void dense(const GpuMat& prevImg, const GpuMat& nextImg, GpuMat& u, GpuMat& v, GpuMat* err = 0);
+
+ Size winSize;
+ int maxLevel;
+ int iters;
+ bool useInitialFlow;
+
+ void releaseMemory();
+ };
+
+The class can calculate an optical flow for a sparse feature set or dense optical flow using the iterative Lucas-Kanade method with pyramids.
+
+.. seealso:: :ocv:func:`calcOpticalFlowPyrLK`
+
+
+
+gpu::PyrLKOpticalFlow::sparse
+-----------------------------
+Calculate an optical flow for a sparse feature set.
+
+.. ocv:function:: void gpu::PyrLKOpticalFlow::sparse(const GpuMat& prevImg, const GpuMat& nextImg, const GpuMat& prevPts, GpuMat& nextPts, GpuMat& status, GpuMat* err = 0)
+
+ :param prevImg: First 8-bit input image (supports both grayscale and color images).
+
+ :param nextImg: Second input image of the same size and the same type as ``prevImg`` .
+
+ :param prevPts: Vector of 2D points for which the flow needs to be found. It must be one row matrix with CV_32FC2 type.
+
+ :param nextPts: Output vector of 2D points (with single-precision floating-point coordinates) containing the calculated new positions of input features in the second image. When ``useInitialFlow`` is true, the vector must have the same size as in the input.
+
+ :param status: Output status vector (CV_8UC1 type). Each element of the vector is set to 1 if the flow for the corresponding features has been found. Otherwise, it is set to 0.
+
+ :param err: Output vector (CV_32FC1 type) that contains the difference between patches around the original and moved points or min eigen value if ``getMinEigenVals`` is checked. It can be NULL, if not needed.
+
+.. seealso:: :ocv:func:`calcOpticalFlowPyrLK`
+
+
+
+gpu::PyrLKOpticalFlow::dense
+-----------------------------
+Calculate dense optical flow.
+
+.. ocv:function:: void gpu::PyrLKOpticalFlow::dense(const GpuMat& prevImg, const GpuMat& nextImg, GpuMat& u, GpuMat& v, GpuMat* err = 0)
+
+ :param prevImg: First 8-bit grayscale input image.
+
+ :param nextImg: Second input image of the same size and the same type as ``prevImg`` .
+
+ :param u: Horizontal component of the optical flow of the same size as input images, 32-bit floating-point, single-channel
+
+ :param v: Vertical component of the optical flow of the same size as input images, 32-bit floating-point, single-channel
+
+ :param err: Output vector (CV_32FC1 type) that contains the difference between patches around the original and moved points or min eigen value if ``getMinEigenVals`` is checked. It can be NULL, if not needed.
+
+
+
+gpu::PyrLKOpticalFlow::releaseMemory
+------------------------------------
+Releases inner buffers memory.
+
+.. ocv:function:: void gpu::PyrLKOpticalFlow::releaseMemory()
+
+
+
+gpu::interpolateFrames
+----------------------
+Interpolates frames (images) using provided optical flow (displacement field).
+
+.. ocv:function:: void gpu::interpolateFrames(const GpuMat& frame0, const GpuMat& frame1, const GpuMat& fu, const GpuMat& fv, const GpuMat& bu, const GpuMat& bv, float pos, GpuMat& newFrame, GpuMat& buf, Stream& stream = Stream::Null())
+
+ :param frame0: First frame (32-bit floating point images, single channel).
+
+ :param frame1: Second frame. Must have the same type and size as ``frame0`` .
+
+ :param fu: Forward horizontal displacement.
+
+ :param fv: Forward vertical displacement.
+
+ :param bu: Backward horizontal displacement.
+
+ :param bv: Backward vertical displacement.
+
+ :param pos: New frame position.
+
+ :param newFrame: Output image.
+
+ :param buf: Temporary buffer, will have width x 6*height size, CV_32FC1 type and contain 6 GpuMat: occlusion masks for first frame, occlusion masks for second, interpolated forward horizontal flow, interpolated forward vertical flow, interpolated backward horizontal flow, interpolated backward vertical flow.
+
+ :param stream: Stream for the asynchronous version.
+
+
+
+.. [Brox2004] T. Brox, A. Bruhn, N. Papenberg, J. Weickert. *High accuracy optical flow estimation based on a theory for warping*. ECCV 2004.
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_GPUOPTFLOW_HPP__
+#define __OPENCV_GPUOPTFLOW_HPP__
+
+#ifndef __cplusplus
+# error gpuoptflow.hpp header must be compiled as C++
+#endif
+
+#include "opencv2/core/gpumat.hpp"
+
+namespace cv { namespace gpu {
+
+class CV_EXPORTS BroxOpticalFlow
+{
+public:
+ BroxOpticalFlow(float alpha_, float gamma_, float scale_factor_, int inner_iterations_, int outer_iterations_, int solver_iterations_) :
+ alpha(alpha_), gamma(gamma_), scale_factor(scale_factor_),
+ inner_iterations(inner_iterations_), outer_iterations(outer_iterations_), solver_iterations(solver_iterations_)
+ {
+ }
+
+ //! Compute optical flow
+ //! frame0 - source frame (supports only CV_32FC1 type)
+ //! frame1 - frame to track (with the same size and type as frame0)
+ //! u - flow horizontal component (along x axis)
+ //! v - flow vertical component (along y axis)
+ void operator ()(const GpuMat& frame0, const GpuMat& frame1, GpuMat& u, GpuMat& v, Stream& stream = Stream::Null());
+
+ //! flow smoothness
+ float alpha;
+
+ //! gradient constancy importance
+ float gamma;
+
+ //! pyramid scale factor
+ float scale_factor;
+
+ //! number of lagged non-linearity iterations (inner loop)
+ int inner_iterations;
+
+ //! number of warping iterations (number of pyramid levels)
+ int outer_iterations;
+
+ //! number of linear system solver iterations
+ int solver_iterations;
+
+ GpuMat buf;
+};
+
+class CV_EXPORTS PyrLKOpticalFlow
+{
+public:
+ PyrLKOpticalFlow();
+
+ void sparse(const GpuMat& prevImg, const GpuMat& nextImg, const GpuMat& prevPts, GpuMat& nextPts,
+ GpuMat& status, GpuMat* err = 0);
+
+ void dense(const GpuMat& prevImg, const GpuMat& nextImg, GpuMat& u, GpuMat& v, GpuMat* err = 0);
+
+ void releaseMemory();
+
+ Size winSize;
+ int maxLevel;
+ int iters;
+ bool useInitialFlow;
+
+private:
+ std::vector<GpuMat> prevPyr_;
+ std::vector<GpuMat> nextPyr_;
+
+ GpuMat buf_;
+
+ GpuMat uPyr_[2];
+ GpuMat vPyr_[2];
+};
+
+class CV_EXPORTS FarnebackOpticalFlow
+{
+public:
+ FarnebackOpticalFlow()
+ {
+ numLevels = 5;
+ pyrScale = 0.5;
+ fastPyramids = false;
+ winSize = 13;
+ numIters = 10;
+ polyN = 5;
+ polySigma = 1.1;
+ flags = 0;
+ }
+
+ int numLevels;
+ double pyrScale;
+ bool fastPyramids;
+ int winSize;
+ int numIters;
+ int polyN;
+ double polySigma;
+ int flags;
+
+ void operator ()(const GpuMat &frame0, const GpuMat &frame1, GpuMat &flowx, GpuMat &flowy, Stream &s = Stream::Null());
+
+ void releaseMemory()
+ {
+ frames_[0].release();
+ frames_[1].release();
+ pyrLevel_[0].release();
+ pyrLevel_[1].release();
+ M_.release();
+ bufM_.release();
+ R_[0].release();
+ R_[1].release();
+ blurredFrame_[0].release();
+ blurredFrame_[1].release();
+ pyramid0_.clear();
+ pyramid1_.clear();
+ }
+
+private:
+ void prepareGaussian(
+ int n, double sigma, float *g, float *xg, float *xxg,
+ double &ig11, double &ig03, double &ig33, double &ig55);
+
+ void setPolynomialExpansionConsts(int n, double sigma);
+
+ void updateFlow_boxFilter(
+ const GpuMat& R0, const GpuMat& R1, GpuMat& flowx, GpuMat &flowy,
+ GpuMat& M, GpuMat &bufM, int blockSize, bool updateMatrices, Stream streams[]);
+
+ void updateFlow_gaussianBlur(
+ const GpuMat& R0, const GpuMat& R1, GpuMat& flowx, GpuMat& flowy,
+ GpuMat& M, GpuMat &bufM, int blockSize, bool updateMatrices, Stream streams[]);
+
+ GpuMat frames_[2];
+ GpuMat pyrLevel_[2], M_, bufM_, R_[2], blurredFrame_[2];
+ std::vector<GpuMat> pyramid0_, pyramid1_;
+};
+
+// Implementation of the Zach, Pock and Bischof Dual TV-L1 Optical Flow method
+//
+// see reference:
+// [1] C. Zach, T. Pock and H. Bischof, "A Duality Based Approach for Realtime TV-L1 Optical Flow".
+// [2] Javier Sanchez, Enric Meinhardt-Llopis and Gabriele Facciolo. "TV-L1 Optical Flow Estimation".
+class CV_EXPORTS OpticalFlowDual_TVL1_GPU
+{
+public:
+ OpticalFlowDual_TVL1_GPU();
+
+ void operator ()(const GpuMat& I0, const GpuMat& I1, GpuMat& flowx, GpuMat& flowy);
+
+ void collectGarbage();
+
+ /**
+ * Time step of the numerical scheme.
+ */
+ double tau;
+
+ /**
+ * Weight parameter for the data term, attachment parameter.
+ * This is the most relevant parameter, which determines the smoothness of the output.
+ * The smaller this parameter is, the smoother the solutions we obtain.
+ * It depends on the range of motions of the images, so its value should be adapted to each image sequence.
+ */
+ double lambda;
+
+ /**
+ * Weight parameter for (u - v)^2, tightness parameter.
+ * It serves as a link between the attachment and the regularization terms.
+ * In theory, it should have a small value in order to maintain both parts in correspondence.
+ * The method is stable for a large range of values of this parameter.
+ */
+ double theta;
+
+ /**
+ * Number of scales used to create the pyramid of images.
+ */
+ int nscales;
+
+ /**
+ * Number of warpings per scale.
+ * Represents the number of times that I1(x+u0) and grad( I1(x+u0) ) are computed per scale.
+ * This is a parameter that assures the stability of the method.
+ * It also affects the running time, so it is a compromise between speed and accuracy.
+ */
+ int warps;
+
+ /**
+ * Stopping criterion threshold used in the numerical scheme, which is a trade-off between precision and running time.
+ * A small value will yield more accurate solutions at the expense of a slower convergence.
+ */
+ double epsilon;
+
+ /**
+ * Stopping criterion iterations number used in the numerical scheme.
+ */
+ int iterations;
+
+ double scaleStep;
+
+ bool useInitialFlow;
+
+private:
+ void procOneScale(const GpuMat& I0, const GpuMat& I1, GpuMat& u1, GpuMat& u2);
+
+ std::vector<GpuMat> I0s;
+ std::vector<GpuMat> I1s;
+ std::vector<GpuMat> u1s;
+ std::vector<GpuMat> u2s;
+
+ GpuMat I1x_buf;
+ GpuMat I1y_buf;
+
+ GpuMat I1w_buf;
+ GpuMat I1wx_buf;
+ GpuMat I1wy_buf;
+
+ GpuMat grad_buf;
+ GpuMat rho_c_buf;
+
+ GpuMat p11_buf;
+ GpuMat p12_buf;
+ GpuMat p21_buf;
+ GpuMat p22_buf;
+
+ GpuMat diff_buf;
+ GpuMat norm_buf;
+};
+
+//! Calculates optical flow for 2 images using block matching algorithm */
+CV_EXPORTS void calcOpticalFlowBM(const GpuMat& prev, const GpuMat& curr,
+ Size block_size, Size shift_size, Size max_range, bool use_previous,
+ GpuMat& velx, GpuMat& vely, GpuMat& buf,
+ Stream& stream = Stream::Null());
+
+class CV_EXPORTS FastOpticalFlowBM
+{
+public:
+ void operator ()(const GpuMat& I0, const GpuMat& I1, GpuMat& flowx, GpuMat& flowy, int search_window = 21, int block_window = 7, Stream& s = Stream::Null());
+
+private:
+ GpuMat buffer;
+ GpuMat extended_I0;
+ GpuMat extended_I1;
+};
+
+//! Interpolate frames (images) using provided optical flow (displacement field).
+//! frame0 - frame 0 (32-bit floating point images, single channel)
+//! frame1 - frame 1 (the same type and size)
+//! fu - forward horizontal displacement
+//! fv - forward vertical displacement
+//! bu - backward horizontal displacement
+//! bv - backward vertical displacement
+//! pos - new frame position
+//! newFrame - new frame
+//! buf - temporary buffer, will have width x 6*height size, CV_32FC1 type and contain 6 GpuMat;
+//! occlusion masks 0, occlusion masks 1,
+//! interpolated forward flow 0, interpolated forward flow 1,
+//! interpolated backward flow 0, interpolated backward flow 1
+//!
+CV_EXPORTS void interpolateFrames(const GpuMat& frame0, const GpuMat& frame1,
+ const GpuMat& fu, const GpuMat& fv,
+ const GpuMat& bu, const GpuMat& bv,
+ float pos, GpuMat& newFrame, GpuMat& buf,
+ Stream& stream = Stream::Null());
+
+CV_EXPORTS void createOpticalFlowNeedleMap(const GpuMat& u, const GpuMat& v, GpuMat& vertex, GpuMat& colors);
+
+}} // namespace cv { namespace gpu {
+
+#endif /* __OPENCV_GPUOPTFLOW_HPP__ */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace perf;
+
+CV_PERF_TEST_MAIN(gpuoptflow, printCudaInfo())
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+#include "opencv2/legacy.hpp"
+
+using namespace std;
+using namespace testing;
+using namespace perf;
+
+//////////////////////////////////////////////////////
+// InterpolateFrames
+
+typedef pair<string, string> pair_string;
+
+DEF_PARAM_TEST_1(ImagePair, pair_string);
+
+PERF_TEST_P(ImagePair, InterpolateFrames,
+ Values<pair_string>(make_pair("gpu/opticalflow/frame0.png", "gpu/opticalflow/frame1.png")))
+{
+ cv::Mat frame0 = readImage(GetParam().first, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(frame0.empty());
+
+ cv::Mat frame1 = readImage(GetParam().second, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(frame1.empty());
+
+ frame0.convertTo(frame0, CV_32FC1, 1.0 / 255.0);
+ frame1.convertTo(frame1, CV_32FC1, 1.0 / 255.0);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_frame0(frame0);
+ const cv::gpu::GpuMat d_frame1(frame1);
+ cv::gpu::GpuMat d_fu, d_fv;
+ cv::gpu::GpuMat d_bu, d_bv;
+
+ cv::gpu::BroxOpticalFlow d_flow(0.197f /*alpha*/, 50.0f /*gamma*/, 0.8f /*scale_factor*/,
+ 10 /*inner_iterations*/, 77 /*outer_iterations*/, 10 /*solver_iterations*/);
+
+ d_flow(d_frame0, d_frame1, d_fu, d_fv);
+ d_flow(d_frame1, d_frame0, d_bu, d_bv);
+
+ cv::gpu::GpuMat newFrame;
+ cv::gpu::GpuMat d_buf;
+
+ TEST_CYCLE() cv::gpu::interpolateFrames(d_frame0, d_frame1, d_fu, d_fv, d_bu, d_bv, 0.5f, newFrame, d_buf);
+
+ GPU_SANITY_CHECK(newFrame);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
+
+//////////////////////////////////////////////////////
+// CreateOpticalFlowNeedleMap
+
+PERF_TEST_P(ImagePair, CreateOpticalFlowNeedleMap,
+ Values<pair_string>(make_pair("gpu/opticalflow/frame0.png", "gpu/opticalflow/frame1.png")))
+{
+ cv::Mat frame0 = readImage(GetParam().first, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(frame0.empty());
+
+ cv::Mat frame1 = readImage(GetParam().second, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(frame1.empty());
+
+ frame0.convertTo(frame0, CV_32FC1, 1.0 / 255.0);
+ frame1.convertTo(frame1, CV_32FC1, 1.0 / 255.0);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_frame0(frame0);
+ const cv::gpu::GpuMat d_frame1(frame1);
+ cv::gpu::GpuMat u;
+ cv::gpu::GpuMat v;
+
+ cv::gpu::BroxOpticalFlow d_flow(0.197f /*alpha*/, 50.0f /*gamma*/, 0.8f /*scale_factor*/,
+ 10 /*inner_iterations*/, 77 /*outer_iterations*/, 10 /*solver_iterations*/);
+
+ d_flow(d_frame0, d_frame1, u, v);
+
+ cv::gpu::GpuMat vertex, colors;
+
+ TEST_CYCLE() cv::gpu::createOpticalFlowNeedleMap(u, v, vertex, colors);
+
+ GPU_SANITY_CHECK(vertex);
+ GPU_SANITY_CHECK(colors);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
+
+//////////////////////////////////////////////////////
+// BroxOpticalFlow
+
+PERF_TEST_P(ImagePair, BroxOpticalFlow,
+ Values<pair_string>(make_pair("gpu/opticalflow/frame0.png", "gpu/opticalflow/frame1.png")))
+{
+ declare.time(300);
+
+ cv::Mat frame0 = readImage(GetParam().first, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(frame0.empty());
+
+ cv::Mat frame1 = readImage(GetParam().second, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(frame1.empty());
+
+ frame0.convertTo(frame0, CV_32FC1, 1.0 / 255.0);
+ frame1.convertTo(frame1, CV_32FC1, 1.0 / 255.0);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_frame0(frame0);
+ const cv::gpu::GpuMat d_frame1(frame1);
+ cv::gpu::GpuMat u;
+ cv::gpu::GpuMat v;
+
+ cv::gpu::BroxOpticalFlow d_flow(0.197f /*alpha*/, 50.0f /*gamma*/, 0.8f /*scale_factor*/,
+ 10 /*inner_iterations*/, 77 /*outer_iterations*/, 10 /*solver_iterations*/);
+
+ TEST_CYCLE() d_flow(d_frame0, d_frame1, u, v);
+
+ GPU_SANITY_CHECK(u);
+ GPU_SANITY_CHECK(v);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
+
+//////////////////////////////////////////////////////
+// PyrLKOpticalFlowSparse
+
+DEF_PARAM_TEST(ImagePair_Gray_NPts_WinSz_Levels_Iters, pair_string, bool, int, int, int, int);
+
+PERF_TEST_P(ImagePair_Gray_NPts_WinSz_Levels_Iters, PyrLKOpticalFlowSparse,
+ Combine(Values<pair_string>(make_pair("gpu/opticalflow/frame0.png", "gpu/opticalflow/frame1.png")),
+ Bool(),
+ Values(8000),
+ Values(21),
+ Values(1, 3),
+ Values(1, 30)))
+{
+ declare.time(20.0);
+
+ const pair_string imagePair = GET_PARAM(0);
+ const bool useGray = GET_PARAM(1);
+ const int points = GET_PARAM(2);
+ const int winSize = GET_PARAM(3);
+ const int levels = GET_PARAM(4);
+ const int iters = GET_PARAM(5);
+
+ const cv::Mat frame0 = readImage(imagePair.first, useGray ? cv::IMREAD_GRAYSCALE : cv::IMREAD_COLOR);
+ ASSERT_FALSE(frame0.empty());
+
+ const cv::Mat frame1 = readImage(imagePair.second, useGray ? cv::IMREAD_GRAYSCALE : cv::IMREAD_COLOR);
+ ASSERT_FALSE(frame1.empty());
+
+ cv::Mat gray_frame;
+ if (useGray)
+ gray_frame = frame0;
+ else
+ cv::cvtColor(frame0, gray_frame, cv::COLOR_BGR2GRAY);
+
+ cv::Mat pts;
+ cv::goodFeaturesToTrack(gray_frame, pts, points, 0.01, 0.0);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_pts(pts.reshape(2, 1));
+
+ cv::gpu::PyrLKOpticalFlow d_pyrLK;
+ d_pyrLK.winSize = cv::Size(winSize, winSize);
+ d_pyrLK.maxLevel = levels - 1;
+ d_pyrLK.iters = iters;
+
+ const cv::gpu::GpuMat d_frame0(frame0);
+ const cv::gpu::GpuMat d_frame1(frame1);
+ cv::gpu::GpuMat nextPts;
+ cv::gpu::GpuMat status;
+
+ TEST_CYCLE() d_pyrLK.sparse(d_frame0, d_frame1, d_pts, nextPts, status);
+
+ GPU_SANITY_CHECK(nextPts);
+ GPU_SANITY_CHECK(status);
+ }
+ else
+ {
+ cv::Mat nextPts;
+ cv::Mat status;
+
+ TEST_CYCLE()
+ {
+ cv::calcOpticalFlowPyrLK(frame0, frame1, pts, nextPts, status, cv::noArray(),
+ cv::Size(winSize, winSize), levels - 1,
+ cv::TermCriteria(cv::TermCriteria::COUNT + cv::TermCriteria::EPS, iters, 0.01));
+ }
+
+ CPU_SANITY_CHECK(nextPts);
+ CPU_SANITY_CHECK(status);
+ }
+}
+
+//////////////////////////////////////////////////////
+// PyrLKOpticalFlowDense
+
+DEF_PARAM_TEST(ImagePair_WinSz_Levels_Iters, pair_string, int, int, int);
+
+PERF_TEST_P(ImagePair_WinSz_Levels_Iters, PyrLKOpticalFlowDense,
+ Combine(Values<pair_string>(make_pair("gpu/opticalflow/frame0.png", "gpu/opticalflow/frame1.png")),
+ Values(3, 5, 7, 9, 13, 17, 21),
+ Values(1, 3),
+ Values(1, 10)))
+{
+ declare.time(30);
+
+ const pair_string imagePair = GET_PARAM(0);
+ const int winSize = GET_PARAM(1);
+ const int levels = GET_PARAM(2);
+ const int iters = GET_PARAM(3);
+
+ const cv::Mat frame0 = readImage(imagePair.first, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(frame0.empty());
+
+ const cv::Mat frame1 = readImage(imagePair.second, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(frame1.empty());
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_frame0(frame0);
+ const cv::gpu::GpuMat d_frame1(frame1);
+ cv::gpu::GpuMat u;
+ cv::gpu::GpuMat v;
+
+ cv::gpu::PyrLKOpticalFlow d_pyrLK;
+ d_pyrLK.winSize = cv::Size(winSize, winSize);
+ d_pyrLK.maxLevel = levels - 1;
+ d_pyrLK.iters = iters;
+
+ TEST_CYCLE() d_pyrLK.dense(d_frame0, d_frame1, u, v);
+
+ GPU_SANITY_CHECK(u);
+ GPU_SANITY_CHECK(v);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
+
+//////////////////////////////////////////////////////
+// FarnebackOpticalFlow
+
+PERF_TEST_P(ImagePair, FarnebackOpticalFlow,
+ Values<pair_string>(make_pair("gpu/opticalflow/frame0.png", "gpu/opticalflow/frame1.png")))
+{
+ declare.time(10);
+
+ const cv::Mat frame0 = readImage(GetParam().first, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(frame0.empty());
+
+ const cv::Mat frame1 = readImage(GetParam().second, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(frame1.empty());
+
+ const int numLevels = 5;
+ const double pyrScale = 0.5;
+ const int winSize = 13;
+ const int numIters = 10;
+ const int polyN = 5;
+ const double polySigma = 1.1;
+ const int flags = 0;
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_frame0(frame0);
+ const cv::gpu::GpuMat d_frame1(frame1);
+ cv::gpu::GpuMat u;
+ cv::gpu::GpuMat v;
+
+ cv::gpu::FarnebackOpticalFlow d_farneback;
+ d_farneback.numLevels = numLevels;
+ d_farneback.pyrScale = pyrScale;
+ d_farneback.winSize = winSize;
+ d_farneback.numIters = numIters;
+ d_farneback.polyN = polyN;
+ d_farneback.polySigma = polySigma;
+ d_farneback.flags = flags;
+
+ TEST_CYCLE() d_farneback(d_frame0, d_frame1, u, v);
+
+ GPU_SANITY_CHECK(u, 1e-4);
+ GPU_SANITY_CHECK(v, 1e-4);
+ }
+ else
+ {
+ cv::Mat flow;
+
+ TEST_CYCLE() cv::calcOpticalFlowFarneback(frame0, frame1, flow, pyrScale, numLevels, winSize, numIters, polyN, polySigma, flags);
+
+ CPU_SANITY_CHECK(flow);
+ }
+}
+
+//////////////////////////////////////////////////////
+// OpticalFlowDual_TVL1
+
+PERF_TEST_P(ImagePair, OpticalFlowDual_TVL1,
+ Values<pair_string>(make_pair("gpu/opticalflow/frame0.png", "gpu/opticalflow/frame1.png")))
+{
+ declare.time(20);
+
+ const cv::Mat frame0 = readImage(GetParam().first, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(frame0.empty());
+
+ const cv::Mat frame1 = readImage(GetParam().second, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(frame1.empty());
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_frame0(frame0);
+ const cv::gpu::GpuMat d_frame1(frame1);
+ cv::gpu::GpuMat u;
+ cv::gpu::GpuMat v;
+
+ cv::gpu::OpticalFlowDual_TVL1_GPU d_alg;
+
+ TEST_CYCLE() d_alg(d_frame0, d_frame1, u, v);
+
+ GPU_SANITY_CHECK(u, 1e-2);
+ GPU_SANITY_CHECK(v, 1e-2);
+ }
+ else
+ {
+ cv::Mat flow;
+
+ cv::Ptr<cv::DenseOpticalFlow> alg = cv::createOptFlow_DualTVL1();
+ alg->set("medianFiltering", 1);
+ alg->set("innerIterations", 1);
+ alg->set("outerIterations", 300);
+
+ TEST_CYCLE() alg->calc(frame0, frame1, flow);
+
+ CPU_SANITY_CHECK(flow);
+ }
+}
+
+//////////////////////////////////////////////////////
+// OpticalFlowBM
+
+void calcOpticalFlowBM(const cv::Mat& prev, const cv::Mat& curr,
+ cv::Size bSize, cv::Size shiftSize, cv::Size maxRange, int usePrevious,
+ cv::Mat& velx, cv::Mat& vely)
+{
+ cv::Size sz((curr.cols - bSize.width + shiftSize.width)/shiftSize.width, (curr.rows - bSize.height + shiftSize.height)/shiftSize.height);
+
+ velx.create(sz, CV_32FC1);
+ vely.create(sz, CV_32FC1);
+
+ CvMat cvprev = prev;
+ CvMat cvcurr = curr;
+
+ CvMat cvvelx = velx;
+ CvMat cvvely = vely;
+
+ cvCalcOpticalFlowBM(&cvprev, &cvcurr, bSize, shiftSize, maxRange, usePrevious, &cvvelx, &cvvely);
+}
+
+PERF_TEST_P(ImagePair, OpticalFlowBM,
+ Values<pair_string>(make_pair("gpu/opticalflow/frame0.png", "gpu/opticalflow/frame1.png")))
+{
+ declare.time(400);
+
+ const cv::Mat frame0 = readImage(GetParam().first, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(frame0.empty());
+
+ const cv::Mat frame1 = readImage(GetParam().second, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(frame1.empty());
+
+ const cv::Size block_size(16, 16);
+ const cv::Size shift_size(1, 1);
+ const cv::Size max_range(16, 16);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_frame0(frame0);
+ const cv::gpu::GpuMat d_frame1(frame1);
+ cv::gpu::GpuMat u, v, buf;
+
+ TEST_CYCLE() cv::gpu::calcOpticalFlowBM(d_frame0, d_frame1, block_size, shift_size, max_range, false, u, v, buf);
+
+ GPU_SANITY_CHECK(u);
+ GPU_SANITY_CHECK(v);
+ }
+ else
+ {
+ cv::Mat u, v;
+
+ TEST_CYCLE() calcOpticalFlowBM(frame0, frame1, block_size, shift_size, max_range, false, u, v);
+
+ CPU_SANITY_CHECK(u);
+ CPU_SANITY_CHECK(v);
+ }
+}
+
+PERF_TEST_P(ImagePair, FastOpticalFlowBM,
+ Values<pair_string>(make_pair("gpu/opticalflow/frame0.png", "gpu/opticalflow/frame1.png")))
+{
+ declare.time(400);
+
+ const cv::Mat frame0 = readImage(GetParam().first, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(frame0.empty());
+
+ const cv::Mat frame1 = readImage(GetParam().second, cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(frame1.empty());
+
+ const cv::Size block_size(16, 16);
+ const cv::Size shift_size(1, 1);
+ const cv::Size max_range(16, 16);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_frame0(frame0);
+ const cv::gpu::GpuMat d_frame1(frame1);
+ cv::gpu::GpuMat u, v;
+
+ cv::gpu::FastOpticalFlowBM fastBM;
+
+ TEST_CYCLE() fastBM(d_frame0, d_frame1, u, v, max_range.width, block_size.width);
+
+ GPU_SANITY_CHECK(u, 2);
+ GPU_SANITY_CHECK(v, 2);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifdef __GNUC__
+# pragma GCC diagnostic ignored "-Wmissing-declarations"
+# if defined __clang__ || defined __APPLE__
+# pragma GCC diagnostic ignored "-Wmissing-prototypes"
+# pragma GCC diagnostic ignored "-Wextra"
+# endif
+#endif
+
+#ifndef __OPENCV_PERF_PRECOMP_HPP__
+#define __OPENCV_PERF_PRECOMP_HPP__
+
+#include "opencv2/ts.hpp"
+#include "opencv2/ts/gpu_perf.hpp"
+
+#include "opencv2/gpuoptflow.hpp"
+#include "opencv2/video.hpp"
+
+#ifdef GTEST_CREATE_SHARED_LIBRARY
+#error no modules except ts should have GTEST_CREATE_SHARED_LIBRARY defined
+#endif
+
+#endif
void cv::gpu::calcOpticalFlowBM(const GpuMat&, const GpuMat&, Size, Size, Size, bool, GpuMat&, GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::FastOpticalFlowBM::operator ()(const GpuMat&, const GpuMat&, GpuMat&, GpuMat&, int, int, Stream&) { throw_no_cuda(); }
-
#else // HAVE_CUDA
namespace optflowbm
// upper side
for (int j = -i; j <= i + 1; ++j, ++ssCount)
{
- ss[ssCount].x = ++x;
- ss[ssCount].y = y;
+ ss[ssCount].x = (short) ++x;
+ ss[ssCount].y = (short) y;
}
// right side
for (int j = -i; j <= i + 1; ++j, ++ssCount)
{
- ss[ssCount].x = x;
- ss[ssCount].y = ++y;
+ ss[ssCount].x = (short) x;
+ ss[ssCount].y = (short) ++y;
}
// bottom side
for (int j = -i; j <= i + 1; ++j, ++ssCount)
{
- ss[ssCount].x = --x;
- ss[ssCount].y = y;
+ ss[ssCount].x = (short) --x;
+ ss[ssCount].y = (short) y;
}
// left side
for (int j = -i; j <= i + 1; ++j, ++ssCount)
{
- ss[ssCount].x = x;
- ss[ssCount].y = --y;
+ ss[ssCount].x = (short) x;
+ ss[ssCount].y = (short) --y;
}
}
// upper side
for (int j = -maxRange.width; j <= maxRange.width; ++j, ++ssCount, ++x)
{
- ss[ssCount].x = x;
- ss[ssCount].y = y;
+ ss[ssCount].x = (short) x;
+ ss[ssCount].y = (short) y;
}
x = xleft;
// bottom side
for (int j = -maxRange.width; j <= maxRange.width; ++j, ++ssCount, ++x)
{
- ss[ssCount].x = x;
- ss[ssCount].y = y;
+ ss[ssCount].x = (short) x;
+ ss[ssCount].y = (short) y;
}
}
}
// left side
for (int j = -maxRange.height; j <= maxRange.height; ++j, ++ssCount, ++y)
{
- ss[ssCount].x = x;
- ss[ssCount].y = y;
+ ss[ssCount].x = (short) x;
+ ss[ssCount].y = (short) y;
}
y = yupper;
// right side
for (int j = -maxRange.height; j <= maxRange.height; ++j, ++ssCount, ++y)
{
- ss[ssCount].x = x;
- ss[ssCount].y = y;
+ ss[ssCount].x = (short) x;
+ ss[ssCount].y = (short) y;
}
}
}
maxX, maxY, acceptLevel, escapeLevel, buf.ptr<short2>(), ssCount, stream);
}
-namespace optflowbm_fast
-{
- void get_buffer_size(int src_cols, int src_rows, int search_window, int block_window, int& buffer_cols, int& buffer_rows);
-
- template <typename T>
- void calc(PtrStepSzb I0, PtrStepSzb I1, PtrStepSzf velx, PtrStepSzf vely, PtrStepi buffer, int search_window, int block_window, cudaStream_t stream);
-}
-
-void cv::gpu::FastOpticalFlowBM::operator ()(const GpuMat& I0, const GpuMat& I1, GpuMat& flowx, GpuMat& flowy, int search_window, int block_window, Stream& stream)
-{
- CV_Assert( I0.type() == CV_8UC1 );
- CV_Assert( I1.size() == I0.size() && I1.type() == I0.type() );
-
- int border_size = search_window / 2 + block_window / 2;
- Size esize = I0.size() + Size(border_size, border_size) * 2;
-
- ensureSizeIsEnough(esize, I0.type(), extended_I0);
- ensureSizeIsEnough(esize, I0.type(), extended_I1);
-
- copyMakeBorder(I0, extended_I0, border_size, border_size, border_size, border_size, cv::BORDER_DEFAULT, Scalar(), stream);
- copyMakeBorder(I1, extended_I1, border_size, border_size, border_size, border_size, cv::BORDER_DEFAULT, Scalar(), stream);
-
- GpuMat I0_hdr = extended_I0(Rect(Point2i(border_size, border_size), I0.size()));
- GpuMat I1_hdr = extended_I1(Rect(Point2i(border_size, border_size), I0.size()));
-
- int bcols, brows;
- optflowbm_fast::get_buffer_size(I0.cols, I0.rows, search_window, block_window, bcols, brows);
-
- ensureSizeIsEnough(brows, bcols, CV_32SC1, buffer);
-
- flowx.create(I0.size(), CV_32FC1);
- flowy.create(I0.size(), CV_32FC1);
-
- optflowbm_fast::calc<uchar>(I0_hdr, I1_hdr, flowx, flowy, buffer, search_window, block_window, StreamAccessor::getStream(stream));
-}
-
#endif // HAVE_CUDA
//
//M*/
-#ifndef __FRAME_QUEUE_H__
-#define __FRAME_QUEUE_H__
-
#include "precomp.hpp"
-#include "thread_wrappers.h"
-#if defined(HAVE_CUDA) && defined(HAVE_NVCUVID)
+using namespace cv;
+using namespace cv::gpu;
-namespace cv { namespace gpu
-{
- namespace detail
- {
- class FrameQueue
- {
- public:
- static const int MaximumSize = 20; // MAX_FRM_CNT;
+#if !defined HAVE_CUDA || defined(CUDA_DISABLER)
- FrameQueue();
+void cv::gpu::FastOpticalFlowBM::operator ()(const GpuMat&, const GpuMat&, GpuMat&, GpuMat&, int, int, Stream&) { throw_no_cuda(); }
- void endDecode() { endOfDecode_ = true; }
- bool isEndOfDecode() const { return endOfDecode_ != 0;}
+#else // HAVE_CUDA
- // Spins until frame becomes available or decoding gets canceled.
- // If the requested frame is available the method returns true.
- // If decoding was interupted before the requested frame becomes
- // available, the method returns false.
- bool waitUntilFrameAvailable(int pictureIndex);
+namespace optflowbm_fast
+{
+ void get_buffer_size(int src_cols, int src_rows, int search_window, int block_window, int& buffer_cols, int& buffer_rows);
- void enqueue(const CUVIDPARSERDISPINFO* picParams);
+ template <typename T>
+ void calc(PtrStepSzb I0, PtrStepSzb I1, PtrStepSzf velx, PtrStepSzf vely, PtrStepi buffer, int search_window, int block_window, cudaStream_t stream);
+}
- // Deque the next frame.
- // Parameters:
- // displayInfo - New frame info gets placed into this object.
- // Returns:
- // true, if a new frame was returned,
- // false, if the queue was empty and no new frame could be returned.
- bool dequeue(CUVIDPARSERDISPINFO& displayInfo);
+void cv::gpu::FastOpticalFlowBM::operator ()(const GpuMat& I0, const GpuMat& I1, GpuMat& flowx, GpuMat& flowy, int search_window, int block_window, Stream& stream)
+{
+ CV_Assert( I0.type() == CV_8UC1 );
+ CV_Assert( I1.size() == I0.size() && I1.type() == I0.type() );
- void releaseFrame(const CUVIDPARSERDISPINFO& picParams) { isFrameInUse_[picParams.picture_index] = false; }
+ int border_size = search_window / 2 + block_window / 2;
+ Size esize = I0.size() + Size(border_size, border_size) * 2;
- private:
- FrameQueue(const FrameQueue&);
- FrameQueue& operator =(const FrameQueue&);
+ ensureSizeIsEnough(esize, I0.type(), extended_I0);
+ ensureSizeIsEnough(esize, I0.type(), extended_I1);
- bool isInUse(int pictureIndex) const { return isFrameInUse_[pictureIndex] != 0; }
+ gpu::copyMakeBorder(I0, extended_I0, border_size, border_size, border_size, border_size, cv::BORDER_DEFAULT, Scalar(), stream);
+ gpu::copyMakeBorder(I1, extended_I1, border_size, border_size, border_size, border_size, cv::BORDER_DEFAULT, Scalar(), stream);
- CriticalSection criticalSection_;
+ GpuMat I0_hdr = extended_I0(Rect(Point2i(border_size, border_size), I0.size()));
+ GpuMat I1_hdr = extended_I1(Rect(Point2i(border_size, border_size), I0.size()));
- volatile int isFrameInUse_[MaximumSize];
- volatile int endOfDecode_;
+ int bcols, brows;
+ optflowbm_fast::get_buffer_size(I0.cols, I0.rows, search_window, block_window, bcols, brows);
- int framesInQueue_;
- int readPosition_;
- CUVIDPARSERDISPINFO displayQueue_[MaximumSize];
- };
- }
-}}
+ ensureSizeIsEnough(brows, bcols, CV_32SC1, buffer);
-#endif // HAVE_CUDA
+ flowx.create(I0.size(), CV_32FC1);
+ flowy.create(I0.size(), CV_32FC1);
+
+ optflowbm_fast::calc<uchar>(I0_hdr, I1_hdr, flowx, flowy, buffer, search_window, block_window, StreamAccessor::getStream(stream));
+}
-#endif // __FRAME_QUEUE_H__
+#endif // HAVE_CUDA
using namespace cv;
using namespace cv::gpu;
-#if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
+#if !defined (HAVE_CUDA) || !defined (HAVE_OPENCV_GPULEGACY) || defined (CUDA_DISABLER)
void cv::gpu::BroxOpticalFlow::operator ()(const GpuMat&, const GpuMat&, GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::interpolateFrames(const GpuMat&, const GpuMat&, const GpuMat&, const GpuMat&, const GpuMat&, const GpuMat&, float, GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
-void cv::gpu::createOpticalFlowNeedleMap(const GpuMat&, const GpuMat&, GpuMat&, GpuMat&) { throw_no_cuda(); }
#else
ncvSafeCall( NCVBroxOpticalFlow(desc, gpuAllocator, frame0Mat, frame1Mat, uMat, vMat, stream) );
}
-void cv::gpu::interpolateFrames(const GpuMat& frame0, const GpuMat& frame1, const GpuMat& fu, const GpuMat& fv, const GpuMat& bu, const GpuMat& bv,
- float pos, GpuMat& newFrame, GpuMat& buf, Stream& s)
-{
- CV_Assert(frame0.type() == CV_32FC1);
- CV_Assert(frame1.size() == frame0.size() && frame1.type() == frame0.type());
- CV_Assert(fu.size() == frame0.size() && fu.type() == frame0.type());
- CV_Assert(fv.size() == frame0.size() && fv.type() == frame0.type());
- CV_Assert(bu.size() == frame0.size() && bu.type() == frame0.type());
- CV_Assert(bv.size() == frame0.size() && bv.type() == frame0.type());
-
- newFrame.create(frame0.size(), frame0.type());
-
- buf.create(6 * frame0.rows, frame0.cols, CV_32FC1);
- buf.setTo(Scalar::all(0));
-
- // occlusion masks
- GpuMat occ0 = buf.rowRange(0 * frame0.rows, 1 * frame0.rows);
- GpuMat occ1 = buf.rowRange(1 * frame0.rows, 2 * frame0.rows);
-
- // interpolated forward flow
- GpuMat fui = buf.rowRange(2 * frame0.rows, 3 * frame0.rows);
- GpuMat fvi = buf.rowRange(3 * frame0.rows, 4 * frame0.rows);
-
- // interpolated backward flow
- GpuMat bui = buf.rowRange(4 * frame0.rows, 5 * frame0.rows);
- GpuMat bvi = buf.rowRange(5 * frame0.rows, 6 * frame0.rows);
-
- size_t step = frame0.step;
-
- CV_Assert(frame1.step == step && fu.step == step && fv.step == step && bu.step == step && bv.step == step && newFrame.step == step && buf.step == step);
-
- cudaStream_t stream = StreamAccessor::getStream(s);
- NppStStreamHandler h(stream);
-
- NppStInterpolationState state;
-
- state.size = NcvSize32u(frame0.cols, frame0.rows);
- state.nStep = static_cast<Ncv32u>(step);
- state.pSrcFrame0 = const_cast<Ncv32f*>(frame0.ptr<Ncv32f>());
- state.pSrcFrame1 = const_cast<Ncv32f*>(frame1.ptr<Ncv32f>());
- state.pFU = const_cast<Ncv32f*>(fu.ptr<Ncv32f>());
- state.pFV = const_cast<Ncv32f*>(fv.ptr<Ncv32f>());
- state.pBU = const_cast<Ncv32f*>(bu.ptr<Ncv32f>());
- state.pBV = const_cast<Ncv32f*>(bv.ptr<Ncv32f>());
- state.pos = pos;
- state.pNewFrame = newFrame.ptr<Ncv32f>();
- state.ppBuffers[0] = occ0.ptr<Ncv32f>();
- state.ppBuffers[1] = occ1.ptr<Ncv32f>();
- state.ppBuffers[2] = fui.ptr<Ncv32f>();
- state.ppBuffers[3] = fvi.ptr<Ncv32f>();
- state.ppBuffers[4] = bui.ptr<Ncv32f>();
- state.ppBuffers[5] = bvi.ptr<Ncv32f>();
-
- ncvSafeCall( nppiStInterpolateFrames(&state) );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
-}
-
-namespace cv { namespace gpu { namespace cudev
-{
- namespace optical_flow
- {
- void NeedleMapAverage_gpu(PtrStepSzf u, PtrStepSzf v, PtrStepSzf u_avg, PtrStepSzf v_avg);
- void CreateOpticalFlowNeedleMap_gpu(PtrStepSzf u_avg, PtrStepSzf v_avg, float* vertex_buffer, float* color_data, float max_flow, float xscale, float yscale);
- }
-}}}
-
-void cv::gpu::createOpticalFlowNeedleMap(const GpuMat& u, const GpuMat& v, GpuMat& vertex, GpuMat& colors)
-{
- using namespace cv::gpu::cudev::optical_flow;
-
- CV_Assert(u.type() == CV_32FC1);
- CV_Assert(v.type() == u.type() && v.size() == u.size());
-
- const int NEEDLE_MAP_SCALE = 16;
-
- const int x_needles = u.cols / NEEDLE_MAP_SCALE;
- const int y_needles = u.rows / NEEDLE_MAP_SCALE;
-
- GpuMat u_avg(y_needles, x_needles, CV_32FC1);
- GpuMat v_avg(y_needles, x_needles, CV_32FC1);
-
- NeedleMapAverage_gpu(u, v, u_avg, v_avg);
-
- const int NUM_VERTS_PER_ARROW = 6;
-
- const int num_arrows = x_needles * y_needles * NUM_VERTS_PER_ARROW;
-
- vertex.create(1, num_arrows, CV_32FC3);
- colors.create(1, num_arrows, CV_32FC3);
-
- colors.setTo(Scalar::all(1.0));
-
- double uMax, vMax;
- minMax(u_avg, 0, &uMax);
- minMax(v_avg, 0, &vMax);
-
- float max_flow = static_cast<float>(std::sqrt(uMax * uMax + vMax * vMax));
-
- CreateOpticalFlowNeedleMap_gpu(u_avg, v_avg, vertex.ptr<float>(), colors.ptr<float>(), max_flow, 1.0f / u.cols, 1.0f / u.rows);
-
- cvtColor(colors, colors, COLOR_HSV2RGB);
-}
-
#endif /* HAVE_CUDA */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/limits.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/reduce.hpp"
+
+using namespace cv::gpu;
+using namespace cv::gpu::cudev;
+
+namespace optflowbm
+{
+ texture<uchar, cudaTextureType2D, cudaReadModeElementType> tex_prev(false, cudaFilterModePoint, cudaAddressModeClamp);
+ texture<uchar, cudaTextureType2D, cudaReadModeElementType> tex_curr(false, cudaFilterModePoint, cudaAddressModeClamp);
+
+ __device__ int cmpBlocks(int X1, int Y1, int X2, int Y2, int2 blockSize)
+ {
+ int s = 0;
+
+ for (int y = 0; y < blockSize.y; ++y)
+ {
+ for (int x = 0; x < blockSize.x; ++x)
+ s += ::abs(tex2D(tex_prev, X1 + x, Y1 + y) - tex2D(tex_curr, X2 + x, Y2 + y));
+ }
+
+ return s;
+ }
+
+ __global__ void calcOptFlowBM(PtrStepSzf velx, PtrStepf vely, const int2 blockSize, const int2 shiftSize, const bool usePrevious,
+ const int maxX, const int maxY, const int acceptLevel, const int escapeLevel,
+ const short2* ss, const int ssCount)
+ {
+ const int j = blockIdx.x * blockDim.x + threadIdx.x;
+ const int i = blockIdx.y * blockDim.y + threadIdx.y;
+
+ if (i >= velx.rows || j >= velx.cols)
+ return;
+
+ const int X1 = j * shiftSize.x;
+ const int Y1 = i * shiftSize.y;
+
+ const int offX = usePrevious ? __float2int_rn(velx(i, j)) : 0;
+ const int offY = usePrevious ? __float2int_rn(vely(i, j)) : 0;
+
+ int X2 = X1 + offX;
+ int Y2 = Y1 + offY;
+
+ int dist = numeric_limits<int>::max();
+
+ if (0 <= X2 && X2 <= maxX && 0 <= Y2 && Y2 <= maxY)
+ dist = cmpBlocks(X1, Y1, X2, Y2, blockSize);
+
+ int countMin = 1;
+ int sumx = offX;
+ int sumy = offY;
+
+ if (dist > acceptLevel)
+ {
+ // do brute-force search
+ for (int k = 0; k < ssCount; ++k)
+ {
+ const short2 ssVal = ss[k];
+
+ const int dx = offX + ssVal.x;
+ const int dy = offY + ssVal.y;
+
+ X2 = X1 + dx;
+ Y2 = Y1 + dy;
+
+ if (0 <= X2 && X2 <= maxX && 0 <= Y2 && Y2 <= maxY)
+ {
+ const int tmpDist = cmpBlocks(X1, Y1, X2, Y2, blockSize);
+ if (tmpDist < acceptLevel)
+ {
+ sumx = dx;
+ sumy = dy;
+ countMin = 1;
+ break;
+ }
+
+ if (tmpDist < dist)
+ {
+ dist = tmpDist;
+ sumx = dx;
+ sumy = dy;
+ countMin = 1;
+ }
+ else if (tmpDist == dist)
+ {
+ sumx += dx;
+ sumy += dy;
+ countMin++;
+ }
+ }
+ }
+
+ if (dist > escapeLevel)
+ {
+ sumx = offX;
+ sumy = offY;
+ countMin = 1;
+ }
+ }
+
+ velx(i, j) = static_cast<float>(sumx) / countMin;
+ vely(i, j) = static_cast<float>(sumy) / countMin;
+ }
+
+ void calc(PtrStepSzb prev, PtrStepSzb curr, PtrStepSzf velx, PtrStepSzf vely, int2 blockSize, int2 shiftSize, bool usePrevious,
+ int maxX, int maxY, int acceptLevel, int escapeLevel, const short2* ss, int ssCount, cudaStream_t stream)
+ {
+ bindTexture(&tex_prev, prev);
+ bindTexture(&tex_curr, curr);
+
+ const dim3 block(32, 8);
+ const dim3 grid(divUp(velx.cols, block.x), divUp(vely.rows, block.y));
+
+ calcOptFlowBM<<<grid, block, 0, stream>>>(velx, vely, blockSize, shiftSize, usePrevious,
+ maxX, maxY, acceptLevel, escapeLevel, ss, ssCount);
+ cudaSafeCall( cudaGetLastError() );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+}
+
+#endif // !defined CUDA_DISABLER
using namespace cv::gpu;
using namespace cv::gpu::cudev;
-namespace optflowbm
-{
- texture<uchar, cudaTextureType2D, cudaReadModeElementType> tex_prev(false, cudaFilterModePoint, cudaAddressModeClamp);
- texture<uchar, cudaTextureType2D, cudaReadModeElementType> tex_curr(false, cudaFilterModePoint, cudaAddressModeClamp);
-
- __device__ int cmpBlocks(int X1, int Y1, int X2, int Y2, int2 blockSize)
- {
- int s = 0;
-
- for (int y = 0; y < blockSize.y; ++y)
- {
- for (int x = 0; x < blockSize.x; ++x)
- s += ::abs(tex2D(tex_prev, X1 + x, Y1 + y) - tex2D(tex_curr, X2 + x, Y2 + y));
- }
-
- return s;
- }
-
- __global__ void calcOptFlowBM(PtrStepSzf velx, PtrStepf vely, const int2 blockSize, const int2 shiftSize, const bool usePrevious,
- const int maxX, const int maxY, const int acceptLevel, const int escapeLevel,
- const short2* ss, const int ssCount)
- {
- const int j = blockIdx.x * blockDim.x + threadIdx.x;
- const int i = blockIdx.y * blockDim.y + threadIdx.y;
-
- if (i >= velx.rows || j >= velx.cols)
- return;
-
- const int X1 = j * shiftSize.x;
- const int Y1 = i * shiftSize.y;
-
- const int offX = usePrevious ? __float2int_rn(velx(i, j)) : 0;
- const int offY = usePrevious ? __float2int_rn(vely(i, j)) : 0;
-
- int X2 = X1 + offX;
- int Y2 = Y1 + offY;
-
- int dist = numeric_limits<int>::max();
-
- if (0 <= X2 && X2 <= maxX && 0 <= Y2 && Y2 <= maxY)
- dist = cmpBlocks(X1, Y1, X2, Y2, blockSize);
-
- int countMin = 1;
- int sumx = offX;
- int sumy = offY;
-
- if (dist > acceptLevel)
- {
- // do brute-force search
- for (int k = 0; k < ssCount; ++k)
- {
- const short2 ssVal = ss[k];
-
- const int dx = offX + ssVal.x;
- const int dy = offY + ssVal.y;
-
- X2 = X1 + dx;
- Y2 = Y1 + dy;
-
- if (0 <= X2 && X2 <= maxX && 0 <= Y2 && Y2 <= maxY)
- {
- const int tmpDist = cmpBlocks(X1, Y1, X2, Y2, blockSize);
- if (tmpDist < acceptLevel)
- {
- sumx = dx;
- sumy = dy;
- countMin = 1;
- break;
- }
-
- if (tmpDist < dist)
- {
- dist = tmpDist;
- sumx = dx;
- sumy = dy;
- countMin = 1;
- }
- else if (tmpDist == dist)
- {
- sumx += dx;
- sumy += dy;
- countMin++;
- }
- }
- }
-
- if (dist > escapeLevel)
- {
- sumx = offX;
- sumy = offY;
- countMin = 1;
- }
- }
-
- velx(i, j) = static_cast<float>(sumx) / countMin;
- vely(i, j) = static_cast<float>(sumy) / countMin;
- }
-
- void calc(PtrStepSzb prev, PtrStepSzb curr, PtrStepSzf velx, PtrStepSzf vely, int2 blockSize, int2 shiftSize, bool usePrevious,
- int maxX, int maxY, int acceptLevel, int escapeLevel, const short2* ss, int ssCount, cudaStream_t stream)
- {
- bindTexture(&tex_prev, prev);
- bindTexture(&tex_curr, curr);
-
- const dim3 block(32, 8);
- const dim3 grid(divUp(velx.cols, block.x), divUp(vely.rows, block.y));
-
- calcOptFlowBM<<<grid, block, 0, stream>>>(velx, vely, blockSize, shiftSize, usePrevious,
- maxX, maxY, acceptLevel, escapeLevel, ss, ssCount);
- cudaSafeCall( cudaGetLastError() );
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
-}
-
-/////////////////////////////////////////////////////////
-// Fast approximate version
-
namespace optflowbm_fast
{
enum
static const caller_t callers[] =
{
- gaussianBlurCaller<BrdReflect101<float> >,
+ 0 /*gaussianBlurCaller<BrdConstant<float> >*/,
gaussianBlurCaller<BrdReplicate<float> >,
+ 0 /*gaussianBlurCaller<BrdReflect<float> >*/,
+ 0 /*gaussianBlurCaller<BrdWrap<float> >*/,
+ gaussianBlurCaller<BrdReflect101<float> >
};
callers[borderMode](src, ksizeHalf, dst, stream);
static const caller_t callers[] =
{
- gaussianBlur5Caller<BrdReflect101<float>,256>,
+ 0 /*gaussianBlur5Caller<BrdConstant<float>,256>*/,
gaussianBlur5Caller<BrdReplicate<float>,256>,
+ 0 /*gaussianBlur5Caller<BrdReflect<float>,256>*/,
+ 0 /*gaussianBlur5Caller<BrdWrap<float>,256>*/,
+ gaussianBlur5Caller<BrdReflect101<float>,256>
};
callers[borderMode](src, ksizeHalf, dst, stream);
static const caller_t callers[] =
{
- gaussianBlur5Caller<BrdReflect101<float>,128>,
+ 0 /*gaussianBlur5Caller<BrdConstant<float>,128>*/,
gaussianBlur5Caller<BrdReplicate<float>,128>,
+ 0 /*gaussianBlur5Caller<BrdReflect<float>,128>*/,
+ 0 /*gaussianBlur5Caller<BrdWrap<float>,128>*/,
+ gaussianBlur5Caller<BrdReflect101<float>,128>
};
callers[borderMode](src, ksizeHalf, dst, stream);
{
if (deviceSupports(FEATURE_SET_COMPUTE_12))
cudev::optflow_farneback::gaussianBlur5Gpu(
- M, blockSize/2, bufM, BORDER_REPLICATE_GPU, S(streams[0]));
+ M, blockSize/2, bufM, BORDER_REPLICATE, S(streams[0]));
else
cudev::optflow_farneback::gaussianBlur5Gpu_CC11(
- M, blockSize/2, bufM, BORDER_REPLICATE_GPU, S(streams[0]));
+ M, blockSize/2, bufM, BORDER_REPLICATE, S(streams[0]));
swap(M, bufM);
cudev::optflow_farneback::updateFlowGpu(M, flowx, flowy, S(streams[0]));
pyramid1_[0] = frames_[1];
for (int i = 1; i <= numLevelsCropped; ++i)
{
- pyrDown(pyramid0_[i - 1], pyramid0_[i], streams[0]);
- pyrDown(pyramid1_[i - 1], pyramid1_[i], streams[1]);
+ gpu::pyrDown(pyramid0_[i - 1], pyramid0_[i], streams[0]);
+ gpu::pyrDown(pyramid1_[i - 1], pyramid1_[i], streams[1]);
}
}
{
if (flags & OPTFLOW_USE_INITIAL_FLOW)
{
-#if ENABLE_GPU_RESIZE
- resize(flowx0, curFlowX, Size(width, height), 0, 0, INTER_LINEAR, streams[0]);
- resize(flowy0, curFlowY, Size(width, height), 0, 0, INTER_LINEAR, streams[1]);
+ gpu::resize(flowx0, curFlowX, Size(width, height), 0, 0, INTER_LINEAR, streams[0]);
+ gpu::resize(flowy0, curFlowY, Size(width, height), 0, 0, INTER_LINEAR, streams[1]);
streams[0].enqueueConvert(curFlowX, curFlowX, curFlowX.depth(), scale);
streams[1].enqueueConvert(curFlowY, curFlowY, curFlowY.depth(), scale);
-#else
- Mat tmp1, tmp2;
- flowx0.download(tmp1);
- resize(tmp1, tmp2, Size(width, height), 0, 0, INTER_AREA);
- tmp2 *= scale;
- curFlowX.upload(tmp2);
- flowy0.download(tmp1);
- resize(tmp1, tmp2, Size(width, height), 0, 0, INTER_AREA);
- tmp2 *= scale;
- curFlowY.upload(tmp2);
-#endif
}
else
{
}
else
{
-#if ENABLE_GPU_RESIZE
- resize(prevFlowX, curFlowX, Size(width, height), 0, 0, INTER_LINEAR, streams[0]);
- resize(prevFlowY, curFlowY, Size(width, height), 0, 0, INTER_LINEAR, streams[1]);
+ gpu::resize(prevFlowX, curFlowX, Size(width, height), 0, 0, INTER_LINEAR, streams[0]);
+ gpu::resize(prevFlowY, curFlowY, Size(width, height), 0, 0, INTER_LINEAR, streams[1]);
streams[0].enqueueConvert(curFlowX, curFlowX, curFlowX.depth(), 1./pyrScale);
streams[1].enqueueConvert(curFlowY, curFlowY, curFlowY.depth(), 1./pyrScale);
-#else
- Mat tmp1, tmp2;
- prevFlowX.download(tmp1);
- resize(tmp1, tmp2, Size(width, height), 0, 0, INTER_LINEAR);
- tmp2 *= 1./pyrScale;
- curFlowX.upload(tmp2);
- prevFlowY.download(tmp1);
- resize(tmp1, tmp2, Size(width, height), 0, 0, INTER_LINEAR);
- tmp2 *= 1./pyrScale;
- curFlowY.upload(tmp2);
-#endif
}
GpuMat M = allocMatFromBuf(5*height, width, CV_32F, M_);
for (int i = 0; i < 2; i++)
{
cudev::optflow_farneback::gaussianBlurGpu(
- frames_[i], smoothSize/2, blurredFrame[i], BORDER_REFLECT101_GPU, S(streams[i]));
-#if ENABLE_GPU_RESIZE
- resize(blurredFrame[i], pyrLevel[i], Size(width, height), INTER_LINEAR, streams[i]);
-#else
- Mat tmp1, tmp2;
- tmp[i].download(tmp1);
- resize(tmp1, tmp2, Size(width, height), INTER_LINEAR);
- I[i].upload(tmp2);
-#endif
+ frames_[i], smoothSize/2, blurredFrame[i], BORDER_REFLECT101, S(streams[i]));
+ gpu::resize(blurredFrame[i], pyrLevel[i], Size(width, height), INTER_LINEAR, streams[i]);
cudev::optflow_farneback::polynomialExpansionGpu(pyrLevel[i], polyN, R[i], S(streams[i]));
}
}
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "precomp.hpp"
+
+using namespace cv;
+using namespace cv::gpu;
+
+#if !defined (HAVE_CUDA) || !defined (HAVE_OPENCV_GPULEGACY) || defined (CUDA_DISABLER)
+
+void cv::gpu::interpolateFrames(const GpuMat&, const GpuMat&, const GpuMat&, const GpuMat&, const GpuMat&, const GpuMat&, float, GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+
+#else
+
+void cv::gpu::interpolateFrames(const GpuMat& frame0, const GpuMat& frame1, const GpuMat& fu, const GpuMat& fv, const GpuMat& bu, const GpuMat& bv,
+ float pos, GpuMat& newFrame, GpuMat& buf, Stream& s)
+{
+ CV_Assert(frame0.type() == CV_32FC1);
+ CV_Assert(frame1.size() == frame0.size() && frame1.type() == frame0.type());
+ CV_Assert(fu.size() == frame0.size() && fu.type() == frame0.type());
+ CV_Assert(fv.size() == frame0.size() && fv.type() == frame0.type());
+ CV_Assert(bu.size() == frame0.size() && bu.type() == frame0.type());
+ CV_Assert(bv.size() == frame0.size() && bv.type() == frame0.type());
+
+ newFrame.create(frame0.size(), frame0.type());
+
+ buf.create(6 * frame0.rows, frame0.cols, CV_32FC1);
+ buf.setTo(Scalar::all(0));
+
+ // occlusion masks
+ GpuMat occ0 = buf.rowRange(0 * frame0.rows, 1 * frame0.rows);
+ GpuMat occ1 = buf.rowRange(1 * frame0.rows, 2 * frame0.rows);
+
+ // interpolated forward flow
+ GpuMat fui = buf.rowRange(2 * frame0.rows, 3 * frame0.rows);
+ GpuMat fvi = buf.rowRange(3 * frame0.rows, 4 * frame0.rows);
+
+ // interpolated backward flow
+ GpuMat bui = buf.rowRange(4 * frame0.rows, 5 * frame0.rows);
+ GpuMat bvi = buf.rowRange(5 * frame0.rows, 6 * frame0.rows);
+
+ size_t step = frame0.step;
+
+ CV_Assert(frame1.step == step && fu.step == step && fv.step == step && bu.step == step && bv.step == step && newFrame.step == step && buf.step == step);
+
+ cudaStream_t stream = StreamAccessor::getStream(s);
+ NppStStreamHandler h(stream);
+
+ NppStInterpolationState state;
+
+ state.size = NcvSize32u(frame0.cols, frame0.rows);
+ state.nStep = static_cast<Ncv32u>(step);
+ state.pSrcFrame0 = const_cast<Ncv32f*>(frame0.ptr<Ncv32f>());
+ state.pSrcFrame1 = const_cast<Ncv32f*>(frame1.ptr<Ncv32f>());
+ state.pFU = const_cast<Ncv32f*>(fu.ptr<Ncv32f>());
+ state.pFV = const_cast<Ncv32f*>(fv.ptr<Ncv32f>());
+ state.pBU = const_cast<Ncv32f*>(bu.ptr<Ncv32f>());
+ state.pBV = const_cast<Ncv32f*>(bv.ptr<Ncv32f>());
+ state.pos = pos;
+ state.pNewFrame = newFrame.ptr<Ncv32f>();
+ state.ppBuffers[0] = occ0.ptr<Ncv32f>();
+ state.ppBuffers[1] = occ1.ptr<Ncv32f>();
+ state.ppBuffers[2] = fui.ptr<Ncv32f>();
+ state.ppBuffers[3] = fvi.ptr<Ncv32f>();
+ state.ppBuffers[4] = bui.ptr<Ncv32f>();
+ state.ppBuffers[5] = bvi.ptr<Ncv32f>();
+
+ ncvSafeCall( nppiStInterpolateFrames(&state) );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+}
+
+#endif /* HAVE_CUDA */
//
//M*/
-#ifndef __THREAD_WRAPPERS_H__
-#define __THREAD_WRAPPERS_H__
-
#include "precomp.hpp"
-#if defined(HAVE_CUDA) && defined(HAVE_NVCUVID)
+using namespace cv;
+using namespace cv::gpu;
+
+#if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
+
+void cv::gpu::createOpticalFlowNeedleMap(const GpuMat&, const GpuMat&, GpuMat&, GpuMat&) { throw_no_cuda(); }
-namespace cv { namespace gpu
+#else
+
+namespace cv { namespace gpu { namespace cudev
{
- namespace detail
+ namespace optical_flow
{
- class CriticalSection
- {
- public:
- CriticalSection();
- ~CriticalSection();
-
- void enter();
- void leave();
-
- class AutoLock
- {
- public:
- explicit AutoLock(CriticalSection& criticalSection) :
- criticalSection_(criticalSection)
- {
- criticalSection_.enter();
- }
-
- ~AutoLock()
- {
- criticalSection_.leave();
- }
-
- private:
- CriticalSection& criticalSection_;
- };
-
- private:
- CriticalSection(const CriticalSection&);
- CriticalSection& operator=(const CriticalSection&);
-
- class Impl;
- std::auto_ptr<Impl> impl_;
- };
-
- class Thread
- {
- public:
- explicit Thread(void (*func)(void* userData), void* userData = 0);
- ~Thread();
-
- void wait();
-
- static void sleep(int ms);
-
- private:
- Thread(const Thread&);
- Thread& operator=(const Thread&);
-
- class Impl;
- std::auto_ptr<Impl> impl_;
- };
-
+ void NeedleMapAverage_gpu(PtrStepSzf u, PtrStepSzf v, PtrStepSzf u_avg, PtrStepSzf v_avg);
+ void CreateOpticalFlowNeedleMap_gpu(PtrStepSzf u_avg, PtrStepSzf v_avg, float* vertex_buffer, float* color_data, float max_flow, float xscale, float yscale);
}
-}}
+}}}
+
+void cv::gpu::createOpticalFlowNeedleMap(const GpuMat& u, const GpuMat& v, GpuMat& vertex, GpuMat& colors)
+{
+ using namespace cv::gpu::cudev::optical_flow;
+
+ CV_Assert(u.type() == CV_32FC1);
+ CV_Assert(v.type() == u.type() && v.size() == u.size());
+
+ const int NEEDLE_MAP_SCALE = 16;
+
+ const int x_needles = u.cols / NEEDLE_MAP_SCALE;
+ const int y_needles = u.rows / NEEDLE_MAP_SCALE;
+
+ GpuMat u_avg(y_needles, x_needles, CV_32FC1);
+ GpuMat v_avg(y_needles, x_needles, CV_32FC1);
+
+ NeedleMapAverage_gpu(u, v, u_avg, v_avg);
+
+ const int NUM_VERTS_PER_ARROW = 6;
+
+ const int num_arrows = x_needles * y_needles * NUM_VERTS_PER_ARROW;
+
+ vertex.create(1, num_arrows, CV_32FC3);
+ colors.create(1, num_arrows, CV_32FC3);
+
+ colors.setTo(Scalar::all(1.0));
+
+ double uMax, vMax;
+ gpu::minMax(u_avg, 0, &uMax);
+ gpu::minMax(v_avg, 0, &vMax);
+
+ float max_flow = static_cast<float>(std::sqrt(uMax * uMax + vMax * vMax));
+
+ CreateOpticalFlowNeedleMap_gpu(u_avg, v_avg, vertex.ptr<float>(), colors.ptr<float>(), max_flow, 1.0f / u.cols, 1.0f / u.rows);
-#endif // HAVE_CUDA
+ cvtColor(colors, colors, COLOR_HSV2RGB);
+}
-#endif // __THREAD_WRAPPERS_H__
+#endif /* HAVE_CUDA */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_PRECOMP_H__
+#define __OPENCV_PRECOMP_H__
+
+#include <limits>
+
+#include "opencv2/gpuoptflow.hpp"
+#include "opencv2/gpuarithm.hpp"
+#include "opencv2/gpuwarping.hpp"
+#include "opencv2/gpuimgproc.hpp"
+#include "opencv2/video.hpp"
+
+#include "opencv2/core/gpu_private.hpp"
+
+#include "opencv2/opencv_modules.hpp"
+
+#ifdef HAVE_OPENCV_GPULEGACY
+# include "opencv2/gpulegacy/private.hpp"
+#endif
+
+#endif /* __OPENCV_PRECOMP_H__ */
GpuMat temp1 = (useInitialFlow ? nextPts : prevPts).reshape(1);
GpuMat temp2 = nextPts.reshape(1);
- multiply(temp1, Scalar::all(1.0 / (1 << maxLevel) / 2.0), temp2);
+ gpu::multiply(temp1, Scalar::all(1.0 / (1 << maxLevel) / 2.0), temp2);
ensureSizeIsEnough(1, prevPts.cols, CV_8UC1, status);
status.setTo(Scalar::all(1));
}
else
{
- cvtColor(prevImg, buf_, COLOR_BGR2BGRA);
+ gpu::cvtColor(prevImg, buf_, COLOR_BGR2BGRA);
buf_.convertTo(prevPyr_[0], CV_32F);
- cvtColor(nextImg, buf_, COLOR_BGR2BGRA);
+ gpu::cvtColor(nextImg, buf_, COLOR_BGR2BGRA);
buf_.convertTo(nextPyr_[0], CV_32F);
}
for (int level = 1; level <= maxLevel; ++level)
{
- pyrDown(prevPyr_[level - 1], prevPyr_[level]);
- pyrDown(nextPyr_[level - 1], nextPyr_[level]);
+ gpu::pyrDown(prevPyr_[level - 1], prevPyr_[level]);
+ gpu::pyrDown(nextPyr_[level - 1], nextPyr_[level]);
}
pyrlk::loadConstants(make_int2(winSize.width, winSize.height), iters);
for (int level = 1; level <= maxLevel; ++level)
{
- pyrDown(prevPyr_[level - 1], prevPyr_[level]);
- pyrDown(nextPyr_[level - 1], nextPyr_[level]);
+ gpu::pyrDown(prevPyr_[level - 1], prevPyr_[level]);
+ gpu::pyrDown(nextPyr_[level - 1], nextPyr_[level]);
}
ensureSizeIsEnough(prevImg.size(), CV_32FC1, uPyr_[0]);
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+CV_GPU_TEST_MAIN("gpu")
EXPECT_TRUE(cv::checkRange(h_v));
};
-INSTANTIATE_TEST_CASE_P(GPU_Video, BroxOpticalFlow, ALL_DEVICES);
-
-//////////////////////////////////////////////////////
-// GoodFeaturesToTrack
-
-namespace
-{
- IMPLEMENT_PARAM_CLASS(MinDistance, double)
-}
-
-PARAM_TEST_CASE(GoodFeaturesToTrack, cv::gpu::DeviceInfo, MinDistance)
-{
- cv::gpu::DeviceInfo devInfo;
- double minDistance;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- minDistance = GET_PARAM(1);
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(GoodFeaturesToTrack, Accuracy)
-{
- cv::Mat image = readImage("opticalflow/frame0.png", cv::IMREAD_GRAYSCALE);
- ASSERT_FALSE(image.empty());
-
- int maxCorners = 1000;
- double qualityLevel = 0.01;
-
- cv::gpu::GoodFeaturesToTrackDetector_GPU detector(maxCorners, qualityLevel, minDistance);
-
- cv::gpu::GpuMat d_pts;
- detector(loadMat(image), d_pts);
-
- ASSERT_FALSE(d_pts.empty());
-
- std::vector<cv::Point2f> pts(d_pts.cols);
- cv::Mat pts_mat(1, d_pts.cols, CV_32FC2, (void*) &pts[0]);
- d_pts.download(pts_mat);
-
- std::vector<cv::Point2f> pts_gold;
- cv::goodFeaturesToTrack(image, pts_gold, maxCorners, qualityLevel, minDistance);
-
- ASSERT_EQ(pts_gold.size(), pts.size());
-
- size_t mistmatch = 0;
- for (size_t i = 0; i < pts.size(); ++i)
- {
- cv::Point2i a = pts_gold[i];
- cv::Point2i b = pts[i];
-
- bool eq = std::abs(a.x - b.x) < 1 && std::abs(a.y - b.y) < 1;
-
- if (!eq)
- ++mistmatch;
- }
-
- double bad_ratio = static_cast<double>(mistmatch) / pts.size();
-
- ASSERT_LE(bad_ratio, 0.01);
-}
-
-GPU_TEST_P(GoodFeaturesToTrack, EmptyCorners)
-{
- int maxCorners = 1000;
- double qualityLevel = 0.01;
-
- cv::gpu::GoodFeaturesToTrackDetector_GPU detector(maxCorners, qualityLevel, minDistance);
-
- cv::gpu::GpuMat src(100, 100, CV_8UC1, cv::Scalar::all(0));
- cv::gpu::GpuMat corners(1, maxCorners, CV_32FC2);
-
- detector(src, corners);
-
- ASSERT_TRUE(corners.empty());
-}
-
-INSTANTIATE_TEST_CASE_P(GPU_Video, GoodFeaturesToTrack, testing::Combine(
- ALL_DEVICES,
- testing::Values(MinDistance(0.0), MinDistance(3.0))));
+INSTANTIATE_TEST_CASE_P(GPU_OptFlow, BroxOpticalFlow, ALL_DEVICES);
//////////////////////////////////////////////////////
// PyrLKOpticalFlow
ASSERT_LE(bad_ratio, 0.01);
}
-INSTANTIATE_TEST_CASE_P(GPU_Video, PyrLKOpticalFlow, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_OptFlow, PyrLKOpticalFlow, testing::Combine(
ALL_DEVICES,
testing::Values(UseGray(true), UseGray(false))));
EXPECT_MAT_SIMILAR(flowxy[1], d_flowy, 0.1);
}
-INSTANTIATE_TEST_CASE_P(GPU_Video, FarnebackOpticalFlow, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_OptFlow, FarnebackOpticalFlow, testing::Combine(
ALL_DEVICES,
testing::Values(PyrScale(0.3), PyrScale(0.5), PyrScale(0.8)),
testing::Values(PolyN(5), PolyN(7)),
EXPECT_MAT_SIMILAR(gold[1], d_flowy, 4e-3);
}
-INSTANTIATE_TEST_CASE_P(GPU_Video, OpticalFlowDual_TVL1, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_OptFlow, OpticalFlowDual_TVL1, testing::Combine(
ALL_DEVICES,
WHOLE_SUBMAT));
EXPECT_MAT_NEAR(vely, d_vely, 0);
}
-INSTANTIATE_TEST_CASE_P(GPU_Video, OpticalFlowBM, ALL_DEVICES);
+INSTANTIATE_TEST_CASE_P(GPU_OptFlow, OpticalFlowBM, ALL_DEVICES);
//////////////////////////////////////////////////////
// FastOpticalFlowBM
EXPECT_LE(err, MAX_RMSE);
}
-INSTANTIATE_TEST_CASE_P(GPU_Video, FastOpticalFlowBM, ALL_DEVICES);
+INSTANTIATE_TEST_CASE_P(GPU_OptFlow, FastOpticalFlowBM, ALL_DEVICES);
#endif // HAVE_CUDA
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifdef __GNUC__
+# pragma GCC diagnostic ignored "-Wmissing-declarations"
+# if defined __clang__ || defined __APPLE__
+# pragma GCC diagnostic ignored "-Wmissing-prototypes"
+# pragma GCC diagnostic ignored "-Wextra"
+# endif
+#endif
+
+#ifndef __OPENCV_TEST_PRECOMP_HPP__
+#define __OPENCV_TEST_PRECOMP_HPP__
+
+#include <fstream>
+
+#include "opencv2/ts.hpp"
+#include "opencv2/ts/gpu_test.hpp"
+
+#include "opencv2/gpuoptflow.hpp"
+#include "opencv2/video.hpp"
+
+#endif
--- /dev/null
+if(ANDROID OR IOS)
+ ocv_module_disable(gpustereo)
+endif()
+
+set(the_description "GPU-accelerated Stereo Correspondence")
+
+ocv_warnings_disable(CMAKE_CXX_FLAGS /wd4127 /wd4324 /wd4512 -Wundef -Wmissing-declarations)
+
+ocv_define_module(gpustereo opencv_calib3d)
--- /dev/null
+************************************************
+gpustereo. GPU-accelerated Stereo Correspondence
+************************************************
+
+.. toctree::
+ :maxdepth: 1
+
+ stereo
-Camera Calibration and 3D Reconstruction
-========================================
+Stereo Correspondence
+=====================
.. highlight:: cpp
-gpu::solvePnPRansac
--------------------
-Finds the object pose from 3D-2D point correspondences.
-
-.. ocv:function:: void gpu::solvePnPRansac(const Mat& object, const Mat& image, const Mat& camera_mat, const Mat& dist_coef, Mat& rvec, Mat& tvec, bool use_extrinsic_guess=false, int num_iters=100, float max_dist=8.0, int min_inlier_count=100, vector<int>* inliers=NULL)
-
- :param object: Single-row matrix of object points.
-
- :param image: Single-row matrix of image points.
-
- :param camera_mat: 3x3 matrix of intrinsic camera parameters.
-
- :param dist_coef: Distortion coefficients. See :ocv:func:`undistortPoints` for details.
-
- :param rvec: Output 3D rotation vector.
-
- :param tvec: Output 3D translation vector.
-
- :param use_extrinsic_guess: Flag to indicate that the function must use ``rvec`` and ``tvec`` as an initial transformation guess. It is not supported for now.
-
- :param num_iters: Maximum number of RANSAC iterations.
-
- :param max_dist: Euclidean distance threshold to detect whether point is inlier or not.
-
- :param min_inlier_count: Flag to indicate that the function must stop if greater or equal number of inliers is achieved. It is not supported for now.
-
- :param inliers: Output vector of inlier indices.
-
-.. seealso:: :ocv:func:`solvePnPRansac`
-
-
-
.. [Felzenszwalb2006] Pedro F. Felzenszwalb algorithm [Pedro F. Felzenszwalb and Daniel P. Huttenlocher. *Efficient belief propagation for early vision*. International Journal of Computer Vision, 70(1), October 2006
.. [Yang2010] Q. Yang, L. Wang, and N. Ahuja. *A constant-space belief propagation algorithm for stereo matching*. In CVPR, 2010.
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_GPUSTEREO_HPP__
+#define __OPENCV_GPUSTEREO_HPP__
+
+#ifndef __cplusplus
+# error gpustereo.hpp header must be compiled as C++
+#endif
+
+#include "opencv2/core/gpumat.hpp"
+
+namespace cv { namespace gpu {
+
+class CV_EXPORTS StereoBM_GPU
+{
+public:
+ enum { BASIC_PRESET = 0, PREFILTER_XSOBEL = 1 };
+
+ enum { DEFAULT_NDISP = 64, DEFAULT_WINSZ = 19 };
+
+ //! the default constructor
+ StereoBM_GPU();
+ //! the full constructor taking the camera-specific preset, number of disparities and the SAD window size. ndisparities must be multiple of 8.
+ StereoBM_GPU(int preset, int ndisparities = DEFAULT_NDISP, int winSize = DEFAULT_WINSZ);
+
+ //! the stereo correspondence operator. Finds the disparity for the specified rectified stereo pair
+ //! Output disparity has CV_8U type.
+ void operator()(const GpuMat& left, const GpuMat& right, GpuMat& disparity, Stream& stream = Stream::Null());
+
+ //! Some heuristics that tries to estmate
+ // if current GPU will be faster than CPU in this algorithm.
+ // It queries current active device.
+ static bool checkIfGpuCallReasonable();
+
+ int preset;
+ int ndisp;
+ int winSize;
+
+ // If avergeTexThreshold == 0 => post procesing is disabled
+ // If avergeTexThreshold != 0 then disparity is set 0 in each point (x,y) where for left image
+ // SumOfHorizontalGradiensInWindow(x, y, winSize) < (winSize * winSize) * avergeTexThreshold
+ // i.e. input left image is low textured.
+ float avergeTexThreshold;
+
+private:
+ GpuMat minSSD, leBuf, riBuf;
+};
+
+// "Efficient Belief Propagation for Early Vision"
+// P.Felzenszwalb
+class CV_EXPORTS StereoBeliefPropagation
+{
+public:
+ enum { DEFAULT_NDISP = 64 };
+ enum { DEFAULT_ITERS = 5 };
+ enum { DEFAULT_LEVELS = 5 };
+
+ static void estimateRecommendedParams(int width, int height, int& ndisp, int& iters, int& levels);
+
+ //! the default constructor
+ explicit StereoBeliefPropagation(int ndisp = DEFAULT_NDISP,
+ int iters = DEFAULT_ITERS,
+ int levels = DEFAULT_LEVELS,
+ int msg_type = CV_32F);
+
+ //! the full constructor taking the number of disparities, number of BP iterations on each level,
+ //! number of levels, truncation of data cost, data weight,
+ //! truncation of discontinuity cost and discontinuity single jump
+ //! DataTerm = data_weight * min(fabs(I2-I1), max_data_term)
+ //! DiscTerm = min(disc_single_jump * fabs(f1-f2), max_disc_term)
+ //! please see paper for more details
+ StereoBeliefPropagation(int ndisp, int iters, int levels,
+ float max_data_term, float data_weight,
+ float max_disc_term, float disc_single_jump,
+ int msg_type = CV_32F);
+
+ //! the stereo correspondence operator. Finds the disparity for the specified rectified stereo pair,
+ //! if disparity is empty output type will be CV_16S else output type will be disparity.type().
+ void operator()(const GpuMat& left, const GpuMat& right, GpuMat& disparity, Stream& stream = Stream::Null());
+
+
+ //! version for user specified data term
+ void operator()(const GpuMat& data, GpuMat& disparity, Stream& stream = Stream::Null());
+
+ int ndisp;
+
+ int iters;
+ int levels;
+
+ float max_data_term;
+ float data_weight;
+ float max_disc_term;
+ float disc_single_jump;
+
+ int msg_type;
+private:
+ GpuMat u, d, l, r, u2, d2, l2, r2;
+ std::vector<GpuMat> datas;
+ GpuMat out;
+};
+
+// "A Constant-Space Belief Propagation Algorithm for Stereo Matching"
+// Qingxiong Yang, Liang Wang, Narendra Ahuja
+// http://vision.ai.uiuc.edu/~qyang6/
+class CV_EXPORTS StereoConstantSpaceBP
+{
+public:
+ enum { DEFAULT_NDISP = 128 };
+ enum { DEFAULT_ITERS = 8 };
+ enum { DEFAULT_LEVELS = 4 };
+ enum { DEFAULT_NR_PLANE = 4 };
+
+ static void estimateRecommendedParams(int width, int height, int& ndisp, int& iters, int& levels, int& nr_plane);
+
+ //! the default constructor
+ explicit StereoConstantSpaceBP(int ndisp = DEFAULT_NDISP,
+ int iters = DEFAULT_ITERS,
+ int levels = DEFAULT_LEVELS,
+ int nr_plane = DEFAULT_NR_PLANE,
+ int msg_type = CV_32F);
+
+ //! the full constructor taking the number of disparities, number of BP iterations on each level,
+ //! number of levels, number of active disparity on the first level, truncation of data cost, data weight,
+ //! truncation of discontinuity cost, discontinuity single jump and minimum disparity threshold
+ StereoConstantSpaceBP(int ndisp, int iters, int levels, int nr_plane,
+ float max_data_term, float data_weight, float max_disc_term, float disc_single_jump,
+ int min_disp_th = 0,
+ int msg_type = CV_32F);
+
+ //! the stereo correspondence operator. Finds the disparity for the specified rectified stereo pair,
+ //! if disparity is empty output type will be CV_16S else output type will be disparity.type().
+ void operator()(const GpuMat& left, const GpuMat& right, GpuMat& disparity, Stream& stream = Stream::Null());
+
+ int ndisp;
+
+ int iters;
+ int levels;
+
+ int nr_plane;
+
+ float max_data_term;
+ float data_weight;
+ float max_disc_term;
+ float disc_single_jump;
+
+ int min_disp_th;
+
+ int msg_type;
+
+ bool use_local_init_data_cost;
+private:
+ GpuMat messages_buffers;
+
+ GpuMat temp;
+ GpuMat out;
+};
+
+// Disparity map refinement using joint bilateral filtering given a single color image.
+// Qingxiong Yang, Liang Wang, Narendra Ahuja
+// http://vision.ai.uiuc.edu/~qyang6/
+class CV_EXPORTS DisparityBilateralFilter
+{
+public:
+ enum { DEFAULT_NDISP = 64 };
+ enum { DEFAULT_RADIUS = 3 };
+ enum { DEFAULT_ITERS = 1 };
+
+ //! the default constructor
+ explicit DisparityBilateralFilter(int ndisp = DEFAULT_NDISP, int radius = DEFAULT_RADIUS, int iters = DEFAULT_ITERS);
+
+ //! the full constructor taking the number of disparities, filter radius,
+ //! number of iterations, truncation of data continuity, truncation of disparity continuity
+ //! and filter range sigma
+ DisparityBilateralFilter(int ndisp, int radius, int iters, float edge_threshold, float max_disc_threshold, float sigma_range);
+
+ //! the disparity map refinement operator. Refine disparity map using joint bilateral filtering given a single color image.
+ //! disparity must have CV_8U or CV_16S type, image must have CV_8UC1 or CV_8UC3 type.
+ void operator()(const GpuMat& disparity, const GpuMat& image, GpuMat& dst, Stream& stream = Stream::Null());
+
+private:
+ int ndisp;
+ int radius;
+ int iters;
+
+ float edge_threshold;
+ float max_disc_threshold;
+ float sigma_range;
+
+ GpuMat table_color;
+ GpuMat table_space;
+};
+
+//! Reprojects disparity image to 3D space.
+//! Supports CV_8U and CV_16S types of input disparity.
+//! The output is a 3- or 4-channel floating-point matrix.
+//! Each element of this matrix will contain the 3D coordinates of the point (x,y,z,1), computed from the disparity map.
+//! Q is the 4x4 perspective transformation matrix that can be obtained with cvStereoRectify.
+CV_EXPORTS void reprojectImageTo3D(const GpuMat& disp, GpuMat& xyzw, const Mat& Q, int dst_cn = 4, Stream& stream = Stream::Null());
+
+//! Does coloring of disparity image: [0..ndisp) -> [0..240, 1, 1] in HSV.
+//! Supported types of input disparity: CV_8U, CV_16S.
+//! Output disparity has CV_8UC4 type in BGRA format (alpha = 255).
+CV_EXPORTS void drawColorDisp(const GpuMat& src_disp, GpuMat& dst_disp, int ndisp, Stream& stream = Stream::Null());
+
+}} // namespace cv { namespace gpu {
+
+#endif /* __OPENCV_GPUSTEREO_HPP__ */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace perf;
+
+CV_PERF_TEST_MAIN(gpustereo, printCudaInfo())
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifdef __GNUC__
+# pragma GCC diagnostic ignored "-Wmissing-declarations"
+# if defined __clang__ || defined __APPLE__
+# pragma GCC diagnostic ignored "-Wmissing-prototypes"
+# pragma GCC diagnostic ignored "-Wextra"
+# endif
+#endif
+
+#ifndef __OPENCV_PERF_PRECOMP_HPP__
+#define __OPENCV_PERF_PRECOMP_HPP__
+
+#include "opencv2/ts.hpp"
+#include "opencv2/ts/gpu_perf.hpp"
+
+#include "opencv2/gpustereo.hpp"
+#include "opencv2/calib3d.hpp"
+
+#ifdef GTEST_CREATE_SHARED_LIBRARY
+#error no modules except ts should have GTEST_CREATE_SHARED_LIBRARY defined
+#endif
+
+#endif
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace std;
+using namespace testing;
+using namespace perf;
+
+//////////////////////////////////////////////////////////////////////
+// StereoBM
+
+typedef std::tr1::tuple<string, string> pair_string;
+DEF_PARAM_TEST_1(ImagePair, pair_string);
+
+PERF_TEST_P(ImagePair, StereoBM,
+ Values(pair_string("gpu/perf/aloe.png", "gpu/perf/aloeR.png")))
+{
+ declare.time(300.0);
+
+ const cv::Mat imgLeft = readImage(GET_PARAM(0), cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(imgLeft.empty());
+
+ const cv::Mat imgRight = readImage(GET_PARAM(1), cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(imgRight.empty());
+
+ const int preset = 0;
+ const int ndisp = 256;
+
+ if (PERF_RUN_GPU())
+ {
+ cv::gpu::StereoBM_GPU d_bm(preset, ndisp);
+
+ const cv::gpu::GpuMat d_imgLeft(imgLeft);
+ const cv::gpu::GpuMat d_imgRight(imgRight);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() d_bm(d_imgLeft, d_imgRight, dst);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ cv::Ptr<cv::StereoBM> bm = cv::createStereoBM(ndisp);
+
+ cv::Mat dst;
+
+ TEST_CYCLE() bm->compute(imgLeft, imgRight, dst);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// StereoBeliefPropagation
+
+PERF_TEST_P(ImagePair, StereoBeliefPropagation,
+ Values(pair_string("gpu/stereobp/aloe-L.png", "gpu/stereobp/aloe-R.png")))
+{
+ declare.time(300.0);
+
+ const cv::Mat imgLeft = readImage(GET_PARAM(0));
+ ASSERT_FALSE(imgLeft.empty());
+
+ const cv::Mat imgRight = readImage(GET_PARAM(1));
+ ASSERT_FALSE(imgRight.empty());
+
+ const int ndisp = 64;
+
+ if (PERF_RUN_GPU())
+ {
+ cv::gpu::StereoBeliefPropagation d_bp(ndisp);
+
+ const cv::gpu::GpuMat d_imgLeft(imgLeft);
+ const cv::gpu::GpuMat d_imgRight(imgRight);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() d_bp(d_imgLeft, d_imgRight, dst);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// StereoConstantSpaceBP
+
+PERF_TEST_P(ImagePair, StereoConstantSpaceBP,
+ Values(pair_string("gpu/stereobm/aloe-L.png", "gpu/stereobm/aloe-R.png")))
+{
+ declare.time(300.0);
+
+ const cv::Mat imgLeft = readImage(GET_PARAM(0), cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(imgLeft.empty());
+
+ const cv::Mat imgRight = readImage(GET_PARAM(1), cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(imgRight.empty());
+
+ const int ndisp = 128;
+
+ if (PERF_RUN_GPU())
+ {
+ cv::gpu::StereoConstantSpaceBP d_csbp(ndisp);
+
+ const cv::gpu::GpuMat d_imgLeft(imgLeft);
+ const cv::gpu::GpuMat d_imgRight(imgRight);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() d_csbp(d_imgLeft, d_imgRight, dst);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// DisparityBilateralFilter
+
+PERF_TEST_P(ImagePair, DisparityBilateralFilter,
+ Values(pair_string("gpu/stereobm/aloe-L.png", "gpu/stereobm/aloe-disp.png")))
+{
+ const cv::Mat img = readImage(GET_PARAM(0), cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(img.empty());
+
+ const cv::Mat disp = readImage(GET_PARAM(1), cv::IMREAD_GRAYSCALE);
+ ASSERT_FALSE(disp.empty());
+
+ const int ndisp = 128;
+
+ if (PERF_RUN_GPU())
+ {
+ cv::gpu::DisparityBilateralFilter d_filter(ndisp);
+
+ const cv::gpu::GpuMat d_img(img);
+ const cv::gpu::GpuMat d_disp(disp);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() d_filter(d_disp, d_img, dst);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// ReprojectImageTo3D
+
+PERF_TEST_P(Sz_Depth, ReprojectImageTo3D,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16S)))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+
+ cv::Mat src(size, depth);
+ declare.in(src, WARMUP_RNG);
+
+ cv::Mat Q(4, 4, CV_32FC1);
+ cv::randu(Q, 0.1, 1.0);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::reprojectImageTo3D(d_src, dst, Q);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::reprojectImageTo3D(src, dst, Q);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// DrawColorDisp
+
+PERF_TEST_P(Sz_Depth, DrawColorDisp,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16S)))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int type = GET_PARAM(1);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::drawColorDisp(d_src, dst, 255);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/transform.hpp"
+#include "opencv2/core/cuda/functional.hpp"
+#include "opencv2/core/cuda/reduce.hpp"
+
+namespace cv { namespace gpu { namespace cudev
+{
+ /////////////////////////////////// reprojectImageTo3D ///////////////////////////////////////////////
+
+ __constant__ float cq[16];
+
+ template <typename T, typename D>
+ __global__ void reprojectImageTo3D(const PtrStepSz<T> disp, PtrStep<D> xyz)
+ {
+ const int x = blockIdx.x * blockDim.x + threadIdx.x;
+ const int y = blockIdx.y * blockDim.y + threadIdx.y;
+
+ if (y >= disp.rows || x >= disp.cols)
+ return;
+
+ const float qx = x * cq[ 0] + y * cq[ 1] + cq[ 3];
+ const float qy = x * cq[ 4] + y * cq[ 5] + cq[ 7];
+ const float qz = x * cq[ 8] + y * cq[ 9] + cq[11];
+ const float qw = x * cq[12] + y * cq[13] + cq[15];
+
+ const T d = disp(y, x);
+
+ const float iW = 1.f / (qw + cq[14] * d);
+
+ D v = VecTraits<D>::all(1.0f);
+ v.x = (qx + cq[2] * d) * iW;
+ v.y = (qy + cq[6] * d) * iW;
+ v.z = (qz + cq[10] * d) * iW;
+
+ xyz(y, x) = v;
+ }
+
+ template <typename T, typename D>
+ void reprojectImageTo3D_gpu(const PtrStepSzb disp, PtrStepSzb xyz, const float* q, cudaStream_t stream)
+ {
+ dim3 block(32, 8);
+ dim3 grid(divUp(disp.cols, block.x), divUp(disp.rows, block.y));
+
+ cudaSafeCall( cudaMemcpyToSymbol(cq, q, 16 * sizeof(float)) );
+
+ reprojectImageTo3D<T, D><<<grid, block, 0, stream>>>((PtrStepSz<T>)disp, (PtrStepSz<D>)xyz);
+ cudaSafeCall( cudaGetLastError() );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+
+ template void reprojectImageTo3D_gpu<uchar, float3>(const PtrStepSzb disp, PtrStepSzb xyz, const float* q, cudaStream_t stream);
+ template void reprojectImageTo3D_gpu<uchar, float4>(const PtrStepSzb disp, PtrStepSzb xyz, const float* q, cudaStream_t stream);
+ template void reprojectImageTo3D_gpu<short, float3>(const PtrStepSzb disp, PtrStepSzb xyz, const float* q, cudaStream_t stream);
+ template void reprojectImageTo3D_gpu<short, float4>(const PtrStepSzb disp, PtrStepSzb xyz, const float* q, cudaStream_t stream);
+
+ /////////////////////////////////// drawColorDisp ///////////////////////////////////////////////
+
+ template <typename T>
+ __device__ unsigned int cvtPixel(T d, int ndisp, float S = 1, float V = 1)
+ {
+ unsigned int H = ((ndisp-d) * 240)/ndisp;
+
+ unsigned int hi = (H/60) % 6;
+ float f = H/60.f - H/60;
+ float p = V * (1 - S);
+ float q = V * (1 - f * S);
+ float t = V * (1 - (1 - f) * S);
+
+ float3 res;
+
+ if (hi == 0) //R = V, G = t, B = p
+ {
+ res.x = p;
+ res.y = t;
+ res.z = V;
+ }
+
+ if (hi == 1) // R = q, G = V, B = p
+ {
+ res.x = p;
+ res.y = V;
+ res.z = q;
+ }
+
+ if (hi == 2) // R = p, G = V, B = t
+ {
+ res.x = t;
+ res.y = V;
+ res.z = p;
+ }
+
+ if (hi == 3) // R = p, G = q, B = V
+ {
+ res.x = V;
+ res.y = q;
+ res.z = p;
+ }
+
+ if (hi == 4) // R = t, G = p, B = V
+ {
+ res.x = V;
+ res.y = p;
+ res.z = t;
+ }
+
+ if (hi == 5) // R = V, G = p, B = q
+ {
+ res.x = q;
+ res.y = p;
+ res.z = V;
+ }
+ const unsigned int b = (unsigned int)(::max(0.f, ::min(res.x, 1.f)) * 255.f);
+ const unsigned int g = (unsigned int)(::max(0.f, ::min(res.y, 1.f)) * 255.f);
+ const unsigned int r = (unsigned int)(::max(0.f, ::min(res.z, 1.f)) * 255.f);
+ const unsigned int a = 255U;
+
+ return (a << 24) + (r << 16) + (g << 8) + b;
+ }
+
+ __global__ void drawColorDisp(uchar* disp, size_t disp_step, uchar* out_image, size_t out_step, int width, int height, int ndisp)
+ {
+ const int x = (blockIdx.x * blockDim.x + threadIdx.x) << 2;
+ const int y = blockIdx.y * blockDim.y + threadIdx.y;
+
+ if(x < width && y < height)
+ {
+ uchar4 d4 = *(uchar4*)(disp + y * disp_step + x);
+
+ uint4 res;
+ res.x = cvtPixel(d4.x, ndisp);
+ res.y = cvtPixel(d4.y, ndisp);
+ res.z = cvtPixel(d4.z, ndisp);
+ res.w = cvtPixel(d4.w, ndisp);
+
+ uint4* line = (uint4*)(out_image + y * out_step);
+ line[x >> 2] = res;
+ }
+ }
+
+ __global__ void drawColorDisp(short* disp, size_t disp_step, uchar* out_image, size_t out_step, int width, int height, int ndisp)
+ {
+ const int x = (blockIdx.x * blockDim.x + threadIdx.x) << 1;
+ const int y = blockIdx.y * blockDim.y + threadIdx.y;
+
+ if(x < width && y < height)
+ {
+ short2 d2 = *(short2*)(disp + y * disp_step + x);
+
+ uint2 res;
+ res.x = cvtPixel(d2.x, ndisp);
+ res.y = cvtPixel(d2.y, ndisp);
+
+ uint2* line = (uint2*)(out_image + y * out_step);
+ line[x >> 1] = res;
+ }
+ }
+
+
+ void drawColorDisp_gpu(const PtrStepSzb& src, const PtrStepSzb& dst, int ndisp, const cudaStream_t& stream)
+ {
+ dim3 threads(16, 16, 1);
+ dim3 grid(1, 1, 1);
+ grid.x = divUp(src.cols, threads.x << 2);
+ grid.y = divUp(src.rows, threads.y);
+
+ drawColorDisp<<<grid, threads, 0, stream>>>(src.data, src.step, dst.data, dst.step, src.cols, src.rows, ndisp);
+ cudaSafeCall( cudaGetLastError() );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+
+ void drawColorDisp_gpu(const PtrStepSz<short>& src, const PtrStepSzb& dst, int ndisp, const cudaStream_t& stream)
+ {
+ dim3 threads(32, 8, 1);
+ dim3 grid(1, 1, 1);
+ grid.x = divUp(src.cols, threads.x << 1);
+ grid.y = divUp(src.rows, threads.y);
+
+ drawColorDisp<<<grid, threads, 0, stream>>>(src.data, src.step / sizeof(short), dst.data, dst.step, src.cols, src.rows, ndisp);
+ cudaSafeCall( cudaGetLastError() );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+}}} // namespace cv { namespace gpu { namespace cudev
+
+
+#endif /* CUDA_DISABLER */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_PRECOMP_H__
+#define __OPENCV_PRECOMP_H__
+
+#include <limits>
+
+#include "opencv2/gpustereo.hpp"
+
+#include "opencv2/core/gpu_private.hpp"
+
+#endif /* __OPENCV_PRECOMP_H__ */
for(int _r = 0; _r < 5; ++_r)
{
*buf_ptrs[_r] = sub2.rowRange(_r * sub2.rows/5, (_r+1) * sub2.rows/5);
- assert(buf_ptrs[_r]->cols == cols && buf_ptrs[_r]->rows == rows * rthis.nr_plane);
+ CV_DbgAssert(buf_ptrs[_r]->cols == cols && buf_ptrs[_r]->rows == rows * rthis.nr_plane);
}
};
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "precomp.hpp"
+
+using namespace cv;
+using namespace cv::gpu;
+
+#if !defined HAVE_CUDA || defined(CUDA_DISABLER)
+
+void cv::gpu::reprojectImageTo3D(const GpuMat&, GpuMat&, const Mat&, int, Stream&) { throw_no_cuda(); }
+void cv::gpu::drawColorDisp(const GpuMat&, GpuMat&, int, Stream&) { throw_no_cuda(); }
+
+#else
+
+////////////////////////////////////////////////////////////////////////
+// reprojectImageTo3D
+
+namespace cv { namespace gpu { namespace cudev
+{
+ template <typename T, typename D>
+ void reprojectImageTo3D_gpu(const PtrStepSzb disp, PtrStepSzb xyz, const float* q, cudaStream_t stream);
+}}}
+
+void cv::gpu::reprojectImageTo3D(const GpuMat& disp, GpuMat& xyz, const Mat& Q, int dst_cn, Stream& stream)
+{
+ using namespace cv::gpu::cudev;
+
+ typedef void (*func_t)(const PtrStepSzb disp, PtrStepSzb xyz, const float* q, cudaStream_t stream);
+ static const func_t funcs[2][4] =
+ {
+ {reprojectImageTo3D_gpu<uchar, float3>, 0, 0, reprojectImageTo3D_gpu<short, float3>},
+ {reprojectImageTo3D_gpu<uchar, float4>, 0, 0, reprojectImageTo3D_gpu<short, float4>}
+ };
+
+ CV_Assert(disp.type() == CV_8U || disp.type() == CV_16S);
+ CV_Assert(Q.type() == CV_32F && Q.rows == 4 && Q.cols == 4 && Q.isContinuous());
+ CV_Assert(dst_cn == 3 || dst_cn == 4);
+
+ xyz.create(disp.size(), CV_MAKE_TYPE(CV_32F, dst_cn));
+
+ funcs[dst_cn == 4][disp.type()](disp, xyz, Q.ptr<float>(), StreamAccessor::getStream(stream));
+}
+
+////////////////////////////////////////////////////////////////////////
+// drawColorDisp
+
+namespace cv { namespace gpu { namespace cudev
+{
+ void drawColorDisp_gpu(const PtrStepSzb& src, const PtrStepSzb& dst, int ndisp, const cudaStream_t& stream);
+ void drawColorDisp_gpu(const PtrStepSz<short>& src, const PtrStepSzb& dst, int ndisp, const cudaStream_t& stream);
+}}}
+
+namespace
+{
+ template <typename T>
+ void drawColorDisp_caller(const GpuMat& src, GpuMat& dst, int ndisp, const cudaStream_t& stream)
+ {
+ using namespace ::cv::gpu::cudev;
+
+ dst.create(src.size(), CV_8UC4);
+
+ drawColorDisp_gpu((PtrStepSz<T>)src, dst, ndisp, stream);
+ }
+
+ typedef void (*drawColorDisp_caller_t)(const GpuMat& src, GpuMat& dst, int ndisp, const cudaStream_t& stream);
+
+ const drawColorDisp_caller_t drawColorDisp_callers[] = {drawColorDisp_caller<unsigned char>, 0, 0, drawColorDisp_caller<short>, 0, 0, 0, 0};
+}
+
+void cv::gpu::drawColorDisp(const GpuMat& src, GpuMat& dst, int ndisp, Stream& stream)
+{
+ CV_Assert(src.type() == CV_8U || src.type() == CV_16S);
+
+ drawColorDisp_callers[src.type()](src, dst, ndisp, StreamAccessor::getStream(stream));
+}
+
+#endif
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+CV_GPU_TEST_MAIN("gpu")
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifdef __GNUC__
+# pragma GCC diagnostic ignored "-Wmissing-declarations"
+# if defined __clang__ || defined __APPLE__
+# pragma GCC diagnostic ignored "-Wmissing-prototypes"
+# pragma GCC diagnostic ignored "-Wextra"
+# endif
+#endif
+
+#ifndef __OPENCV_TEST_PRECOMP_HPP__
+#define __OPENCV_TEST_PRECOMP_HPP__
+
+#include "opencv2/ts.hpp"
+#include "opencv2/ts/gpu_test.hpp"
+
+#include "opencv2/gpustereo.hpp"
+#include "opencv2/calib3d.hpp"
+
+#endif
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+#ifdef HAVE_CUDA
+
+using namespace cvtest;
+
+//////////////////////////////////////////////////////////////////////////
+// StereoBM
+
+struct StereoBM : testing::TestWithParam<cv::gpu::DeviceInfo>
+{
+ cv::gpu::DeviceInfo devInfo;
+
+ virtual void SetUp()
+ {
+ devInfo = GetParam();
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(StereoBM, Regression)
+{
+ cv::Mat left_image = readImage("stereobm/aloe-L.png", cv::IMREAD_GRAYSCALE);
+ cv::Mat right_image = readImage("stereobm/aloe-R.png", cv::IMREAD_GRAYSCALE);
+ cv::Mat disp_gold = readImage("stereobm/aloe-disp.png", cv::IMREAD_GRAYSCALE);
+
+ ASSERT_FALSE(left_image.empty());
+ ASSERT_FALSE(right_image.empty());
+ ASSERT_FALSE(disp_gold.empty());
+
+ cv::gpu::StereoBM_GPU bm(0, 128, 19);
+ cv::gpu::GpuMat disp;
+
+ bm(loadMat(left_image), loadMat(right_image), disp);
+
+ EXPECT_MAT_NEAR(disp_gold, disp, 0.0);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Stereo, StereoBM, ALL_DEVICES);
+
+//////////////////////////////////////////////////////////////////////////
+// StereoBeliefPropagation
+
+struct StereoBeliefPropagation : testing::TestWithParam<cv::gpu::DeviceInfo>
+{
+ cv::gpu::DeviceInfo devInfo;
+
+ virtual void SetUp()
+ {
+ devInfo = GetParam();
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(StereoBeliefPropagation, Regression)
+{
+ cv::Mat left_image = readImage("stereobp/aloe-L.png");
+ cv::Mat right_image = readImage("stereobp/aloe-R.png");
+ cv::Mat disp_gold = readImage("stereobp/aloe-disp.png", cv::IMREAD_GRAYSCALE);
+
+ ASSERT_FALSE(left_image.empty());
+ ASSERT_FALSE(right_image.empty());
+ ASSERT_FALSE(disp_gold.empty());
+
+ cv::gpu::StereoBeliefPropagation bp(64, 8, 2, 25, 0.1f, 15, 1, CV_16S);
+ cv::gpu::GpuMat disp;
+
+ bp(loadMat(left_image), loadMat(right_image), disp);
+
+ cv::Mat h_disp(disp);
+ h_disp.convertTo(h_disp, disp_gold.depth());
+
+ EXPECT_MAT_NEAR(disp_gold, h_disp, 0.0);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Stereo, StereoBeliefPropagation, ALL_DEVICES);
+
+//////////////////////////////////////////////////////////////////////////
+// StereoConstantSpaceBP
+
+struct StereoConstantSpaceBP : testing::TestWithParam<cv::gpu::DeviceInfo>
+{
+ cv::gpu::DeviceInfo devInfo;
+
+ virtual void SetUp()
+ {
+ devInfo = GetParam();
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(StereoConstantSpaceBP, Regression)
+{
+ cv::Mat left_image = readImage("csstereobp/aloe-L.png");
+ cv::Mat right_image = readImage("csstereobp/aloe-R.png");
+
+ cv::Mat disp_gold;
+
+ if (supportFeature(devInfo, cv::gpu::FEATURE_SET_COMPUTE_20))
+ disp_gold = readImage("csstereobp/aloe-disp.png", cv::IMREAD_GRAYSCALE);
+ else
+ disp_gold = readImage("csstereobp/aloe-disp_CC1X.png", cv::IMREAD_GRAYSCALE);
+
+ ASSERT_FALSE(left_image.empty());
+ ASSERT_FALSE(right_image.empty());
+ ASSERT_FALSE(disp_gold.empty());
+
+ cv::gpu::StereoConstantSpaceBP csbp(128, 16, 4, 4);
+ cv::gpu::GpuMat disp;
+
+ csbp(loadMat(left_image), loadMat(right_image), disp);
+
+ cv::Mat h_disp(disp);
+ h_disp.convertTo(h_disp, disp_gold.depth());
+
+ EXPECT_MAT_NEAR(disp_gold, h_disp, 1.0);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Stereo, StereoConstantSpaceBP, ALL_DEVICES);
+
+////////////////////////////////////////////////////////////////////////////////
+// reprojectImageTo3D
+
+PARAM_TEST_CASE(ReprojectImageTo3D, cv::gpu::DeviceInfo, cv::Size, MatDepth, UseRoi)
+{
+ cv::gpu::DeviceInfo devInfo;
+ cv::Size size;
+ int depth;
+ bool useRoi;
+
+ virtual void SetUp()
+ {
+ devInfo = GET_PARAM(0);
+ size = GET_PARAM(1);
+ depth = GET_PARAM(2);
+ useRoi = GET_PARAM(3);
+
+ cv::gpu::setDevice(devInfo.deviceID());
+ }
+};
+
+GPU_TEST_P(ReprojectImageTo3D, Accuracy)
+{
+ cv::Mat disp = randomMat(size, depth, 5.0, 30.0);
+ cv::Mat Q = randomMat(cv::Size(4, 4), CV_32FC1, 0.1, 1.0);
+
+ cv::gpu::GpuMat dst;
+ cv::gpu::reprojectImageTo3D(loadMat(disp, useRoi), dst, Q, 3);
+
+ cv::Mat dst_gold;
+ cv::reprojectImageTo3D(disp, dst_gold, Q, false);
+
+ EXPECT_MAT_NEAR(dst_gold, dst, 1e-5);
+}
+
+INSTANTIATE_TEST_CASE_P(GPU_Stereo, ReprojectImageTo3D, testing::Combine(
+ ALL_DEVICES,
+ DIFFERENT_SIZES,
+ testing::Values(MatDepth(CV_8U), MatDepth(CV_16S)),
+ WHOLE_SUBMAT));
+
+#endif // HAVE_CUDA
--- /dev/null
+if(ANDROID OR IOS)
+ ocv_module_disable(gpuwarping)
+endif()
+
+set(the_description "GPU-accelerated Image Warping")
+
+ocv_warnings_disable(CMAKE_CXX_FLAGS /wd4127 /wd4324 /wd4512 -Wundef -Wmissing-declarations)
+
+ocv_define_module(gpuwarping opencv_imgproc OPTIONAL opencv_gpulegacy)
--- /dev/null
+*****************************************
+gpuwarping. GPU-accelerated Image Warping
+*****************************************
+
+.. toctree::
+ :maxdepth: 1
+
+ warping
--- /dev/null
+Image Warping
+=============
+
+.. highlight:: cpp
+
+
+
+gpu::remap
+--------------
+Applies a generic geometrical transformation to an image.
+
+.. ocv:function:: void gpu::remap( const GpuMat& src, GpuMat& dst, const GpuMat& xmap, const GpuMat& ymap, int interpolation, int borderMode=BORDER_CONSTANT, Scalar borderValue=Scalar(), Stream& stream=Stream::Null() )
+
+ :param src: Source image.
+
+ :param dst: Destination image with the size the same as ``xmap`` and the type the same as ``src`` .
+
+ :param xmap: X values. Only ``CV_32FC1`` type is supported.
+
+ :param ymap: Y values. Only ``CV_32FC1`` type is supported.
+
+ :param interpolation: Interpolation method (see :ocv:func:`resize` ). ``INTER_NEAREST`` , ``INTER_LINEAR`` and ``INTER_CUBIC`` are supported for now.
+
+ :param borderMode: Pixel extrapolation method (see :ocv:func:`borderInterpolate` ). ``BORDER_REFLECT101`` , ``BORDER_REPLICATE`` , ``BORDER_CONSTANT`` , ``BORDER_REFLECT`` and ``BORDER_WRAP`` are supported for now.
+
+ :param borderValue: Value used in case of a constant border. By default, it is 0.
+
+ :param stream: Stream for the asynchronous version.
+
+The function transforms the source image using the specified map:
+
+.. math::
+
+ \texttt{dst} (x,y) = \texttt{src} (xmap(x,y), ymap(x,y))
+
+Values of pixels with non-integer coordinates are computed using the bilinear interpolation.
+
+.. seealso:: :ocv:func:`remap`
+
+
+
+gpu::resize
+---------------
+Resizes an image.
+
+.. ocv:function:: void gpu::resize(const GpuMat& src, GpuMat& dst, Size dsize, double fx=0, double fy=0, int interpolation = INTER_LINEAR, Stream& stream = Stream::Null())
+
+ :param src: Source image.
+
+ :param dst: Destination image with the same type as ``src`` . The size is ``dsize`` (when it is non-zero) or the size is computed from ``src.size()`` , ``fx`` , and ``fy`` .
+
+ :param dsize: Destination image size. If it is zero, it is computed as:
+
+ .. math::
+ \texttt{dsize = Size(round(fx*src.cols), round(fy*src.rows))}
+
+ Either ``dsize`` or both ``fx`` and ``fy`` must be non-zero.
+
+ :param fx: Scale factor along the horizontal axis. If it is zero, it is computed as:
+
+ .. math::
+
+ \texttt{(double)dsize.width/src.cols}
+
+ :param fy: Scale factor along the vertical axis. If it is zero, it is computed as:
+
+ .. math::
+
+ \texttt{(double)dsize.height/src.rows}
+
+ :param interpolation: Interpolation method. ``INTER_NEAREST`` , ``INTER_LINEAR`` and ``INTER_CUBIC`` are supported for now.
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`resize`
+
+
+
+gpu::warpAffine
+-------------------
+Applies an affine transformation to an image.
+
+.. ocv:function:: void gpu::warpAffine( const GpuMat& src, GpuMat& dst, const Mat& M, Size dsize, int flags=INTER_LINEAR, int borderMode=BORDER_CONSTANT, Scalar borderValue=Scalar(), Stream& stream=Stream::Null() )
+
+ :param src: Source image. ``CV_8U`` , ``CV_16U`` , ``CV_32S`` , or ``CV_32F`` depth and 1, 3, or 4 channels are supported.
+
+ :param dst: Destination image with the same type as ``src`` . The size is ``dsize`` .
+
+ :param M: *2x3* transformation matrix.
+
+ :param dsize: Size of the destination image.
+
+ :param flags: Combination of interpolation methods (see :ocv:func:`resize`) and the optional flag ``WARP_INVERSE_MAP`` specifying that ``M`` is an inverse transformation ( ``dst=>src`` ). Only ``INTER_NEAREST`` , ``INTER_LINEAR`` , and ``INTER_CUBIC`` interpolation methods are supported.
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`warpAffine`
+
+
+
+gpu::buildWarpAffineMaps
+------------------------
+Builds transformation maps for affine transformation.
+
+.. ocv:function:: void gpu::buildWarpAffineMaps(const Mat& M, bool inverse, Size dsize, GpuMat& xmap, GpuMat& ymap, Stream& stream = Stream::Null())
+
+ :param M: *2x3* transformation matrix.
+
+ :param inverse: Flag specifying that ``M`` is an inverse transformation ( ``dst=>src`` ).
+
+ :param dsize: Size of the destination image.
+
+ :param xmap: X values with ``CV_32FC1`` type.
+
+ :param ymap: Y values with ``CV_32FC1`` type.
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`gpu::warpAffine` , :ocv:func:`gpu::remap`
+
+
+
+gpu::warpPerspective
+------------------------
+Applies a perspective transformation to an image.
+
+.. ocv:function:: void gpu::warpPerspective( const GpuMat& src, GpuMat& dst, const Mat& M, Size dsize, int flags=INTER_LINEAR, int borderMode=BORDER_CONSTANT, Scalar borderValue=Scalar(), Stream& stream=Stream::Null() )
+
+ :param src: Source image. ``CV_8U`` , ``CV_16U`` , ``CV_32S`` , or ``CV_32F`` depth and 1, 3, or 4 channels are supported.
+
+ :param dst: Destination image with the same type as ``src`` . The size is ``dsize`` .
+
+ :param M: *3x3* transformation matrix.
+
+ :param dsize: Size of the destination image.
+
+ :param flags: Combination of interpolation methods (see :ocv:func:`resize` ) and the optional flag ``WARP_INVERSE_MAP`` specifying that ``M`` is the inverse transformation ( ``dst => src`` ). Only ``INTER_NEAREST`` , ``INTER_LINEAR`` , and ``INTER_CUBIC`` interpolation methods are supported.
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`warpPerspective`
+
+
+
+gpu::buildWarpPerspectiveMaps
+-----------------------------
+Builds transformation maps for perspective transformation.
+
+.. ocv:function:: void gpu::buildWarpAffineMaps(const Mat& M, bool inverse, Size dsize, GpuMat& xmap, GpuMat& ymap, Stream& stream = Stream::Null())
+
+ :param M: *3x3* transformation matrix.
+
+ :param inverse: Flag specifying that ``M`` is an inverse transformation ( ``dst=>src`` ).
+
+ :param dsize: Size of the destination image.
+
+ :param xmap: X values with ``CV_32FC1`` type.
+
+ :param ymap: Y values with ``CV_32FC1`` type.
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`gpu::warpPerspective` , :ocv:func:`gpu::remap`
+
+
+
+gpu::rotate
+---------------
+Rotates an image around the origin (0,0) and then shifts it.
+
+.. ocv:function:: void gpu::rotate(const GpuMat& src, GpuMat& dst, Size dsize, double angle, double xShift = 0, double yShift = 0, int interpolation = INTER_LINEAR, Stream& stream = Stream::Null())
+
+ :param src: Source image. Supports 1, 3 or 4 channels images with ``CV_8U`` , ``CV_16U`` or ``CV_32F`` depth.
+
+ :param dst: Destination image with the same type as ``src`` . The size is ``dsize`` .
+
+ :param dsize: Size of the destination image.
+
+ :param angle: Angle of rotation in degrees.
+
+ :param xShift: Shift along the horizontal axis.
+
+ :param yShift: Shift along the vertical axis.
+
+ :param interpolation: Interpolation method. Only ``INTER_NEAREST`` , ``INTER_LINEAR`` , and ``INTER_CUBIC`` are supported.
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`gpu::warpAffine`
+
+
+
+gpu::buildWarpPlaneMaps
+-----------------------
+Builds plane warping maps.
+
+.. ocv:function:: void gpu::buildWarpPlaneMaps( Size src_size, Rect dst_roi, const Mat & K, const Mat& R, const Mat & T, float scale, GpuMat& map_x, GpuMat& map_y, Stream& stream=Stream::Null() )
+
+ :param stream: Stream for the asynchronous version.
+
+
+
+gpu::buildWarpCylindricalMaps
+-----------------------------
+Builds cylindrical warping maps.
+
+.. ocv:function:: void gpu::buildWarpCylindricalMaps( Size src_size, Rect dst_roi, const Mat & K, const Mat& R, float scale, GpuMat& map_x, GpuMat& map_y, Stream& stream=Stream::Null() )
+
+ :param stream: Stream for the asynchronous version.
+
+
+
+gpu::buildWarpSphericalMaps
+---------------------------
+Builds spherical warping maps.
+
+.. ocv:function:: void gpu::buildWarpSphericalMaps( Size src_size, Rect dst_roi, const Mat & K, const Mat& R, float scale, GpuMat& map_x, GpuMat& map_y, Stream& stream=Stream::Null() )
+
+ :param stream: Stream for the asynchronous version.
+
+
+
+gpu::pyrDown
+-------------------
+Smoothes an image and downsamples it.
+
+.. ocv:function:: void gpu::pyrDown(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null())
+
+ :param src: Source image.
+
+ :param dst: Destination image. Will have ``Size((src.cols+1)/2, (src.rows+1)/2)`` size and the same type as ``src`` .
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`pyrDown`
+
+
+
+gpu::pyrUp
+-------------------
+Upsamples an image and then smoothes it.
+
+.. ocv:function:: void gpu::pyrUp(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null())
+
+ :param src: Source image.
+
+ :param dst: Destination image. Will have ``Size(src.cols*2, src.rows*2)`` size and the same type as ``src`` .
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso:: :ocv:func:`pyrUp`
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_GPUWARPING_HPP__
+#define __OPENCV_GPUWARPING_HPP__
+
+#ifndef __cplusplus
+# error gpuwarping.hpp header must be compiled as C++
+#endif
+
+#include "opencv2/core/gpumat.hpp"
+#include "opencv2/imgproc.hpp"
+
+namespace cv { namespace gpu {
+
+//! DST[x,y] = SRC[xmap[x,y],ymap[x,y]]
+//! supports only CV_32FC1 map type
+CV_EXPORTS void remap(const GpuMat& src, GpuMat& dst, const GpuMat& xmap, const GpuMat& ymap,
+ int interpolation, int borderMode = BORDER_CONSTANT, Scalar borderValue = Scalar(),
+ Stream& stream = Stream::Null());
+
+//! resizes the image
+//! Supports INTER_NEAREST, INTER_LINEAR, INTER_CUBIC, INTER_AREA
+CV_EXPORTS void resize(const GpuMat& src, GpuMat& dst, Size dsize, double fx=0, double fy=0, int interpolation = INTER_LINEAR, Stream& stream = Stream::Null());
+
+//! warps the image using affine transformation
+//! Supports INTER_NEAREST, INTER_LINEAR, INTER_CUBIC
+CV_EXPORTS void warpAffine(const GpuMat& src, GpuMat& dst, const Mat& M, Size dsize, int flags = INTER_LINEAR,
+ int borderMode = BORDER_CONSTANT, Scalar borderValue = Scalar(), Stream& stream = Stream::Null());
+
+CV_EXPORTS void buildWarpAffineMaps(const Mat& M, bool inverse, Size dsize, GpuMat& xmap, GpuMat& ymap, Stream& stream = Stream::Null());
+
+//! warps the image using perspective transformation
+//! Supports INTER_NEAREST, INTER_LINEAR, INTER_CUBIC
+CV_EXPORTS void warpPerspective(const GpuMat& src, GpuMat& dst, const Mat& M, Size dsize, int flags = INTER_LINEAR,
+ int borderMode = BORDER_CONSTANT, Scalar borderValue = Scalar(), Stream& stream = Stream::Null());
+
+CV_EXPORTS void buildWarpPerspectiveMaps(const Mat& M, bool inverse, Size dsize, GpuMat& xmap, GpuMat& ymap, Stream& stream = Stream::Null());
+
+//! builds plane warping maps
+CV_EXPORTS void buildWarpPlaneMaps(Size src_size, Rect dst_roi, const Mat &K, const Mat& R, const Mat &T, float scale,
+ GpuMat& map_x, GpuMat& map_y, Stream& stream = Stream::Null());
+
+//! builds cylindrical warping maps
+CV_EXPORTS void buildWarpCylindricalMaps(Size src_size, Rect dst_roi, const Mat &K, const Mat& R, float scale,
+ GpuMat& map_x, GpuMat& map_y, Stream& stream = Stream::Null());
+
+//! builds spherical warping maps
+CV_EXPORTS void buildWarpSphericalMaps(Size src_size, Rect dst_roi, const Mat &K, const Mat& R, float scale,
+ GpuMat& map_x, GpuMat& map_y, Stream& stream = Stream::Null());
+
+//! rotates an image around the origin (0,0) and then shifts it
+//! supports INTER_NEAREST, INTER_LINEAR, INTER_CUBIC
+//! supports 1, 3 or 4 channels images with CV_8U, CV_16U or CV_32F depth
+CV_EXPORTS void rotate(const GpuMat& src, GpuMat& dst, Size dsize, double angle, double xShift = 0, double yShift = 0,
+ int interpolation = INTER_LINEAR, Stream& stream = Stream::Null());
+
+//! smoothes the source image and downsamples it
+CV_EXPORTS void pyrDown(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null());
+
+//! upsamples the source image and then smoothes it
+CV_EXPORTS void pyrUp(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null());
+
+class CV_EXPORTS ImagePyramid
+{
+public:
+ inline ImagePyramid() : nLayers_(0) {}
+ inline ImagePyramid(const GpuMat& img, int nLayers, Stream& stream = Stream::Null())
+ {
+ build(img, nLayers, stream);
+ }
+
+ void build(const GpuMat& img, int nLayers, Stream& stream = Stream::Null());
+
+ void getLayer(GpuMat& outImg, Size outRoi, Stream& stream = Stream::Null()) const;
+
+ inline void release()
+ {
+ layer0_.release();
+ pyramid_.clear();
+ nLayers_ = 0;
+ }
+
+private:
+ GpuMat layer0_;
+ std::vector<GpuMat> pyramid_;
+ int nLayers_;
+};
+
+}} // namespace cv { namespace gpu {
+
+#endif /* __OPENCV_GPUWARPING_HPP__ */
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace perf;
+
+CV_PERF_TEST_MAIN(gpuwarping, printCudaInfo())
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifdef __GNUC__
+# pragma GCC diagnostic ignored "-Wmissing-declarations"
+# if defined __clang__ || defined __APPLE__
+# pragma GCC diagnostic ignored "-Wmissing-prototypes"
+# pragma GCC diagnostic ignored "-Wextra"
+# endif
+#endif
+
+#ifndef __OPENCV_PERF_PRECOMP_HPP__
+#define __OPENCV_PERF_PRECOMP_HPP__
+
+#include "opencv2/ts.hpp"
+#include "opencv2/ts/gpu_perf.hpp"
+
+#include "opencv2/gpuwarping.hpp"
+#include "opencv2/imgproc.hpp"
+
+#ifdef GTEST_CREATE_SHARED_LIBRARY
+#error no modules except ts should have GTEST_CREATE_SHARED_LIBRARY defined
+#endif
+
+#endif
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "perf_precomp.hpp"
+
+using namespace std;
+using namespace testing;
+using namespace perf;
+
+//////////////////////////////////////////////////////////////////////
+// Remap
+
+enum { HALF_SIZE=0, UPSIDE_DOWN, REFLECTION_X, REFLECTION_BOTH };
+CV_ENUM(RemapMode, HALF_SIZE, UPSIDE_DOWN, REFLECTION_X, REFLECTION_BOTH)
+
+void generateMap(cv::Mat& map_x, cv::Mat& map_y, int remapMode)
+{
+ for (int j = 0; j < map_x.rows; ++j)
+ {
+ for (int i = 0; i < map_x.cols; ++i)
+ {
+ switch (remapMode)
+ {
+ case HALF_SIZE:
+ if (i > map_x.cols*0.25 && i < map_x.cols*0.75 && j > map_x.rows*0.25 && j < map_x.rows*0.75)
+ {
+ map_x.at<float>(j,i) = 2.f * (i - map_x.cols * 0.25f) + 0.5f;
+ map_y.at<float>(j,i) = 2.f * (j - map_x.rows * 0.25f) + 0.5f;
+ }
+ else
+ {
+ map_x.at<float>(j,i) = 0.f;
+ map_y.at<float>(j,i) = 0.f;
+ }
+ break;
+ case UPSIDE_DOWN:
+ map_x.at<float>(j,i) = static_cast<float>(i);
+ map_y.at<float>(j,i) = static_cast<float>(map_x.rows - j);
+ break;
+ case REFLECTION_X:
+ map_x.at<float>(j,i) = static_cast<float>(map_x.cols - i);
+ map_y.at<float>(j,i) = static_cast<float>(j);
+ break;
+ case REFLECTION_BOTH:
+ map_x.at<float>(j,i) = static_cast<float>(map_x.cols - i);
+ map_y.at<float>(j,i) = static_cast<float>(map_x.rows - j);
+ break;
+ } // end of switch
+ }
+ }
+}
+
+DEF_PARAM_TEST(Sz_Depth_Cn_Inter_Border_Mode, cv::Size, MatDepth, MatCn, Interpolation, BorderMode, RemapMode);
+
+PERF_TEST_P(Sz_Depth_Cn_Inter_Border_Mode, Remap,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_32F),
+ GPU_CHANNELS_1_3_4,
+ Values(Interpolation(cv::INTER_NEAREST), Interpolation(cv::INTER_LINEAR), Interpolation(cv::INTER_CUBIC)),
+ ALL_BORDER_MODES,
+ RemapMode::all()))
+{
+ declare.time(20.0);
+
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+ const int interpolation = GET_PARAM(3);
+ const int borderMode = GET_PARAM(4);
+ const int remapMode = GET_PARAM(5);
+
+ const int type = CV_MAKE_TYPE(depth, channels);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ cv::Mat xmap(size, CV_32FC1);
+ cv::Mat ymap(size, CV_32FC1);
+ generateMap(xmap, ymap, remapMode);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ const cv::gpu::GpuMat d_xmap(xmap);
+ const cv::gpu::GpuMat d_ymap(ymap);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::remap(d_src, dst, d_xmap, d_ymap, interpolation, borderMode);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::remap(src, dst, xmap, ymap, interpolation, borderMode);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// Resize
+
+DEF_PARAM_TEST(Sz_Depth_Cn_Inter_Scale, cv::Size, MatDepth, MatCn, Interpolation, double);
+
+PERF_TEST_P(Sz_Depth_Cn_Inter_Scale, Resize,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_32F),
+ GPU_CHANNELS_1_3_4,
+ Values(Interpolation(cv::INTER_NEAREST), Interpolation(cv::INTER_LINEAR), Interpolation(cv::INTER_CUBIC)),
+ Values(0.5, 0.3, 2.0)))
+{
+ declare.time(20.0);
+
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+ const int interpolation = GET_PARAM(3);
+ const double f = GET_PARAM(4);
+
+ const int type = CV_MAKE_TYPE(depth, channels);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::resize(d_src, dst, cv::Size(), f, f, interpolation);
+
+ GPU_SANITY_CHECK(dst, 1e-3, ERROR_RELATIVE);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::resize(src, dst, cv::Size(), f, f, interpolation);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// ResizeArea
+
+DEF_PARAM_TEST(Sz_Depth_Cn_Scale, cv::Size, MatDepth, MatCn, double);
+
+PERF_TEST_P(Sz_Depth_Cn_Scale, ResizeArea,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_32F),
+ GPU_CHANNELS_1_3_4,
+ Values(0.2, 0.1, 0.05)))
+{
+ declare.time(1.0);
+
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+ const int interpolation = cv::INTER_AREA;
+ const double f = GET_PARAM(3);
+
+ const int type = CV_MAKE_TYPE(depth, channels);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::resize(d_src, dst, cv::Size(), f, f, interpolation);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::resize(src, dst, cv::Size(), f, f, interpolation);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// WarpAffine
+
+DEF_PARAM_TEST(Sz_Depth_Cn_Inter_Border, cv::Size, MatDepth, MatCn, Interpolation, BorderMode);
+
+PERF_TEST_P(Sz_Depth_Cn_Inter_Border, WarpAffine,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_32F),
+ GPU_CHANNELS_1_3_4,
+ Values(Interpolation(cv::INTER_NEAREST), Interpolation(cv::INTER_LINEAR), Interpolation(cv::INTER_CUBIC)),
+ ALL_BORDER_MODES))
+{
+ declare.time(20.0);
+
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+ const int interpolation = GET_PARAM(3);
+ const int borderMode = GET_PARAM(4);
+
+ const int type = CV_MAKE_TYPE(depth, channels);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ const double aplha = CV_PI / 4;
+ const double mat[2 * 3] =
+ {
+ std::cos(aplha), -std::sin(aplha), src.cols / 2,
+ std::sin(aplha), std::cos(aplha), 0
+ };
+ const cv::Mat M(2, 3, CV_64F, (void*) mat);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::warpAffine(d_src, dst, M, size, interpolation, borderMode);
+
+ GPU_SANITY_CHECK(dst, 1);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::warpAffine(src, dst, M, size, interpolation, borderMode);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// WarpPerspective
+
+PERF_TEST_P(Sz_Depth_Cn_Inter_Border, WarpPerspective,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_32F),
+ GPU_CHANNELS_1_3_4,
+ Values(Interpolation(cv::INTER_NEAREST), Interpolation(cv::INTER_LINEAR), Interpolation(cv::INTER_CUBIC)),
+ ALL_BORDER_MODES))
+{
+ declare.time(20.0);
+
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+ const int interpolation = GET_PARAM(3);
+ const int borderMode = GET_PARAM(4);
+
+ const int type = CV_MAKE_TYPE(depth, channels);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ const double aplha = CV_PI / 4;
+ double mat[3][3] = { {std::cos(aplha), -std::sin(aplha), src.cols / 2},
+ {std::sin(aplha), std::cos(aplha), 0},
+ {0.0, 0.0, 1.0}};
+ const cv::Mat M(3, 3, CV_64F, (void*) mat);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::warpPerspective(d_src, dst, M, size, interpolation, borderMode);
+
+ GPU_SANITY_CHECK(dst, 1);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::warpPerspective(src, dst, M, size, interpolation, borderMode);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// BuildWarpPlaneMaps
+
+PERF_TEST_P(Sz, BuildWarpPlaneMaps,
+ GPU_TYPICAL_MAT_SIZES)
+{
+ const cv::Size size = GetParam();
+
+ const cv::Mat K = cv::Mat::eye(3, 3, CV_32FC1);
+ const cv::Mat R = cv::Mat::ones(3, 3, CV_32FC1);
+ const cv::Mat T = cv::Mat::zeros(1, 3, CV_32F);
+
+ if (PERF_RUN_GPU())
+ {
+ cv::gpu::GpuMat map_x;
+ cv::gpu::GpuMat map_y;
+
+ TEST_CYCLE() cv::gpu::buildWarpPlaneMaps(size, cv::Rect(0, 0, size.width, size.height), K, R, T, 1.0, map_x, map_y);
+
+ GPU_SANITY_CHECK(map_x);
+ GPU_SANITY_CHECK(map_y);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// BuildWarpCylindricalMaps
+
+PERF_TEST_P(Sz, BuildWarpCylindricalMaps,
+ GPU_TYPICAL_MAT_SIZES)
+{
+ const cv::Size size = GetParam();
+
+ const cv::Mat K = cv::Mat::eye(3, 3, CV_32FC1);
+ const cv::Mat R = cv::Mat::ones(3, 3, CV_32FC1);
+
+ if (PERF_RUN_GPU())
+ {
+ cv::gpu::GpuMat map_x;
+ cv::gpu::GpuMat map_y;
+
+ TEST_CYCLE() cv::gpu::buildWarpCylindricalMaps(size, cv::Rect(0, 0, size.width, size.height), K, R, 1.0, map_x, map_y);
+
+ GPU_SANITY_CHECK(map_x);
+ GPU_SANITY_CHECK(map_y);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// BuildWarpSphericalMaps
+
+PERF_TEST_P(Sz, BuildWarpSphericalMaps,
+ GPU_TYPICAL_MAT_SIZES)
+{
+ const cv::Size size = GetParam();
+
+ const cv::Mat K = cv::Mat::eye(3, 3, CV_32FC1);
+ const cv::Mat R = cv::Mat::ones(3, 3, CV_32FC1);
+
+ if (PERF_RUN_GPU())
+ {
+ cv::gpu::GpuMat map_x;
+ cv::gpu::GpuMat map_y;
+
+ TEST_CYCLE() cv::gpu::buildWarpSphericalMaps(size, cv::Rect(0, 0, size.width, size.height), K, R, 1.0, map_x, map_y);
+
+ GPU_SANITY_CHECK(map_x);
+ GPU_SANITY_CHECK(map_y);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// Rotate
+
+DEF_PARAM_TEST(Sz_Depth_Cn_Inter, cv::Size, MatDepth, MatCn, Interpolation);
+
+PERF_TEST_P(Sz_Depth_Cn_Inter, Rotate,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_32F),
+ GPU_CHANNELS_1_3_4,
+ Values(Interpolation(cv::INTER_NEAREST), Interpolation(cv::INTER_LINEAR), Interpolation(cv::INTER_CUBIC))))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+ const int interpolation = GET_PARAM(3);
+
+ const int type = CV_MAKE_TYPE(depth, channels);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::rotate(d_src, dst, size, 30.0, 0, 0, interpolation);
+
+ GPU_SANITY_CHECK(dst, 1e-3, ERROR_RELATIVE);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// PyrDown
+
+PERF_TEST_P(Sz_Depth_Cn, PyrDown,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_32F),
+ GPU_CHANNELS_1_3_4))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+
+ const int type = CV_MAKE_TYPE(depth, channels);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::pyrDown(d_src, dst);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::pyrDown(src, dst);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// PyrUp
+
+PERF_TEST_P(Sz_Depth_Cn, PyrUp,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_32F),
+ GPU_CHANNELS_1_3_4))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+
+ const int type = CV_MAKE_TYPE(depth, channels);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ TEST_CYCLE() cv::gpu::pyrUp(d_src, dst);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ cv::Mat dst;
+
+ TEST_CYCLE() cv::pyrUp(src, dst);
+
+ CPU_SANITY_CHECK(dst);
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// ImagePyramidBuild
+
+PERF_TEST_P(Sz_Depth_Cn, ImagePyramidBuild,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_32F),
+ GPU_CHANNELS_1_3_4))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+
+ const int type = CV_MAKE_TYPE(depth, channels);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ const int nLayers = 5;
+ const cv::Size dstSize(size.width / 2 + 10, size.height / 2 + 10);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+
+ cv::gpu::ImagePyramid d_pyr;
+
+ TEST_CYCLE() d_pyr.build(d_src, nLayers);
+
+ cv::gpu::GpuMat dst;
+ d_pyr.getLayer(dst, dstSize);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
+
+//////////////////////////////////////////////////////////////////////
+// ImagePyramidGetLayer
+
+PERF_TEST_P(Sz_Depth_Cn, ImagePyramidGetLayer,
+ Combine(GPU_TYPICAL_MAT_SIZES,
+ Values(CV_8U, CV_16U, CV_32F),
+ GPU_CHANNELS_1_3_4))
+{
+ const cv::Size size = GET_PARAM(0);
+ const int depth = GET_PARAM(1);
+ const int channels = GET_PARAM(2);
+
+ const int type = CV_MAKE_TYPE(depth, channels);
+
+ cv::Mat src(size, type);
+ declare.in(src, WARMUP_RNG);
+
+ const int nLayers = 3;
+ const cv::Size dstSize(size.width / 2 + 10, size.height / 2 + 10);
+
+ if (PERF_RUN_GPU())
+ {
+ const cv::gpu::GpuMat d_src(src);
+ cv::gpu::GpuMat dst;
+
+ cv::gpu::ImagePyramid d_pyr(d_src, nLayers);
+
+ TEST_CYCLE() d_pyr.getLayer(dst, dstSize);
+
+ GPU_SANITY_CHECK(dst);
+ }
+ else
+ {
+ FAIL_NO_CPU();
+ }
+}
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#if !defined CUDA_DISABLER
+
+#include "opencv2/core/cuda/common.hpp"
+#include "opencv2/core/cuda/vec_traits.hpp"
+#include "opencv2/core/cuda/vec_math.hpp"
+#include "opencv2/core/cuda/saturate_cast.hpp"
+#include "opencv2/core/cuda/border_interpolate.hpp"
+
+namespace cv { namespace gpu { namespace cudev
+{
+ namespace imgproc
+ {
+ // TODO use intrinsics like __sinf and so on
+
+ namespace build_warp_maps
+ {
+
+ __constant__ float ck_rinv[9];
+ __constant__ float cr_kinv[9];
+ __constant__ float ct[3];
+ __constant__ float cscale;
+ }
+
+
+ class PlaneMapper
+ {
+ public:
+ static __device__ __forceinline__ void mapBackward(float u, float v, float &x, float &y)
+ {
+ using namespace build_warp_maps;
+
+ float x_ = u / cscale - ct[0];
+ float y_ = v / cscale - ct[1];
+
+ float z;
+ x = ck_rinv[0] * x_ + ck_rinv[1] * y_ + ck_rinv[2] * (1 - ct[2]);
+ y = ck_rinv[3] * x_ + ck_rinv[4] * y_ + ck_rinv[5] * (1 - ct[2]);
+ z = ck_rinv[6] * x_ + ck_rinv[7] * y_ + ck_rinv[8] * (1 - ct[2]);
+
+ x /= z;
+ y /= z;
+ }
+ };
+
+
+ class CylindricalMapper
+ {
+ public:
+ static __device__ __forceinline__ void mapBackward(float u, float v, float &x, float &y)
+ {
+ using namespace build_warp_maps;
+
+ u /= cscale;
+ float x_ = ::sinf(u);
+ float y_ = v / cscale;
+ float z_ = ::cosf(u);
+
+ float z;
+ x = ck_rinv[0] * x_ + ck_rinv[1] * y_ + ck_rinv[2] * z_;
+ y = ck_rinv[3] * x_ + ck_rinv[4] * y_ + ck_rinv[5] * z_;
+ z = ck_rinv[6] * x_ + ck_rinv[7] * y_ + ck_rinv[8] * z_;
+
+ if (z > 0) { x /= z; y /= z; }
+ else x = y = -1;
+ }
+ };
+
+
+ class SphericalMapper
+ {
+ public:
+ static __device__ __forceinline__ void mapBackward(float u, float v, float &x, float &y)
+ {
+ using namespace build_warp_maps;
+
+ v /= cscale;
+ u /= cscale;
+
+ float sinv = ::sinf(v);
+ float x_ = sinv * ::sinf(u);
+ float y_ = -::cosf(v);
+ float z_ = sinv * ::cosf(u);
+
+ float z;
+ x = ck_rinv[0] * x_ + ck_rinv[1] * y_ + ck_rinv[2] * z_;
+ y = ck_rinv[3] * x_ + ck_rinv[4] * y_ + ck_rinv[5] * z_;
+ z = ck_rinv[6] * x_ + ck_rinv[7] * y_ + ck_rinv[8] * z_;
+
+ if (z > 0) { x /= z; y /= z; }
+ else x = y = -1;
+ }
+ };
+
+
+ template <typename Mapper>
+ __global__ void buildWarpMapsKernel(int tl_u, int tl_v, int cols, int rows,
+ PtrStepf map_x, PtrStepf map_y)
+ {
+ int du = blockIdx.x * blockDim.x + threadIdx.x;
+ int dv = blockIdx.y * blockDim.y + threadIdx.y;
+ if (du < cols && dv < rows)
+ {
+ float u = tl_u + du;
+ float v = tl_v + dv;
+ float x, y;
+ Mapper::mapBackward(u, v, x, y);
+ map_x.ptr(dv)[du] = x;
+ map_y.ptr(dv)[du] = y;
+ }
+ }
+
+
+ void buildWarpPlaneMaps(int tl_u, int tl_v, PtrStepSzf map_x, PtrStepSzf map_y,
+ const float k_rinv[9], const float r_kinv[9], const float t[3],
+ float scale, cudaStream_t stream)
+ {
+ cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::ck_rinv, k_rinv, 9*sizeof(float)));
+ cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::cr_kinv, r_kinv, 9*sizeof(float)));
+ cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::ct, t, 3*sizeof(float)));
+ cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::cscale, &scale, sizeof(float)));
+
+ int cols = map_x.cols;
+ int rows = map_x.rows;
+
+ dim3 threads(32, 8);
+ dim3 grid(divUp(cols, threads.x), divUp(rows, threads.y));
+
+ buildWarpMapsKernel<PlaneMapper><<<grid,threads>>>(tl_u, tl_v, cols, rows, map_x, map_y);
+ cudaSafeCall(cudaGetLastError());
+ if (stream == 0)
+ cudaSafeCall(cudaDeviceSynchronize());
+ }
+
+
+ void buildWarpCylindricalMaps(int tl_u, int tl_v, PtrStepSzf map_x, PtrStepSzf map_y,
+ const float k_rinv[9], const float r_kinv[9], float scale,
+ cudaStream_t stream)
+ {
+ cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::ck_rinv, k_rinv, 9*sizeof(float)));
+ cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::cr_kinv, r_kinv, 9*sizeof(float)));
+ cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::cscale, &scale, sizeof(float)));
+
+ int cols = map_x.cols;
+ int rows = map_x.rows;
+
+ dim3 threads(32, 8);
+ dim3 grid(divUp(cols, threads.x), divUp(rows, threads.y));
+
+ buildWarpMapsKernel<CylindricalMapper><<<grid,threads>>>(tl_u, tl_v, cols, rows, map_x, map_y);
+ cudaSafeCall(cudaGetLastError());
+ if (stream == 0)
+ cudaSafeCall(cudaDeviceSynchronize());
+ }
+
+
+ void buildWarpSphericalMaps(int tl_u, int tl_v, PtrStepSzf map_x, PtrStepSzf map_y,
+ const float k_rinv[9], const float r_kinv[9], float scale,
+ cudaStream_t stream)
+ {
+ cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::ck_rinv, k_rinv, 9*sizeof(float)));
+ cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::cr_kinv, r_kinv, 9*sizeof(float)));
+ cudaSafeCall(cudaMemcpyToSymbol(build_warp_maps::cscale, &scale, sizeof(float)));
+
+ int cols = map_x.cols;
+ int rows = map_x.rows;
+
+ dim3 threads(32, 8);
+ dim3 grid(divUp(cols, threads.x), divUp(rows, threads.y));
+
+ buildWarpMapsKernel<SphericalMapper><<<grid,threads>>>(tl_u, tl_v, cols, rows, map_x, map_y);
+ cudaSafeCall(cudaGetLastError());
+ if (stream == 0)
+ cudaSafeCall(cudaDeviceSynchronize());
+ }
+ } // namespace imgproc
+}}} // namespace cv { namespace gpu { namespace cudev {
+
+
+#endif /* CUDA_DISABLER */
static const caller_t callers[3][5] =
{
{
- RemapDispatcher<PointFilter, BrdReflect101, T>::call,
- RemapDispatcher<PointFilter, BrdReplicate, T>::call,
RemapDispatcher<PointFilter, BrdConstant, T>::call,
+ RemapDispatcher<PointFilter, BrdReplicate, T>::call,
RemapDispatcher<PointFilter, BrdReflect, T>::call,
- RemapDispatcher<PointFilter, BrdWrap, T>::call
+ RemapDispatcher<PointFilter, BrdWrap, T>::call,
+ RemapDispatcher<PointFilter, BrdReflect101, T>::call
},
{
- RemapDispatcher<LinearFilter, BrdReflect101, T>::call,
- RemapDispatcher<LinearFilter, BrdReplicate, T>::call,
RemapDispatcher<LinearFilter, BrdConstant, T>::call,
+ RemapDispatcher<LinearFilter, BrdReplicate, T>::call,
RemapDispatcher<LinearFilter, BrdReflect, T>::call,
- RemapDispatcher<LinearFilter, BrdWrap, T>::call
+ RemapDispatcher<LinearFilter, BrdWrap, T>::call,
+ RemapDispatcher<LinearFilter, BrdReflect101, T>::call
},
{
- RemapDispatcher<CubicFilter, BrdReflect101, T>::call,
- RemapDispatcher<CubicFilter, BrdReplicate, T>::call,
RemapDispatcher<CubicFilter, BrdConstant, T>::call,
+ RemapDispatcher<CubicFilter, BrdReplicate, T>::call,
RemapDispatcher<CubicFilter, BrdReflect, T>::call,
- RemapDispatcher<CubicFilter, BrdWrap, T>::call
+ RemapDispatcher<CubicFilter, BrdWrap, T>::call,
+ RemapDispatcher<CubicFilter, BrdReflect101, T>::call
}
};
static const func_t funcs[3][5] =
{
{
- WarpDispatcher<Transform, PointFilter, BrdReflect101, T>::call,
- WarpDispatcher<Transform, PointFilter, BrdReplicate, T>::call,
WarpDispatcher<Transform, PointFilter, BrdConstant, T>::call,
+ WarpDispatcher<Transform, PointFilter, BrdReplicate, T>::call,
WarpDispatcher<Transform, PointFilter, BrdReflect, T>::call,
- WarpDispatcher<Transform, PointFilter, BrdWrap, T>::call
+ WarpDispatcher<Transform, PointFilter, BrdWrap, T>::call,
+ WarpDispatcher<Transform, PointFilter, BrdReflect101, T>::call
},
{
- WarpDispatcher<Transform, LinearFilter, BrdReflect101, T>::call,
- WarpDispatcher<Transform, LinearFilter, BrdReplicate, T>::call,
WarpDispatcher<Transform, LinearFilter, BrdConstant, T>::call,
+ WarpDispatcher<Transform, LinearFilter, BrdReplicate, T>::call,
WarpDispatcher<Transform, LinearFilter, BrdReflect, T>::call,
- WarpDispatcher<Transform, LinearFilter, BrdWrap, T>::call
+ WarpDispatcher<Transform, LinearFilter, BrdWrap, T>::call,
+ WarpDispatcher<Transform, LinearFilter, BrdReflect101, T>::call
},
{
- WarpDispatcher<Transform, CubicFilter, BrdReflect101, T>::call,
- WarpDispatcher<Transform, CubicFilter, BrdReplicate, T>::call,
WarpDispatcher<Transform, CubicFilter, BrdConstant, T>::call,
+ WarpDispatcher<Transform, CubicFilter, BrdReplicate, T>::call,
WarpDispatcher<Transform, CubicFilter, BrdReflect, T>::call,
- WarpDispatcher<Transform, CubicFilter, BrdWrap, T>::call
+ WarpDispatcher<Transform, CubicFilter, BrdWrap, T>::call,
+ WarpDispatcher<Transform, CubicFilter, BrdReflect101, T>::call
}
};
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_PRECOMP_H__
+#define __OPENCV_PRECOMP_H__
+
+#include "opencv2/gpuwarping.hpp"
+
+#include "opencv2/core/gpu_private.hpp"
+
+#include "opencv2/opencv_modules.hpp"
+
+#ifdef HAVE_OPENCV_GPULEGACY
+# include "opencv2/gpulegacy.hpp"
+# include "opencv2/gpulegacy/private.hpp"
+#endif
+
+#endif /* __OPENCV_PRECOMP_H__ */
#if !defined HAVE_CUDA || defined(CUDA_DISABLER)
void cv::gpu::pyrDown(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+
void cv::gpu::pyrUp(const GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+
void cv::gpu::ImagePyramid::build(const GpuMat&, int, Stream&) { throw_no_cuda(); }
void cv::gpu::ImagePyramid::getLayer(GpuMat&, Size, Stream&) const { throw_no_cuda(); }
//////////////////////////////////////////////////////////////////////////////
// ImagePyramid
-namespace cv { namespace gpu { namespace cudev
-{
- namespace pyramid
- {
- template <typename T> void kernelDownsampleX2_gpu(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- template <typename T> void kernelInterpolateFrom1_gpu(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
- }
-}}}
-
void cv::gpu::ImagePyramid::build(const GpuMat& img, int numLayers, Stream& stream)
{
- using namespace cv::gpu::cudev::pyramid;
-
- typedef void (*func_t)(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
-
- static const func_t funcs[6][4] =
- {
- {kernelDownsampleX2_gpu<uchar1> , 0 /*kernelDownsampleX2_gpu<uchar2>*/ , kernelDownsampleX2_gpu<uchar3> , kernelDownsampleX2_gpu<uchar4> },
- {0 /*kernelDownsampleX2_gpu<char1>*/ , 0 /*kernelDownsampleX2_gpu<char2>*/ , 0 /*kernelDownsampleX2_gpu<char3>*/ , 0 /*kernelDownsampleX2_gpu<char4>*/ },
- {kernelDownsampleX2_gpu<ushort1> , 0 /*kernelDownsampleX2_gpu<ushort2>*/, kernelDownsampleX2_gpu<ushort3> , kernelDownsampleX2_gpu<ushort4> },
- {0 /*kernelDownsampleX2_gpu<short1>*/ , 0 /*kernelDownsampleX2_gpu<short2>*/ , 0 /*kernelDownsampleX2_gpu<short3>*/, 0 /*kernelDownsampleX2_gpu<short4>*/},
- {0 /*kernelDownsampleX2_gpu<int1>*/ , 0 /*kernelDownsampleX2_gpu<int2>*/ , 0 /*kernelDownsampleX2_gpu<int3>*/ , 0 /*kernelDownsampleX2_gpu<int4>*/ },
- {kernelDownsampleX2_gpu<float1> , 0 /*kernelDownsampleX2_gpu<float2>*/ , kernelDownsampleX2_gpu<float3> , kernelDownsampleX2_gpu<float4> }
- };
-
+#ifndef HAVE_OPENCV_GPULEGACY
+ (void) img;
+ (void) numLayers;
+ (void) stream;
+ throw_no_cuda();
+#else
CV_Assert(img.depth() <= CV_32F && img.channels() <= 4);
- const func_t func = funcs[img.depth()][img.channels() - 1];
- CV_Assert(func != 0);
-
layer0_ = img;
Size szLastLayer = img.size();
nLayers_ = 1;
const GpuMat& prevLayer = i == 0 ? layer0_ : pyramid_[i - 1];
- func(prevLayer, pyramid_[i], StreamAccessor::getStream(stream));
+ cudev::pyramid::downsampleX2(prevLayer, pyramid_[i], img.depth(), img.channels(), StreamAccessor::getStream(stream));
szLastLayer = szCurLayer;
}
+#endif
}
void cv::gpu::ImagePyramid::getLayer(GpuMat& outImg, Size outRoi, Stream& stream) const
{
- using namespace cv::gpu::cudev::pyramid;
-
- typedef void (*func_t)(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
-
- static const func_t funcs[6][4] =
- {
- {kernelInterpolateFrom1_gpu<uchar1> , 0 /*kernelInterpolateFrom1_gpu<uchar2>*/ , kernelInterpolateFrom1_gpu<uchar3> , kernelInterpolateFrom1_gpu<uchar4> },
- {0 /*kernelInterpolateFrom1_gpu<char1>*/ , 0 /*kernelInterpolateFrom1_gpu<char2>*/ , 0 /*kernelInterpolateFrom1_gpu<char3>*/ , 0 /*kernelInterpolateFrom1_gpu<char4>*/ },
- {kernelInterpolateFrom1_gpu<ushort1> , 0 /*kernelInterpolateFrom1_gpu<ushort2>*/, kernelInterpolateFrom1_gpu<ushort3> , kernelInterpolateFrom1_gpu<ushort4> },
- {0 /*kernelInterpolateFrom1_gpu<short1>*/, 0 /*kernelInterpolateFrom1_gpu<short2>*/ , 0 /*kernelInterpolateFrom1_gpu<short3>*/, 0 /*kernelInterpolateFrom1_gpu<short4>*/},
- {0 /*kernelInterpolateFrom1_gpu<int1>*/ , 0 /*kernelInterpolateFrom1_gpu<int2>*/ , 0 /*kernelInterpolateFrom1_gpu<int3>*/ , 0 /*kernelInterpolateFrom1_gpu<int4>*/ },
- {kernelInterpolateFrom1_gpu<float1> , 0 /*kernelInterpolateFrom1_gpu<float2>*/ , kernelInterpolateFrom1_gpu<float3> , kernelInterpolateFrom1_gpu<float4> }
- };
-
+#ifndef HAVE_OPENCV_GPULEGACY
+ (void) outImg;
+ (void) outRoi;
+ (void) stream;
+ throw_no_cuda();
+#else
CV_Assert(outRoi.width <= layer0_.cols && outRoi.height <= layer0_.rows && outRoi.width > 0 && outRoi.height > 0);
ensureSizeIsEnough(outRoi, layer0_.type(), outImg);
- const func_t func = funcs[outImg.depth()][outImg.channels() - 1];
- CV_Assert(func != 0);
-
if (outRoi.width == layer0_.cols && outRoi.height == layer0_.rows)
{
if (stream)
lastLayer = curLayer;
}
- func(lastLayer, outImg, StreamAccessor::getStream(stream));
+ cudev::pyramid::interpolateFrom1(lastLayer, outImg, outImg.depth(), outImg.channels(), StreamAccessor::getStream(stream));
+#endif
}
#endif // HAVE_CUDA
const func_t func = funcs[src.depth()][src.channels() - 1];
CV_Assert(func != 0);
- int gpuBorderType;
- CV_Assert(tryConvertToGpuBorderType(borderMode, gpuBorderType));
-
dst.create(xmap.size(), src.type());
Scalar_<float> borderValueFloat;
src.locateROI(wholeSize, ofs);
func(src, PtrStepSzb(wholeSize.height, wholeSize.width, src.datastart, src.step), ofs.x, ofs.y, xmap, ymap,
- dst, interpolation, gpuBorderType, borderValueFloat.val, StreamAccessor::getStream(stream), deviceSupports(FEATURE_SET_COMPUTE_20));
+ dst, interpolation, borderMode, borderValueFloat.val, StreamAccessor::getStream(stream), deviceSupports(FEATURE_SET_COMPUTE_20));
}
#endif // HAVE_CUDA
#if !defined HAVE_CUDA || defined(CUDA_DISABLER)
-void cv::gpu::resize(const GpuMat& src, GpuMat& dst, Size dsize, double fx, double fy, int interpolation, Stream& s)
-{
- (void)src;
- (void)dst;
- (void)dsize;
- (void)fx;
- (void)fy;
- (void)interpolation;
- (void)s;
-
- throw_no_cuda();
-}
+void cv::gpu::resize(const GpuMat&, GpuMat&, Size, double, double, int, Stream&) { throw_no_cuda(); }
#else // HAVE_CUDA
#include "precomp.hpp"
-#if !defined HAVE_CUDA || defined(CUDA_DISABLER)
+using namespace cv;
+using namespace cv::gpu;
+#if !defined HAVE_CUDA || defined(CUDA_DISABLER)
void cv::gpu::warpAffine(const GpuMat&, GpuMat&, const Mat&, Size, int, int, Scalar, Stream&) { throw_no_cuda(); }
void cv::gpu::buildWarpAffineMaps(const Mat&, bool, Size, GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
void cv::gpu::warpPerspective(const GpuMat&, GpuMat&, const Mat&, Size, int, int, Scalar, Stream&) { throw_no_cuda(); }
void cv::gpu::buildWarpPerspectiveMaps(const Mat&, bool, Size, GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+void cv::gpu::buildWarpPlaneMaps(Size, Rect, const Mat&, const Mat&, const Mat&, float, GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+void cv::gpu::buildWarpCylindricalMaps(Size, Rect, const Mat&, const Mat&, float, GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+void cv::gpu::buildWarpSphericalMaps(Size, Rect, const Mat&, const Mat&, float, GpuMat&, GpuMat&, Stream&) { throw_no_cuda(); }
+
+void cv::gpu::rotate(const GpuMat&, GpuMat&, Size, double, double, double, int, Stream&) { throw_no_cuda(); }
+
#else // HAVE_CUDA
namespace cv { namespace gpu { namespace cudev
namespace
{
- template<int DEPTH> struct NppTypeTraits;
- template<> struct NppTypeTraits<CV_8U> { typedef Npp8u npp_t; };
- template<> struct NppTypeTraits<CV_8S> { typedef Npp8s npp_t; };
- template<> struct NppTypeTraits<CV_16U> { typedef Npp16u npp_t; };
- template<> struct NppTypeTraits<CV_16S> { typedef Npp16s npp_t; typedef Npp16sc npp_complex_type; };
- template<> struct NppTypeTraits<CV_32S> { typedef Npp32s npp_t; typedef Npp32sc npp_complex_type; };
- template<> struct NppTypeTraits<CV_32F> { typedef Npp32f npp_t; typedef Npp32fc npp_complex_type; };
- template<> struct NppTypeTraits<CV_64F> { typedef Npp64f npp_t; typedef Npp64fc npp_complex_type; };
-
template <int DEPTH> struct NppWarpFunc
{
- typedef typename NppTypeTraits<DEPTH>::npp_t npp_t;
+ typedef typename NPPTypeTraits<DEPTH>::npp_type npp_type;
- typedef NppStatus (*func_t)(const npp_t* pSrc, NppiSize srcSize, int srcStep, NppiRect srcRoi, npp_t* pDst,
+ typedef NppStatus (*func_t)(const npp_type* pSrc, NppiSize srcSize, int srcStep, NppiRect srcRoi, npp_type* pDst,
int dstStep, NppiRect dstRoi, const double coeffs[][3],
int interpolation);
};
template <int DEPTH, typename NppWarpFunc<DEPTH>::func_t func> struct NppWarp
{
- typedef typename NppWarpFunc<DEPTH>::npp_t npp_t;
+ typedef typename NppWarpFunc<DEPTH>::npp_type npp_type;
static void call(const cv::gpu::GpuMat& src, cv::gpu::GpuMat& dst, double coeffs[][3], int interpolation, cudaStream_t stream)
{
cv::gpu::NppStreamHandler h(stream);
- nppSafeCall( func(src.ptr<npp_t>(), srcsz, static_cast<int>(src.step), srcroi,
- dst.ptr<npp_t>(), static_cast<int>(dst.step), dstroi,
+ nppSafeCall( func(src.ptr<npp_type>(), srcsz, static_cast<int>(src.step), srcroi,
+ dst.ptr<npp_type>(), static_cast<int>(dst.step), dstroi,
coeffs, npp_inter[interpolation]) );
if (stream == 0)
const func_t func = funcs[src.depth()][src.channels() - 1];
CV_Assert(func != 0);
- int gpuBorderType;
- CV_Assert(tryConvertToGpuBorderType(borderMode, gpuBorderType));
-
float coeffs[2 * 3];
Mat coeffsMat(2, 3, CV_32F, (void*)coeffs);
borderValueFloat = borderValue;
func(src, PtrStepSzb(wholeSize.height, wholeSize.width, src.datastart, src.step), ofs.x, ofs.y, coeffs,
- dst, interpolation, gpuBorderType, borderValueFloat.val, StreamAccessor::getStream(s), deviceSupports(FEATURE_SET_COMPUTE_20));
+ dst, interpolation, borderMode, borderValueFloat.val, StreamAccessor::getStream(s), deviceSupports(FEATURE_SET_COMPUTE_20));
}
}
const func_t func = funcs[src.depth()][src.channels() - 1];
CV_Assert(func != 0);
- int gpuBorderType;
- CV_Assert(tryConvertToGpuBorderType(borderMode, gpuBorderType));
-
float coeffs[3 * 3];
Mat coeffsMat(3, 3, CV_32F, (void*)coeffs);
borderValueFloat = borderValue;
func(src, PtrStepSzb(wholeSize.height, wholeSize.width, src.datastart, src.step), ofs.x, ofs.y, coeffs,
- dst, interpolation, gpuBorderType, borderValueFloat.val, StreamAccessor::getStream(s), deviceSupports(FEATURE_SET_COMPUTE_20));
+ dst, interpolation, borderMode, borderValueFloat.val, StreamAccessor::getStream(s), deviceSupports(FEATURE_SET_COMPUTE_20));
+ }
+}
+
+//////////////////////////////////////////////////////////////////////////////
+// buildWarpPlaneMaps
+
+namespace cv { namespace gpu { namespace cudev
+{
+ namespace imgproc
+ {
+ void buildWarpPlaneMaps(int tl_u, int tl_v, PtrStepSzf map_x, PtrStepSzf map_y,
+ const float k_rinv[9], const float r_kinv[9], const float t[3], float scale,
+ cudaStream_t stream);
+ }
+}}}
+
+void cv::gpu::buildWarpPlaneMaps(Size src_size, Rect dst_roi, const Mat &K, const Mat& R, const Mat &T,
+ float scale, GpuMat& map_x, GpuMat& map_y, Stream& stream)
+{
+ (void)src_size;
+ using namespace ::cv::gpu::cudev::imgproc;
+
+ CV_Assert(K.size() == Size(3,3) && K.type() == CV_32F);
+ CV_Assert(R.size() == Size(3,3) && R.type() == CV_32F);
+ CV_Assert((T.size() == Size(3,1) || T.size() == Size(1,3)) && T.type() == CV_32F && T.isContinuous());
+
+ Mat K_Rinv = K * R.t();
+ Mat R_Kinv = R * K.inv();
+ CV_Assert(K_Rinv.isContinuous());
+ CV_Assert(R_Kinv.isContinuous());
+
+ map_x.create(dst_roi.size(), CV_32F);
+ map_y.create(dst_roi.size(), CV_32F);
+ cudev::imgproc::buildWarpPlaneMaps(dst_roi.tl().x, dst_roi.tl().y, map_x, map_y, K_Rinv.ptr<float>(), R_Kinv.ptr<float>(),
+ T.ptr<float>(), scale, StreamAccessor::getStream(stream));
+}
+
+//////////////////////////////////////////////////////////////////////////////
+// buildWarpCylyndricalMaps
+
+namespace cv { namespace gpu { namespace cudev
+{
+ namespace imgproc
+ {
+ void buildWarpCylindricalMaps(int tl_u, int tl_v, PtrStepSzf map_x, PtrStepSzf map_y,
+ const float k_rinv[9], const float r_kinv[9], float scale,
+ cudaStream_t stream);
+ }
+}}}
+
+void cv::gpu::buildWarpCylindricalMaps(Size src_size, Rect dst_roi, const Mat &K, const Mat& R, float scale,
+ GpuMat& map_x, GpuMat& map_y, Stream& stream)
+{
+ (void)src_size;
+ using namespace ::cv::gpu::cudev::imgproc;
+
+ CV_Assert(K.size() == Size(3,3) && K.type() == CV_32F);
+ CV_Assert(R.size() == Size(3,3) && R.type() == CV_32F);
+
+ Mat K_Rinv = K * R.t();
+ Mat R_Kinv = R * K.inv();
+ CV_Assert(K_Rinv.isContinuous());
+ CV_Assert(R_Kinv.isContinuous());
+
+ map_x.create(dst_roi.size(), CV_32F);
+ map_y.create(dst_roi.size(), CV_32F);
+ cudev::imgproc::buildWarpCylindricalMaps(dst_roi.tl().x, dst_roi.tl().y, map_x, map_y, K_Rinv.ptr<float>(), R_Kinv.ptr<float>(), scale, StreamAccessor::getStream(stream));
+}
+
+
+//////////////////////////////////////////////////////////////////////////////
+// buildWarpSphericalMaps
+
+namespace cv { namespace gpu { namespace cudev
+{
+ namespace imgproc
+ {
+ void buildWarpSphericalMaps(int tl_u, int tl_v, PtrStepSzf map_x, PtrStepSzf map_y,
+ const float k_rinv[9], const float r_kinv[9], float scale,
+ cudaStream_t stream);
}
+}}}
+
+void cv::gpu::buildWarpSphericalMaps(Size src_size, Rect dst_roi, const Mat &K, const Mat& R, float scale,
+ GpuMat& map_x, GpuMat& map_y, Stream& stream)
+{
+ (void)src_size;
+ using namespace ::cv::gpu::cudev::imgproc;
+
+ CV_Assert(K.size() == Size(3,3) && K.type() == CV_32F);
+ CV_Assert(R.size() == Size(3,3) && R.type() == CV_32F);
+
+ Mat K_Rinv = K * R.t();
+ Mat R_Kinv = R * K.inv();
+ CV_Assert(K_Rinv.isContinuous());
+ CV_Assert(R_Kinv.isContinuous());
+
+ map_x.create(dst_roi.size(), CV_32F);
+ map_y.create(dst_roi.size(), CV_32F);
+ cudev::imgproc::buildWarpSphericalMaps(dst_roi.tl().x, dst_roi.tl().y, map_x, map_y, K_Rinv.ptr<float>(), R_Kinv.ptr<float>(), scale, StreamAccessor::getStream(stream));
+}
+
+////////////////////////////////////////////////////////////////////////
+// rotate
+
+namespace
+{
+ template <int DEPTH> struct NppRotateFunc
+ {
+ typedef typename NPPTypeTraits<DEPTH>::npp_type npp_type;
+
+ typedef NppStatus (*func_t)(const npp_type* pSrc, NppiSize oSrcSize, int nSrcStep, NppiRect oSrcROI,
+ npp_type* pDst, int nDstStep, NppiRect oDstROI,
+ double nAngle, double nShiftX, double nShiftY, int eInterpolation);
+ };
+
+ template <int DEPTH, typename NppRotateFunc<DEPTH>::func_t func> struct NppRotate
+ {
+ typedef typename NppRotateFunc<DEPTH>::npp_type npp_type;
+
+ static void call(const GpuMat& src, GpuMat& dst, Size dsize, double angle, double xShift, double yShift, int interpolation, cudaStream_t stream)
+ {
+ (void)dsize;
+ static const int npp_inter[] = {NPPI_INTER_NN, NPPI_INTER_LINEAR, NPPI_INTER_CUBIC};
+
+ NppStreamHandler h(stream);
+
+ NppiSize srcsz;
+ srcsz.height = src.rows;
+ srcsz.width = src.cols;
+ NppiRect srcroi;
+ srcroi.x = srcroi.y = 0;
+ srcroi.height = src.rows;
+ srcroi.width = src.cols;
+ NppiRect dstroi;
+ dstroi.x = dstroi.y = 0;
+ dstroi.height = dst.rows;
+ dstroi.width = dst.cols;
+
+ nppSafeCall( func(src.ptr<npp_type>(), srcsz, static_cast<int>(src.step), srcroi,
+ dst.ptr<npp_type>(), static_cast<int>(dst.step), dstroi, angle, xShift, yShift, npp_inter[interpolation]) );
+
+ if (stream == 0)
+ cudaSafeCall( cudaDeviceSynchronize() );
+ }
+ };
+}
+
+void cv::gpu::rotate(const GpuMat& src, GpuMat& dst, Size dsize, double angle, double xShift, double yShift, int interpolation, Stream& stream)
+{
+ typedef void (*func_t)(const GpuMat& src, GpuMat& dst, Size dsize, double angle, double xShift, double yShift, int interpolation, cudaStream_t stream);
+
+ static const func_t funcs[6][4] =
+ {
+ {NppRotate<CV_8U, nppiRotate_8u_C1R>::call, 0, NppRotate<CV_8U, nppiRotate_8u_C3R>::call, NppRotate<CV_8U, nppiRotate_8u_C4R>::call},
+ {0,0,0,0},
+ {NppRotate<CV_16U, nppiRotate_16u_C1R>::call, 0, NppRotate<CV_16U, nppiRotate_16u_C3R>::call, NppRotate<CV_16U, nppiRotate_16u_C4R>::call},
+ {0,0,0,0},
+ {0,0,0,0},
+ {NppRotate<CV_32F, nppiRotate_32f_C1R>::call, 0, NppRotate<CV_32F, nppiRotate_32f_C3R>::call, NppRotate<CV_32F, nppiRotate_32f_C4R>::call}
+ };
+
+ CV_Assert(src.depth() == CV_8U || src.depth() == CV_16U || src.depth() == CV_32F);
+ CV_Assert(src.channels() == 1 || src.channels() == 3 || src.channels() == 4);
+ CV_Assert(interpolation == INTER_NEAREST || interpolation == INTER_LINEAR || interpolation == INTER_CUBIC);
+
+ dst.create(dsize, src.type());
+ dst.setTo(Scalar::all(0));
+
+ funcs[src.depth()][src.channels() - 1](src, dst, dsize, angle, xShift, yShift, interpolation, StreamAccessor::getStream(stream));
}
#endif // HAVE_CUDA
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
+
+CV_GPU_TEST_MAIN("gpu")
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#include "test_precomp.hpp"
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifdef __GNUC__
+# pragma GCC diagnostic ignored "-Wmissing-declarations"
+# if defined __clang__ || defined __APPLE__
+# pragma GCC diagnostic ignored "-Wmissing-prototypes"
+# pragma GCC diagnostic ignored "-Wextra"
+# endif
+#endif
+
+#ifndef __OPENCV_TEST_PRECOMP_HPP__
+#define __OPENCV_TEST_PRECOMP_HPP__
+
+#include "opencv2/ts.hpp"
+#include "opencv2/ts/gpu_test.hpp"
+
+#include "opencv2/gpuwarping.hpp"
+#include "opencv2/imgproc.hpp"
+
+#include "interpolation.hpp"
+
+#endif
EXPECT_MAT_NEAR(dst_gold, dst, src.depth() == CV_32F ? 1e-4 : 1.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, PyrDown, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Warping, PyrDown, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatType(CV_8UC1), MatType(CV_8UC3), MatType(CV_8UC4), MatType(CV_16UC1), MatType(CV_16UC3), MatType(CV_16UC4), MatType(CV_32FC1), MatType(CV_32FC3), MatType(CV_32FC4)),
EXPECT_MAT_NEAR(dst_gold, dst, src.depth() == CV_32F ? 1e-4 : 1.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, PyrUp, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Warping, PyrUp, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatType(CV_8UC1), MatType(CV_8UC3), MatType(CV_8UC4), MatType(CV_16UC1), MatType(CV_16UC3), MatType(CV_16UC4), MatType(CV_32FC1), MatType(CV_32FC3), MatType(CV_32FC4)),
EXPECT_MAT_NEAR(dst_gold, dst, src.depth() == CV_32F ? 1e-3 : 1.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, Remap, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Warping, Remap, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatType(CV_8UC1), MatType(CV_8UC3), MatType(CV_8UC4), MatType(CV_32FC1), MatType(CV_32FC3), MatType(CV_32FC4)),
EXPECT_MAT_NEAR(dst_gold, dst, src.depth() == CV_32F ? 1e-2 : 1.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, Resize, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Warping, Resize, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatType(CV_8UC3), MatType(CV_16UC1), MatType(CV_16UC3), MatType(CV_16UC4), MatType(CV_32FC1), MatType(CV_32FC3), MatType(CV_32FC4)),
EXPECT_MAT_NEAR(dst_gold, dst, src.depth() == CV_32F ? 1e-2 : 1.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, ResizeSameAsHost, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Warping, ResizeSameAsHost, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatType(CV_8UC3), MatType(CV_16UC1), MatType(CV_16UC3), MatType(CV_16UC4), MatType(CV_32FC1), MatType(CV_32FC3), MatType(CV_32FC4)),
EXPECT_MAT_SIMILAR(dst_gold, dst, 1e-1);
}
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, ResizeNPP, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Warping, ResizeNPP, testing::Combine(
ALL_DEVICES,
testing::Values(MatType(CV_8UC1), MatType(CV_8UC4)),
testing::Values(0.3, 0.5, 1.5, 2.0),
EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, BuildWarpAffineMaps, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Warping, BuildWarpAffineMaps, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
DIRECT_INVERSE));
EXPECT_MAT_NEAR(dst_gold, dst, src.depth() == CV_32F ? 1e-1 : 1.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, WarpAffine, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Warping, WarpAffine, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatType(CV_8UC1), MatType(CV_8UC3), MatType(CV_8UC4), MatType(CV_16UC1), MatType(CV_16UC3), MatType(CV_16UC4), MatType(CV_32FC1), MatType(CV_32FC3), MatType(CV_32FC4)),
EXPECT_MAT_SIMILAR(dst_gold, dst, 2e-2);
}
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, WarpAffineNPP, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Warping, WarpAffineNPP, testing::Combine(
ALL_DEVICES,
testing::Values(MatType(CV_8UC1), MatType(CV_8UC3), MatType(CV_8UC4), MatType(CV_32FC1), MatType(CV_32FC3), MatType(CV_32FC4)),
DIRECT_INVERSE,
EXPECT_MAT_NEAR(dst_gold, dst, 0.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, BuildWarpPerspectiveMaps, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Warping, BuildWarpPerspectiveMaps, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
DIRECT_INVERSE));
EXPECT_MAT_NEAR(dst_gold, dst, src.depth() == CV_32F ? 1e-1 : 1.0);
}
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, WarpPerspective, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Warping, WarpPerspective, testing::Combine(
ALL_DEVICES,
DIFFERENT_SIZES,
testing::Values(MatType(CV_8UC1), MatType(CV_8UC3), MatType(CV_8UC4), MatType(CV_16UC1), MatType(CV_16UC3), MatType(CV_16UC4), MatType(CV_32FC1), MatType(CV_32FC3), MatType(CV_32FC4)),
EXPECT_MAT_SIMILAR(dst_gold, dst, 2e-2);
}
-INSTANTIATE_TEST_CASE_P(GPU_ImgProc, WarpPerspectiveNPP, testing::Combine(
+INSTANTIATE_TEST_CASE_P(GPU_Warping, WarpPerspectiveNPP, testing::Combine(
ALL_DEVICES,
testing::Values(MatType(CV_8UC1), MatType(CV_8UC3), MatType(CV_8UC4), MatType(CV_32FC1), MatType(CV_32FC3), MatType(CV_32FC4)),
DIRECT_INVERSE,
endif()
set(the_description "Functionality with possible limitations on the use")
-ocv_define_module(nonfree opencv_imgproc opencv_features2d opencv_calib3d OPTIONAL opencv_gpu opencv_ocl)
+ocv_warnings_disable(CMAKE_CXX_FLAGS -Wundef)
+ocv_define_module(nonfree opencv_imgproc opencv_features2d opencv_calib3d OPTIONAL opencv_gpuarithm opencv_ocl)
#ifndef __OPENCV_NONFREE_GPU_HPP__
#define __OPENCV_NONFREE_GPU_HPP__
-#include "opencv2/opencv_modules.hpp"
-
-#if defined(HAVE_OPENCV_GPU)
-
-#include "opencv2/gpu.hpp"
+#include "opencv2/core/gpumat.hpp"
namespace cv { namespace gpu {
GpuMat maxPosBuffer;
};
-} // namespace gpu
-
-} // namespace cv
-
-#endif // defined(HAVE_OPENCV_GPU)
+}} // namespace cv { namespace gpu {
#endif // __OPENCV_NONFREE_GPU_HPP__
#include "perf_precomp.hpp"
-#if defined(HAVE_OPENCV_GPU) && defined(HAVE_CUDA)
+#ifdef HAVE_CUDA
#include "opencv2/ts/gpu_perf.hpp"
//////////////////////////////////////////////////////////////////////
// SURF
+#ifdef HAVE_OPENCV_GPUARITHM
+
DEF_PARAM_TEST_1(Image, string);
PERF_TEST_P(Image, GPU_SURF,
}
}
-#endif
+#endif // HAVE_OPENCV_GPUARITHM
+
+#endif // HAVE_CUDA
# include "opencv2/nonfree/ocl.hpp"
#endif
-#ifdef HAVE_OPENCV_GPU
- #include "opencv2/nonfree/gpu.hpp"
- #include "opencv2/ts/gpu_perf.hpp"
+#ifdef HAVE_CUDA
+# include "opencv2/nonfree/gpu.hpp"
#endif
#ifdef GTEST_CREATE_SHARED_LIBRARY
#include "opencv2/opencv_modules.hpp"
-#ifdef HAVE_OPENCV_GPU
+#ifdef HAVE_OPENCV_GPUARITHM
#include "opencv2/core/cuda/common.hpp"
#include "opencv2/core/cuda/limits.hpp"
} // namespace surf
}}} // namespace cv { namespace gpu { namespace cudev
-
-#endif /* CUDA_DISABLER */
+#endif // HAVE_OPENCV_GPUARITHM
#include "opencv2/core/utility.hpp"
#include "opencv2/core/private.hpp"
+#include "opencv2/nonfree/gpu.hpp"
+#include "opencv2/core/gpu_private.hpp"
+
#include "opencv2/opencv_modules.hpp"
-#ifdef HAVE_OPENCV_GPU
-# include "opencv2/nonfree/gpu.hpp"
-# include "opencv2/core/gpu_private.hpp"
+#ifdef HAVE_OPENCV_GPUARITHM
+# include "opencv2/gpuarithm.hpp"
#endif
#ifdef HAVE_OPENCV_OCL
#include "precomp.hpp"
-#if defined(HAVE_OPENCV_GPU)
-
using namespace cv;
using namespace cv::gpu;
-#if !defined (HAVE_CUDA)
+#if !defined (HAVE_CUDA) || !defined (HAVE_OPENCV_GPUARITHM)
cv::gpu::SURF_GPU::SURF_GPU() { throw_no_cuda(); }
cv::gpu::SURF_GPU::SURF_GPU(double, int, int, bool, float, bool) { throw_no_cuda(); }
bindImgTex(img);
- integralBuffered(img, surf_.sum, surf_.intBuffer);
+ gpu::integralBuffered(img, surf_.sum, surf_.intBuffer);
sumOffset = bindSumTex(surf_.sum);
if (use_mask)
{
min(mask, 1.0, surf_.mask1);
- integralBuffered(surf_.mask1, surf_.maskSum, surf_.intBuffer);
+ gpu::integralBuffered(surf_.mask1, surf_.maskSum, surf_.intBuffer);
maskOffset = bindMaskSumTex(surf_.maskSum);
}
}
}
#endif // !defined (HAVE_CUDA)
-
-#endif // defined(HAVE_OPENCV_GPU)
#include "test_precomp.hpp"
-#if defined(HAVE_OPENCV_GPU) && defined(HAVE_CUDA)
+#ifdef HAVE_CUDA
using namespace cvtest;
/////////////////////////////////////////////////////////////////////////////////////////////////
// SURF
+#ifdef HAVE_OPENCV_GPUARITHM
+
namespace
{
IMPLEMENT_PARAM_CLASS(SURF_HessianThreshold, double)
testing::Values(SURF_Extended(false), SURF_Extended(true)),
testing::Values(SURF_Upright(false), SURF_Upright(true))));
-#endif
+#endif // HAVE_OPENCV_GPUARITHM
+
+#endif // HAVE_CUDA
# include "opencv2/nonfree/ocl.hpp"
#endif
-#ifdef HAVE_OPENCV_GPU
+#ifdef HAVE_CUDA
# include "opencv2/nonfree/gpu.hpp"
#endif
set(the_description "Computational Photography")
-ocv_define_module(photo opencv_imgproc)
+
+if(HAVE_CUDA)
+ ocv_warnings_disable(CMAKE_CXX_FLAGS -Wundef -Wmissing-declarations)
+endif()
+
+ocv_define_module(photo opencv_imgproc OPTIONAL opencv_gpuarithm opencv_gpuimgproc)
The function converts images to CIELAB colorspace and then separately denoise L and AB components with given h parameters using ``fastNlMeansDenoisingMulti`` function.
+
+
+gpu::nonLocalMeans
+-------------------
+Performs pure non local means denoising without any simplification, and thus it is not fast.
+
+.. ocv:function:: void gpu::nonLocalMeans(const GpuMat& src, GpuMat& dst, float h, int search_window = 21, int block_size = 7, int borderMode = BORDER_DEFAULT, Stream& s = Stream::Null())
+
+ :param src: Source image. Supports only CV_8UC1, CV_8UC2 and CV_8UC3.
+
+ :param dst: Destination image.
+
+ :param h: Filter sigma regulating filter strength for color.
+
+ :param search_window: Size of search window.
+
+ :param block_size: Size of block used for computing weights.
+
+ :param borderMode: Border type. See :ocv:func:`borderInterpolate` for details. ``BORDER_REFLECT101`` , ``BORDER_REPLICATE`` , ``BORDER_CONSTANT`` , ``BORDER_REFLECT`` and ``BORDER_WRAP`` are supported for now.
+
+ :param stream: Stream for the asynchronous version.
+
+.. seealso::
+
+ :ocv:func:`fastNlMeansDenoising`
+
+
+
+gpu::FastNonLocalMeansDenoising
+-------------------------------
+.. ocv:class:: gpu::FastNonLocalMeansDenoising
+
+ ::
+
+ class FastNonLocalMeansDenoising
+ {
+ public:
+ //! Simple method, recommended for grayscale images (though it supports multichannel images)
+ void simpleMethod(const GpuMat& src, GpuMat& dst, float h, int search_window = 21, int block_size = 7, Stream& s = Stream::Null())
+ //! Processes luminance and color components separatelly
+ void labMethod(const GpuMat& src, GpuMat& dst, float h_luminance, float h_color, int search_window = 21, int block_size = 7, Stream& s = Stream::Null())
+ };
+
+The class implements fast approximate Non Local Means Denoising algorithm.
+
+
+
+gpu::FastNonLocalMeansDenoising::simpleMethod()
+-----------------------------------------------
+Perform image denoising using Non-local Means Denoising algorithm http://www.ipol.im/pub/algo/bcm_non_local_means_denoising with several computational optimizations. Noise expected to be a gaussian white noise
+
+.. ocv:function:: void gpu::FastNonLocalMeansDenoising::simpleMethod(const GpuMat& src, GpuMat& dst, float h, int search_window = 21, int block_size = 7, Stream& s = Stream::Null())
+
+ :param src: Input 8-bit 1-channel, 2-channel or 3-channel image.
+
+ :param dst: Output image with the same size and type as ``src`` .
+
+ :param h: Parameter regulating filter strength. Big h value perfectly removes noise but also removes image details, smaller h value preserves details but also preserves some noise
+
+ :param search_window: Size in pixels of the window that is used to compute weighted average for given pixel. Should be odd. Affect performance linearly: greater search_window - greater denoising time. Recommended value 21 pixels
+
+ :param block_size: Size in pixels of the template patch that is used to compute weights. Should be odd. Recommended value 7 pixels
+
+ :param stream: Stream for the asynchronous invocations.
+
+This function expected to be applied to grayscale images. For colored images look at ``FastNonLocalMeansDenoising::labMethod``.
+
+.. seealso::
+
+ :ocv:func:`fastNlMeansDenoising`
+
+
+
+gpu::FastNonLocalMeansDenoising::labMethod()
+--------------------------------------------
+Modification of ``FastNonLocalMeansDenoising::simpleMethod`` for color images
+
+.. ocv:function:: void gpu::FastNonLocalMeansDenoising::labMethod(const GpuMat& src, GpuMat& dst, float h_luminance, float h_color, int search_window = 21, int block_size = 7, Stream& s = Stream::Null())
+
+ :param src: Input 8-bit 3-channel image.
+
+ :param dst: Output image with the same size and type as ``src`` .
+
+ :param h_luminance: Parameter regulating filter strength. Big h value perfectly removes noise but also removes image details, smaller h value preserves details but also preserves some noise
+
+ :param float: The same as h but for color components. For most images value equals 10 will be enought to remove colored noise and do not distort colors
+
+ :param search_window: Size in pixels of the window that is used to compute weighted average for given pixel. Should be odd. Affect performance linearly: greater search_window - greater denoising time. Recommended value 21 pixels
+
+ :param block_size: Size in pixels of the template patch that is used to compute weights. Should be odd. Recommended value 7 pixels
+
+ :param stream: Stream for the asynchronous invocations.
+
+The function converts image to CIELAB colorspace and then separately denoise L and AB components with given h parameters using ``FastNonLocalMeansDenoising::simpleMethod`` function.
+
+.. seealso::
+
+ :ocv:func:`fastNlMeansDenoisingColored`
+
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+// By downloading, copying, installing or using the software you agree to this license.
+// If you do not agree to this license, do not download, install,
+// copy or use the software.
+//
+//
+// License Agreement
+// For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2008-2012, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistribution's of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+//
+// * Redistribution's in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation
+// and/or other materials provided with the distribution.
+//
+// * The name of the copyright holders may not be used to endorse or promote products
+// derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_PHOTO_GPU_HPP__
+#define __OPENCV_PHOTO_GPU_HPP__
+
+#include "opencv2/core/gpumat.hpp"
+
+namespace cv { namespace gpu {
+
+//! Brute force non-local means algorith (slow but universal)
+CV_EXPORTS void nonLocalMeans(const GpuMat& src, GpuMat& dst, float h, int search_window = 21, int block_size = 7, int borderMode = BORDER_DEFAULT, Stream& s = Stream::Null());
+
+//! Fast (but approximate)version of non-local means algorith similar to CPU function (running sums technique)
+class CV_EXPORTS FastNonLocalMeansDenoising
+{
+public:
+ //! Simple method, recommended for grayscale images (though it supports multichannel images)
+ void simpleMethod(const GpuMat& src, GpuMat& dst, float h, int search_window = 21, int block_size = 7, Stream& s = Stream::Null());
+
+ //! Processes luminance and color components separatelly
+ void labMethod(const GpuMat& src, GpuMat& dst, float h_luminance, float h_color, int search_window = 21, int block_size = 7, Stream& s = Stream::Null());
+
+private:
+
+ GpuMat buffer, extended_src_buffer;
+ GpuMat lab, l, ab;
+};
+
+}} // namespace cv { namespace gpu {
+
+#endif /* __OPENCV_PHOTO_GPU_HPP__ */
#include "perf_precomp.hpp"
+#include "opencv2/photo/gpu.hpp"
+#include "opencv2/ts/gpu_perf.hpp"
+
+#include "opencv2/opencv_modules.hpp"
+
+#if defined (HAVE_CUDA) && defined(HAVE_OPENCV_GPUARITHM) && defined(HAVE_OPENCV_GPUIMGPROC)
+
using namespace std;
using namespace testing;
using namespace perf;
#define GPU_DENOISING_IMAGE_SIZES testing::Values(perf::szVGA, perf::sz720p)
//////////////////////////////////////////////////////////////////////
-// BilateralFilter
-
-DEF_PARAM_TEST(Sz_Depth_Cn_KernelSz, cv::Size, MatDepth, MatCn, int);
-
-PERF_TEST_P(Sz_Depth_Cn_KernelSz, Denoising_BilateralFilter,
- Combine(GPU_DENOISING_IMAGE_SIZES,
- Values(CV_8U, CV_32F),
- GPU_CHANNELS_1_3,
- Values(3, 5, 9)))
-{
- declare.time(60.0);
-
- const cv::Size size = GET_PARAM(0);
- const int depth = GET_PARAM(1);
- const int channels = GET_PARAM(2);
- const int kernel_size = GET_PARAM(3);
-
- const float sigma_color = 7;
- const float sigma_spatial = 5;
- const int borderMode = cv::BORDER_REFLECT101;
-
- const int type = CV_MAKE_TYPE(depth, channels);
-
- cv::Mat src(size, type);
- declare.in(src, WARMUP_RNG);
-
- if (PERF_RUN_GPU())
- {
- const cv::gpu::GpuMat d_src(src);
- cv::gpu::GpuMat dst;
-
- TEST_CYCLE() cv::gpu::bilateralFilter(d_src, dst, kernel_size, sigma_color, sigma_spatial, borderMode);
-
- GPU_SANITY_CHECK(dst);
- }
- else
- {
- cv::Mat dst;
-
- TEST_CYCLE() cv::bilateralFilter(src, dst, kernel_size, sigma_color, sigma_spatial, borderMode);
-
- CPU_SANITY_CHECK(dst);
- }
-}
-
-//////////////////////////////////////////////////////////////////////
// nonLocalMeans
DEF_PARAM_TEST(Sz_Depth_Cn_WinSz_BlockSz, cv::Size, MatDepth, MatCn, int, int);
-PERF_TEST_P(Sz_Depth_Cn_WinSz_BlockSz, Denoising_NonLocalMeans,
+PERF_TEST_P(Sz_Depth_Cn_WinSz_BlockSz, GPU_NonLocalMeans,
Combine(GPU_DENOISING_IMAGE_SIZES,
Values<MatDepth>(CV_8U),
GPU_CHANNELS_1_3,
DEF_PARAM_TEST(Sz_Depth_Cn_WinSz_BlockSz, cv::Size, MatDepth, MatCn, int, int);
-PERF_TEST_P(Sz_Depth_Cn_WinSz_BlockSz, Denoising_FastNonLocalMeans,
+PERF_TEST_P(Sz_Depth_Cn_WinSz_BlockSz, GPU_FastNonLocalMeans,
Combine(GPU_DENOISING_IMAGE_SIZES,
Values<MatDepth>(CV_8U),
GPU_CHANNELS_1_3,
DEF_PARAM_TEST(Sz_Depth_WinSz_BlockSz, cv::Size, MatDepth, int, int);
-PERF_TEST_P(Sz_Depth_WinSz_BlockSz, Denoising_FastNonLocalMeansColored,
+PERF_TEST_P(Sz_Depth_WinSz_BlockSz, GPU_FastNonLocalMeansColored,
Combine(GPU_DENOISING_IMAGE_SIZES,
Values<MatDepth>(CV_8U),
Values(21),
CPU_SANITY_CHECK(dst);
}
}
+
+#endif
//
//M*/
-#if !defined CUDA_DISABLER
-
#include "opencv2/core/cuda/common.hpp"
#include "opencv2/core/cuda/vec_traits.hpp"
#include "opencv2/core/cuda/vec_math.hpp"
static func_t funcs[] =
{
- nlm_caller<T, BrdReflect101>,
- nlm_caller<T, BrdReplicate>,
nlm_caller<T, BrdConstant>,
+ nlm_caller<T, BrdReplicate>,
nlm_caller<T, BrdReflect>,
nlm_caller<T, BrdWrap>,
+ nlm_caller<T, BrdReflect101>
};
funcs[borderMode](src, dst, search_radius, block_radius, h, stream);
}
}
}
}}}
-
-
-#endif /* CUDA_DISABLER */
#include "precomp.hpp"
+#include "opencv2/photo/gpu.hpp"
+#include "opencv2/core/gpu_private.hpp"
+
+#include "opencv2/opencv_modules.hpp"
+
+#ifdef HAVE_OPENCV_GPUARITHM
+# include "opencv2/gpuarithm.hpp"
+#endif
+
+#ifdef HAVE_OPENCV_GPUIMGPROC
+# include "opencv2/gpuimgproc.hpp"
+#endif
+
using namespace cv;
using namespace cv::gpu;
-#if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
+#if !defined (HAVE_CUDA) || !defined(HAVE_OPENCV_GPUARITHM) || !defined(HAVE_OPENCV_GPUIMGPROC)
-void cv::gpu::bilateralFilter(const GpuMat&, GpuMat&, int, float, float, int, Stream&) { throw_no_cuda(); }
void cv::gpu::nonLocalMeans(const GpuMat&, GpuMat&, float, int, int, int, Stream&) { throw_no_cuda(); }
-
void cv::gpu::FastNonLocalMeansDenoising::simpleMethod(const GpuMat&, GpuMat&, float, int, int, Stream&) { throw_no_cuda(); }
void cv::gpu::FastNonLocalMeansDenoising::labMethod( const GpuMat&, GpuMat&, float, float, int, int, Stream&) { throw_no_cuda(); }
-
#else
//////////////////////////////////////////////////////////////////////////////////
namespace imgproc
{
template<typename T>
- void bilateral_filter_gpu(const PtrStepSzb& src, PtrStepSzb dst, int kernel_size, float sigma_spatial, float sigma_color, int borderMode, cudaStream_t stream);
-
- template<typename T>
void nlm_bruteforce_gpu(const PtrStepSzb& src, PtrStepSzb dst, int search_radius, int block_radius, float h, int borderMode, cudaStream_t stream);
}
}}}
-void cv::gpu::bilateralFilter(const GpuMat& src, GpuMat& dst, int kernel_size, float sigma_color, float sigma_spatial, int borderMode, Stream& s)
-{
- using cv::gpu::cudev::imgproc::bilateral_filter_gpu;
-
- typedef void (*func_t)(const PtrStepSzb& src, PtrStepSzb dst, int kernel_size, float sigma_spatial, float sigma_color, int borderMode, cudaStream_t s);
-
- static const func_t funcs[6][4] =
- {
- {bilateral_filter_gpu<uchar> , 0 /*bilateral_filter_gpu<uchar2>*/ , bilateral_filter_gpu<uchar3> , bilateral_filter_gpu<uchar4> },
- {0 /*bilateral_filter_gpu<schar>*/, 0 /*bilateral_filter_gpu<schar2>*/ , 0 /*bilateral_filter_gpu<schar3>*/, 0 /*bilateral_filter_gpu<schar4>*/},
- {bilateral_filter_gpu<ushort> , 0 /*bilateral_filter_gpu<ushort2>*/, bilateral_filter_gpu<ushort3> , bilateral_filter_gpu<ushort4> },
- {bilateral_filter_gpu<short> , 0 /*bilateral_filter_gpu<short2>*/ , bilateral_filter_gpu<short3> , bilateral_filter_gpu<short4> },
- {0 /*bilateral_filter_gpu<int>*/ , 0 /*bilateral_filter_gpu<int2>*/ , 0 /*bilateral_filter_gpu<int3>*/ , 0 /*bilateral_filter_gpu<int4>*/ },
- {bilateral_filter_gpu<float> , 0 /*bilateral_filter_gpu<float2>*/ , bilateral_filter_gpu<float3> , bilateral_filter_gpu<float4> }
- };
-
- sigma_color = (sigma_color <= 0 ) ? 1 : sigma_color;
- sigma_spatial = (sigma_spatial <= 0 ) ? 1 : sigma_spatial;
-
-
- int radius = (kernel_size <= 0) ? cvRound(sigma_spatial*1.5) : kernel_size/2;
- kernel_size = std::max(radius, 1)*2 + 1;
-
- CV_Assert(src.depth() <= CV_32F && src.channels() <= 4);
- const func_t func = funcs[src.depth()][src.channels() - 1];
- CV_Assert(func != 0);
-
- CV_Assert(borderMode == BORDER_REFLECT101 || borderMode == BORDER_REPLICATE || borderMode == BORDER_CONSTANT || borderMode == BORDER_REFLECT || borderMode == BORDER_WRAP);
-
- int gpuBorderType;
- CV_Assert(tryConvertToGpuBorderType(borderMode, gpuBorderType));
-
- dst.create(src.size(), src.type());
- func(src, dst, kernel_size, sigma_spatial, sigma_color, gpuBorderType, StreamAccessor::getStream(s));
-}
-
void cv::gpu::nonLocalMeans(const GpuMat& src, GpuMat& dst, float h, int search_window, int block_window, int borderMode, Stream& s)
{
using cv::gpu::cudev::imgproc::nlm_bruteforce_gpu;
int b = borderMode;
CV_Assert(b == BORDER_REFLECT101 || b == BORDER_REPLICATE || b == BORDER_CONSTANT || b == BORDER_REFLECT || b == BORDER_WRAP);
- int gpuBorderType;
- CV_Assert(tryConvertToGpuBorderType(borderMode, gpuBorderType));
-
dst.create(src.size(), src.type());
- func(src, dst, search_window/2, block_window/2, h, gpuBorderType, StreamAccessor::getStream(s));
+ func(src, dst, search_window/2, block_window/2, h, borderMode, StreamAccessor::getStream(s));
}
-
-//////////////////////////////////////////////////////////////////////////////////
-//// Non Local Means Denosing (fast approxinate)
-
-
namespace cv { namespace gpu { namespace cudev
{
namespace imgproc
cv::gpu::cvtColor(lab, dst, cv::COLOR_Lab2BGR, 0, s);
}
-
#endif
-
-
#include "test_precomp.hpp"
-#ifdef HAVE_CUDA
+#include "opencv2/photo/gpu.hpp"
+#include "opencv2/ts/gpu_test.hpp"
-using namespace cvtest;
-
-////////////////////////////////////////////////////////
-// BilateralFilter
-
-PARAM_TEST_CASE(BilateralFilter, cv::gpu::DeviceInfo, cv::Size, MatType)
-{
- cv::gpu::DeviceInfo devInfo;
- cv::Size size;
- int type;
- int kernel_size;
- float sigma_color;
- float sigma_spatial;
-
- virtual void SetUp()
- {
- devInfo = GET_PARAM(0);
- size = GET_PARAM(1);
- type = GET_PARAM(2);
-
- kernel_size = 5;
- sigma_color = 10.f;
- sigma_spatial = 3.5f;
-
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(BilateralFilter, Accuracy)
-{
- cv::Mat src = randomMat(size, type);
-
- src.convertTo(src, type);
- cv::gpu::GpuMat dst;
-
- cv::gpu::bilateralFilter(loadMat(src), dst, kernel_size, sigma_color, sigma_spatial);
-
- cv::Mat dst_gold;
- cv::bilateralFilter(src, dst_gold, kernel_size, sigma_color, sigma_spatial);
-
- EXPECT_MAT_NEAR(dst_gold, dst, src.depth() == CV_32F ? 1e-3 : 1.0);
-}
+#include "opencv2/opencv_modules.hpp"
-INSTANTIATE_TEST_CASE_P(GPU_Denoising, BilateralFilter, testing::Combine(
- ALL_DEVICES,
- testing::Values(cv::Size(128, 128), cv::Size(113, 113), cv::Size(639, 481)),
- testing::Values(MatType(CV_8UC1), MatType(CV_8UC3), MatType(CV_32FC1), MatType(CV_32FC3))
- ));
+#if defined (HAVE_CUDA) && defined(HAVE_OPENCV_GPUARITHM) && defined(HAVE_OPENCV_GPUIMGPROC)
+using namespace cvtest;
////////////////////////////////////////////////////////
// Brute Force Non local means
-struct BruteForceNonLocalMeans: testing::TestWithParam<cv::gpu::DeviceInfo>
-{
- cv::gpu::DeviceInfo devInfo;
-
- virtual void SetUp()
- {
- devInfo = GetParam();
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(BruteForceNonLocalMeans, Regression)
+TEST(GPU_BruteForceNonLocalMeans, Regression)
{
using cv::gpu::GpuMat;
- cv::Mat bgr = readImage("denoising/lena_noised_gaussian_sigma=20_multi_0.png", cv::IMREAD_COLOR);
+ cv::Mat bgr = readImage("../gpu/denoising/lena_noised_gaussian_sigma=20_multi_0.png", cv::IMREAD_COLOR);
ASSERT_FALSE(bgr.empty());
cv::Mat gray;
cv::gpu::nonLocalMeans(GpuMat(gray), dgray, 20);
#if 0
- dumpImage("denoising/nlm_denoised_lena_bgr.png", cv::Mat(dbgr));
- dumpImage("denoising/nlm_denoised_lena_gray.png", cv::Mat(dgray));
+ dumpImage("../gpu/denoising/nlm_denoised_lena_bgr.png", cv::Mat(dbgr));
+ dumpImage("../gpu/denoising/nlm_denoised_lena_gray.png", cv::Mat(dgray));
#endif
- cv::Mat bgr_gold = readImage("denoising/nlm_denoised_lena_bgr.png", cv::IMREAD_COLOR);
- cv::Mat gray_gold = readImage("denoising/nlm_denoised_lena_gray.png", cv::IMREAD_GRAYSCALE);
+ cv::Mat bgr_gold = readImage("../gpu/denoising/nlm_denoised_lena_bgr.png", cv::IMREAD_COLOR);
+ cv::Mat gray_gold = readImage("../gpu/denoising/nlm_denoised_lena_gray.png", cv::IMREAD_GRAYSCALE);
ASSERT_FALSE(bgr_gold.empty() || gray_gold.empty());
EXPECT_MAT_NEAR(bgr_gold, dbgr, 1e-4);
EXPECT_MAT_NEAR(gray_gold, dgray, 1e-4);
}
-INSTANTIATE_TEST_CASE_P(GPU_Denoising, BruteForceNonLocalMeans, ALL_DEVICES);
-
////////////////////////////////////////////////////////
// Fast Force Non local means
-struct FastNonLocalMeans: testing::TestWithParam<cv::gpu::DeviceInfo>
-{
- cv::gpu::DeviceInfo devInfo;
-
- virtual void SetUp()
- {
- devInfo = GetParam();
- cv::gpu::setDevice(devInfo.deviceID());
- }
-};
-
-GPU_TEST_P(FastNonLocalMeans, Regression)
+TEST(GPU_FastNonLocalMeans, Regression)
{
using cv::gpu::GpuMat;
- cv::Mat bgr = readImage("denoising/lena_noised_gaussian_sigma=20_multi_0.png", cv::IMREAD_COLOR);
+ cv::Mat bgr = readImage("../gpu/denoising/lena_noised_gaussian_sigma=20_multi_0.png", cv::IMREAD_COLOR);
ASSERT_FALSE(bgr.empty());
cv::Mat gray;
fnlmd.labMethod(GpuMat(bgr), dbgr, 20, 10);
#if 0
- dumpImage("denoising/fnlm_denoised_lena_bgr.png", cv::Mat(dbgr));
- dumpImage("denoising/fnlm_denoised_lena_gray.png", cv::Mat(dgray));
+ dumpImage("../gpu/denoising/fnlm_denoised_lena_bgr.png", cv::Mat(dbgr));
+ dumpImage("../gpu/denoising/fnlm_denoised_lena_gray.png", cv::Mat(dgray));
#endif
- cv::Mat bgr_gold = readImage("denoising/fnlm_denoised_lena_bgr.png", cv::IMREAD_COLOR);
- cv::Mat gray_gold = readImage("denoising/fnlm_denoised_lena_gray.png", cv::IMREAD_GRAYSCALE);
+ cv::Mat bgr_gold = readImage("../gpu/denoising/fnlm_denoised_lena_bgr.png", cv::IMREAD_COLOR);
+ cv::Mat gray_gold = readImage("../gpu/denoising/fnlm_denoised_lena_gray.png", cv::IMREAD_GRAYSCALE);
ASSERT_FALSE(bgr_gold.empty() || gray_gold.empty());
EXPECT_MAT_NEAR(bgr_gold, dbgr, 1);
EXPECT_MAT_NEAR(gray_gold, dgray, 1);
}
-INSTANTIATE_TEST_CASE_P(GPU_Denoising, FastNonLocalMeans, ALL_DEVICES);
-
#endif // HAVE_CUDA
flds.createLevels(image.rows, image.cols);
flds.preprocessor->apply(image, flds.shrunk);
- integral(flds.shrunk, flds.hogluv, flds.integralBuffer, s);
+ ::integral(flds.shrunk, flds.hogluv, flds.integralBuffer, s);
}
else
{
set(the_description "Images stitching")
-ocv_define_module(stitching opencv_imgproc opencv_features2d opencv_calib3d opencv_objdetect OPTIONAL opencv_gpu opencv_nonfree)
+ocv_define_module(stitching opencv_imgproc opencv_features2d opencv_calib3d opencv_objdetect
+ OPTIONAL opencv_gpu opencv_gpuarithm opencv_gpufilters opencv_gpufeatures2d opencv_nonfree)
#include "opencv2/features2d.hpp"
#include "opencv2/opencv_modules.hpp"
-#if defined(HAVE_OPENCV_NONFREE) && defined(HAVE_OPENCV_GPU)
-#include "opencv2/nonfree/gpu.hpp"
+
+#ifdef HAVE_OPENCV_NONFREE
+# include "opencv2/nonfree/gpu.hpp"
#endif
namespace cv {
};
-#if defined(HAVE_OPENCV_NONFREE) && defined(HAVE_OPENCV_GPU)
+#ifdef HAVE_OPENCV_NONFREE
class CV_EXPORTS SurfFeaturesFinderGpu : public FeaturesFinder
{
public:
#define __OPENCV_STITCHING_WARPERS_HPP__
#include "opencv2/core.hpp"
+#include "opencv2/core/gpumat.hpp"
#include "opencv2/imgproc.hpp"
#include "opencv2/opencv_modules.hpp"
-#ifdef HAVE_OPENCV_GPU
-# include "opencv2/gpu.hpp"
-#endif
namespace cv {
namespace detail {
};
-#ifdef HAVE_OPENCV_GPU
+#ifdef HAVE_OPENCV_GPUWARPING
class CV_EXPORTS PlaneWarperGpu : public PlaneWarper
{
public:
-#ifdef HAVE_OPENCV_GPU
+#ifdef HAVE_OPENCV_GPUWARPING
class PlaneWarperGpu: public WarperCreator
{
public:
MultiBandBlender::MultiBandBlender(int try_gpu, int num_bands, int weight_type)
{
setNumBands(num_bands);
-#ifdef HAVE_OPENCV_GPU
+
+#if defined(HAVE_OPENCV_GPUARITHM) && defined(HAVE_OPENCV_GPUWARPING)
can_use_gpu_ = try_gpu && gpu::getCudaEnabledDeviceCount();
#else
- (void)try_gpu;
+ (void) try_gpu;
can_use_gpu_ = false;
#endif
+
CV_Assert(weight_type == CV_32F || weight_type == CV_16S);
weight_type_ = weight_type;
}
void createLaplacePyrGpu(const Mat &img, int num_levels, std::vector<Mat> &pyr)
{
-#ifdef HAVE_OPENCV_GPU
+#if defined(HAVE_OPENCV_GPUARITHM) && defined(HAVE_OPENCV_GPUWARPING)
pyr.resize(num_levels + 1);
std::vector<gpu::GpuMat> gpu_pyr(num_levels + 1);
void restoreImageFromLaplacePyrGpu(std::vector<Mat> &pyr)
{
-#ifdef HAVE_OPENCV_GPU
+#if defined(HAVE_OPENCV_GPUARITHM) && defined(HAVE_OPENCV_GPUWARPING)
if (pyr.empty())
return;
using namespace cv;
using namespace cv::detail;
-
-#ifdef HAVE_OPENCV_GPU
using namespace cv::gpu;
-#endif
#ifdef HAVE_OPENCV_NONFREE
#include "opencv2/nonfree.hpp"
float match_conf_;
};
-#ifdef HAVE_OPENCV_GPU
+#ifdef HAVE_OPENCV_GPUFEATURES2D
class GpuMatcher : public FeaturesMatcher
{
public:
LOG("1->2 & 2->1 matches: " << matches_info.matches.size() << endl);
}
-#ifdef HAVE_OPENCV_GPU
+#ifdef HAVE_OPENCV_GPUFEATURES2D
void GpuMatcher::match(const ImageFeatures &features1, const ImageFeatures &features2, MatchesInfo& matches_info)
{
matches_info.matches.clear();
}
}
-#if defined(HAVE_OPENCV_NONFREE) && defined(HAVE_OPENCV_GPU)
+#ifdef HAVE_OPENCV_NONFREE
SurfFeaturesFinderGpu::SurfFeaturesFinderGpu(double hess_thresh, int num_octaves, int num_layers,
int num_octaves_descr, int num_layers_descr)
{
BestOf2NearestMatcher::BestOf2NearestMatcher(bool try_use_gpu, float match_conf, int num_matches_thresh1, int num_matches_thresh2)
{
-#ifdef HAVE_OPENCV_GPU
+ (void)try_use_gpu;
+
+#ifdef HAVE_OPENCV_GPUFEATURES2D
if (try_use_gpu && getCudaEnabledDeviceCount() > 0)
+ {
impl_ = new GpuMatcher(match_conf);
+ }
else
-#else
- (void)try_use_gpu;
#endif
+ {
impl_ = new CpuMatcher(match_conf);
+ }
is_thread_safe_ = impl_->isThreadSafe();
num_matches_thresh1_ = num_matches_thresh1;
#include "opencv2/imgproc.hpp"
#include "opencv2/features2d.hpp"
#include "opencv2/calib3d.hpp"
+
+#ifdef HAVE_OPENCV_GPUARITHM
+# include "opencv2/gpuarithm.hpp"
+#endif
+
+#ifdef HAVE_OPENCV_GPUWARPING
+# include "opencv2/gpuwarping.hpp"
+#endif
+
+#ifdef HAVE_OPENCV_GPUFEATURES2D
+# include "opencv2/gpufeatures2d.hpp"
+#endif
+
#ifdef HAVE_OPENCV_GPU
# include "opencv2/gpu.hpp"
-# ifdef HAVE_OPENCV_NONFREE
-# include "opencv2/nonfree/gpu.hpp"
-# endif
+#endif
+
+#ifdef HAVE_OPENCV_NONFREE
+# include "opencv2/nonfree/gpu.hpp"
#endif
#include "../../imgproc/src/gcgraph.hpp"
#ifdef HAVE_OPENCV_GPU
if (try_use_gpu && gpu::getCudaEnabledDeviceCount() > 0)
{
-#if defined(HAVE_OPENCV_NONFREE)
+#ifdef HAVE_OPENCV_NONFREE
stitcher.setFeaturesFinder(new detail::SurfFeaturesFinderGpu());
#else
stitcher.setFeaturesFinder(new detail::OrbFeaturesFinder());
}
-#ifdef HAVE_OPENCV_GPU
+#ifdef HAVE_OPENCV_GPUWARPING
Rect PlaneWarperGpu::buildMaps(Size src_size, const Mat &K, const Mat &R, gpu::GpuMat &xmap, gpu::GpuMat &ymap)
{
return buildMaps(src_size, K, R, Mat::zeros(3, 1, CV_32F), xmap, ymap);
set(the_description "Super Resolution")
ocv_warnings_disable(CMAKE_CXX_FLAGS /wd4127 -Wundef)
-ocv_define_module(superres opencv_imgproc opencv_video OPTIONAL opencv_gpu opencv_highgui)
+ocv_define_module(superres opencv_imgproc opencv_video
+ OPTIONAL opencv_highgui
+ opencv_gpuarithm opencv_gpufilters opencv_gpuwarping opencv_gpuimgproc opencv_gpuoptflow opencv_gpucodec)
using namespace cv::superres;
using namespace cv::superres::detail;
-#if !defined(HAVE_CUDA) || !defined(HAVE_OPENCV_GPU)
+#if !defined(HAVE_CUDA) || !defined(HAVE_OPENCV_GPUARITHM) || !defined(HAVE_OPENCV_GPUWARPING) || !defined(HAVE_OPENCV_GPUFILTERS)
Ptr<SuperResolution> cv::superres::createSuperResolution_BTVL1_GPU()
{
btvKernelSize_ = 7;
blurKernelSize_ = 5;
blurSigma_ = 0.0;
+
+#ifdef HAVE_OPENCV_GPUOPTFLOW
opticalFlow_ = createOptFlow_Farneback_GPU();
+#else
+ opticalFlow_ = createOptFlow_Farneback();
+#endif
curBlurKernelSize_ = -1;
curBlurSigma_ = -1.0;
//
//M*/
+#include "opencv2/opencv_modules.hpp"
+
+#if defined(HAVE_OPENCV_GPUARITHM) && defined(HAVE_OPENCV_GPUWARPING) && defined(HAVE_OPENCV_GPUFILTERS)
+
#include "opencv2/core/cuda/common.hpp"
#include "opencv2/core/cuda/transform.hpp"
#include "opencv2/core/cuda/vec_traits.hpp"
template void calcBtvRegularization<3>(PtrStepSzb src, PtrStepSzb dst, int ksize);
template void calcBtvRegularization<4>(PtrStepSzb src, PtrStepSzb dst, int ksize);
}
+
+#endif
//////////////////////////////////////////////////////
// VideoFrameSource_GPU
-#ifndef HAVE_OPENCV_GPU
+#ifndef HAVE_OPENCV_GPUCODEC
Ptr<FrameSource> cv::superres::createFrameSource_Video_GPU(const String& fileName)
{
return Ptr<FrameSource>();
}
-#else // HAVE_OPENCV_GPU
+#else // HAVE_OPENCV_GPUCODEC
namespace
{
return new VideoFrameSource(fileName);
}
-#endif // HAVE_OPENCV_GPU
+#endif // HAVE_OPENCV_GPUCODEC
switch (src.kind())
{
case _InputArray::GPU_MAT:
- #ifdef HAVE_OPENCV_GPU
+ #ifdef HAVE_OPENCV_GPUIMGPROC
gpu::cvtColor(src.getGpuMat(), dst.getGpuMatRef(), code, cn);
#else
CV_Error(cv::Error::StsNotImplemented, "The called functionality is disabled for current build or platform");
///////////////////////////////////////////////////////////////////
// GpuOpticalFlow
-#ifndef HAVE_OPENCV_GPU
+#ifndef HAVE_OPENCV_GPUOPTFLOW
Ptr<DenseOpticalFlowExt> cv::superres::createOptFlow_Farneback_GPU()
{
return Ptr<DenseOpticalFlowExt>();
}
-#else // HAVE_OPENCV_GPU
+#else // HAVE_OPENCV_GPUOPTFLOW
namespace
{
return new DualTVL1_GPU;
}
-#endif // HAVE_OPENCV_GPU
+#endif // HAVE_OPENCV_GPUOPTFLOW
#include "opencv2/video/tracking.hpp"
#include "opencv2/core/private.hpp"
-#ifdef HAVE_OPENCV_GPU
-# include "opencv2/gpu.hpp"
-# include "opencv2/core/gpu_private.hpp"
+#include "opencv2/core/gpu_private.hpp"
+
+#ifdef HAVE_OPENCV_GPUARITHM
+# include "opencv2/gpuarithm.hpp"
+#endif
+
+#ifdef HAVE_OPENCV_GPUWARPING
+# include "opencv2/gpuwarping.hpp"
+#endif
+
+#ifdef HAVE_OPENCV_GPUFILTERS
+# include "opencv2/gpufilters.hpp"
+#endif
+
+#ifdef HAVE_OPENCV_GPUIMGPROC
+# include "opencv2/gpuimgproc.hpp"
+#endif
+
+#ifdef HAVE_OPENCV_GPUOPTFLOW
+# include "opencv2/gpuoptflow.hpp"
+#endif
+
+#ifdef HAVE_OPENCV_GPUCODEC
+# include "opencv2/gpucodec.hpp"
#endif
#ifdef HAVE_OPENCV_HIGHGUI
RunTest(cv::superres::createSuperResolution_BTVL1());
}
-#if defined(HAVE_OPENCV_GPU) && defined(HAVE_CUDA)
+#if defined(HAVE_CUDA) && defined(HAVE_OPENCV_GPUARITHM) && defined(HAVE_OPENCV_GPUWARPING) && defined(HAVE_OPENCV_GPUFILTERS)
TEST_F(SuperResolution, BTVL1_GPU)
{
#ifndef __OPENCV_GPU_TEST_UTILITY_HPP__
#define __OPENCV_GPU_TEST_UTILITY_HPP__
+#include <stdexcept>
#include "opencv2/core.hpp"
#include "opencv2/core/gpumat.hpp"
#include "opencv2/highgui.hpp"
CV_EXPORTS void PrintTo(const DeviceInfo& info, std::ostream* os);
}}
+#ifdef HAVE_CUDA
+
+#define CV_GPU_TEST_MAIN(resourcesubdir) \
+ int main(int argc, char* argv[]) \
+ { \
+ try \
+ { \
+ cv::CommandLineParser cmd(argc, argv, \
+ "{ h help ? | | Print help}" \
+ "{ i info | | Print information about system and exit }" \
+ "{ device | -1 | Device on which tests will be executed (-1 means all devices) }" \
+ ); \
+ if (cmd.has("help")) \
+ { \
+ cmd.printMessage(); \
+ return 0; \
+ } \
+ cvtest::printCudaInfo(); \
+ if (cmd.has("info")) \
+ { \
+ return 0; \
+ } \
+ int device = cmd.get<int>("device"); \
+ if (device < 0) \
+ { \
+ cvtest::DeviceManager::instance().loadAll(); \
+ std::cout << "Run tests on all supported devices \n" << std::endl; \
+ } \
+ else \
+ { \
+ cvtest::DeviceManager::instance().load(device); \
+ cv::gpu::DeviceInfo info(device); \
+ std::cout << "Run tests on device " << device << " [" << info.name() << "] \n" << std::endl; \
+ } \
+ cvtest::TS::ptr()->init( resourcesubdir ); \
+ testing::InitGoogleTest(&argc, argv); \
+ return RUN_ALL_TESTS(); \
+ } \
+ catch (const std::exception& e) \
+ { \
+ std::cerr << e.what() << std::endl; \
+ return -1; \
+ } \
+ catch (...) \
+ { \
+ std::cerr << "Unknown error" << std::endl; \
+ return -1; \
+ } \
+ return 0; \
+ }
+
+#else // HAVE_CUDA
+
+#define CV_GPU_TEST_MAIN(resourcesubdir) \
+ int main() \
+ { \
+ printf("OpenCV was built without CUDA support\n"); \
+ return 0; \
+ }
+
+#endif // HAVE_CUDA
+
+
#endif // __OPENCV_GPU_TEST_UTILITY_HPP__
set(the_description "Video stabilization")
-ocv_define_module(videostab opencv_imgproc opencv_features2d opencv_video opencv_photo opencv_calib3d OPTIONAL opencv_gpu opencv_highgui)
+ocv_define_module(videostab opencv_imgproc opencv_features2d opencv_video opencv_photo opencv_calib3d
+ OPTIONAL opencv_gpu opencv_gpuwarping opencv_gpuoptflow opencv_highgui)
#include "opencv2/videostab/motion_core.hpp"
#include "opencv2/videostab/outlier_rejection.hpp"
-#ifdef HAVE_OPENCV_GPU
- #include "opencv2/gpu.hpp"
+#ifdef HAVE_OPENCV_GPUIMGPROC
+# include "opencv2/gpuimgproc.hpp"
#endif
namespace cv
std::vector<Point2f> pointsPrevGood_, pointsGood_;
};
-#ifdef HAVE_OPENCV_GPU
+#if defined(HAVE_OPENCV_GPUIMGPROC) && defined(HAVE_OPENCV_GPU) && defined(HAVE_OPENCV_GPUOPTFLOW)
+
class CV_EXPORTS KeypointBasedMotionEstimatorGpu : public ImageMotionEstimatorBase
{
public:
std::vector<Point2f> hostPointsPrevTmp_, hostPointsTmp_;
std::vector<uchar> rejectionStatus_;
};
-#endif
+
+#endif // defined(HAVE_OPENCV_GPUIMGPROC) && defined(HAVE_OPENCV_GPU) && defined(HAVE_OPENCV_GPUOPTFLOW)
CV_EXPORTS Mat getMotion(int from, int to, const std::vector<Mat> &motions);
#include "opencv2/core.hpp"
#include "opencv2/opencv_modules.hpp"
-#ifdef HAVE_OPENCV_GPU
- #include "opencv2/gpu.hpp"
+#ifdef HAVE_OPENCV_GPUOPTFLOW
+ #include "opencv2/gpuoptflow.hpp"
#endif
namespace cv
OutputArray status, OutputArray errors);
};
-#ifdef HAVE_OPENCV_GPU
+#ifdef HAVE_OPENCV_GPUOPTFLOW
+
class CV_EXPORTS SparsePyrLkOptFlowEstimatorGpu
: public PyrLkOptFlowEstimatorBase, public ISparseOptFlowEstimator
{
gpu::PyrLKOpticalFlow optFlowEstimator_;
gpu::GpuMat frame0_, frame1_, flowX_, flowY_, errors_;
};
+
#endif
} // namespace videostab
#include <vector>
#include "opencv2/core.hpp"
+#include "opencv2/core/gpumat.hpp"
#include "opencv2/videostab/global_motion.hpp"
#include "opencv2/videostab/log.hpp"
-#ifdef HAVE_OPENCV_GPU
- #include "opencv2/gpu.hpp"
-#endif
-
namespace cv
{
namespace videostab
#include "opencv2/opencv_modules.hpp"
#include "clp.hpp"
+#ifdef HAVE_OPENCV_GPU
+# include "opencv2/gpu.hpp"
+#endif
+
namespace cv
{
namespace videostab
}
-#ifdef HAVE_OPENCV_GPU
+#if defined(HAVE_OPENCV_GPUIMGPROC) && defined(HAVE_OPENCV_GPU) && defined(HAVE_OPENCV_GPUOPTFLOW)
+
KeypointBasedMotionEstimatorGpu::KeypointBasedMotionEstimatorGpu(Ptr<MotionEstimatorBase> estimator)
: ImageMotionEstimatorBase(estimator->motionModel()), motionEstimator_(estimator)
{
// estimate motion
return motionEstimator_->estimate(hostPointsPrev_, hostPoints_, ok);
}
-#endif // HAVE_OPENCV_GPU
+
+#endif // defined(HAVE_OPENCV_GPUIMGPROC) && defined(HAVE_OPENCV_GPU) && defined(HAVE_OPENCV_GPUOPTFLOW)
Mat getMotion(int from, int to, const std::vector<Mat> &motions)
MotionInpainter::MotionInpainter()
{
-#ifdef HAVE_OPENCV_GPU
+#ifdef HAVE_OPENCV_GPUOPTFLOW
setOptFlowEstimator(new DensePyrLkOptFlowEstimatorGpu());
#else
CV_Error(Error::StsNotImplemented, "Current implementation of MotionInpainter requires GPU");
}
-#ifdef HAVE_OPENCV_GPU
+#ifdef HAVE_OPENCV_GPUOPTFLOW
+
SparsePyrLkOptFlowEstimatorGpu::SparsePyrLkOptFlowEstimatorGpu()
{
CV_Assert(gpu::getCudaEnabledDeviceCount() > 0);
flowX_.download(flowX.getMatRef());
flowY_.download(flowY.getMatRef());
}
-#endif // HAVE_OPENCV_GPU
+
+#endif // HAVE_OPENCV_GPUOPTFLOW
} // namespace videostab
} // namespace cv
#include "opencv2/videostab/wobble_suppression.hpp"
#include "opencv2/videostab/ring_buffer.hpp"
+#ifdef HAVE_OPENCV_GPUWARPING
+# include "opencv2/gpuwarping.hpp"
+#endif
+
+#ifdef HAVE_OPENCV_GPU
+# include "opencv2/gpu.hpp"
+#endif
+
+
namespace cv
{
namespace videostab
}
-#ifdef HAVE_OPENCV_GPU
+#ifdef HAVE_OPENCV_GPUWARPING
void MoreAccurateMotionWobbleSuppressorGpu::suppress(int idx, const gpu::GpuMat &frame, gpu::GpuMat &result)
{
CV_Assert(motions_ && stabilizationMotions_);
ocv_include_directories("${OpenCV_SOURCE_DIR}/include")#for opencv.hpp
ocv_include_modules(${OPENCV_CPP_SAMPLES_REQUIRED_DEPS})
- if(HAVE_opencv_gpu)
- ocv_include_directories("${OpenCV_SOURCE_DIR}/modules/gpu/include")
+ if(HAVE_opencv_gpuoptflow)
+ ocv_include_directories("${OpenCV_SOURCE_DIR}/modules/gpuoptflow/include")
+ endif()
+ if(HAVE_opencv_gpuimgproc)
+ ocv_include_directories("${OpenCV_SOURCE_DIR}/modules/gpuimgproc/include")
+ endif()
+ if(HAVE_opencv_gpuarithm)
+ ocv_include_directories("${OpenCV_SOURCE_DIR}/modules/gpuarithm/include")
+ endif()
+ if(HAVE_opencv_gpufilters)
+ ocv_include_directories("${OpenCV_SOURCE_DIR}/modules/gpufilters/include")
endif()
if(CMAKE_COMPILER_IS_GNUCXX AND NOT ENABLE_NOISY_WARNINGS)
add_executable(${the_target} ${srcs})
target_link_libraries(${the_target} ${OPENCV_LINKER_LIBS} ${OPENCV_CPP_SAMPLES_REQUIRED_DEPS})
- if(HAVE_opencv_gpu)
- target_link_libraries(${the_target} opencv_gpu)
+ if("${srcs}" MATCHES "gpu/")
+ target_link_libraries(${the_target} opencv_gpuarithm opencv_gpufilters)
endif()
set_target_properties(${the_target} PROPERTIES
ocv_list_filterout(cpp_samples Qt_sample)
endif()
- if(NOT HAVE_opencv_gpu)
+ if(NOT HAVE_opencv_gpuarithm OR NOT HAVE_opencv_gpufilters)
ocv_list_filterout(cpp_samples "/gpu/")
endif()
Ptr<FeaturesFinder> finder;
if (features_type == "surf")
{
-#if defined(HAVE_OPENCV_NONFREE) && defined(HAVE_OPENCV_GPU)
+#ifdef HAVE_OPENCV_NONFREE
if (try_gpu && gpu::getCudaEnabledDeviceCount() > 0)
finder = new SurfFeaturesFinderGpu();
else
// Warp images and their masks
Ptr<WarperCreator> warper_creator;
-#ifdef HAVE_OPENCV_GPU
+#ifdef HAVE_OPENCV_GPUWARPING
if (try_gpu && gpu::getCudaEnabledDeviceCount() > 0)
{
if (warp_type == "plane") warper_creator = new cv::PlaneWarperGpu();
#include <opencv2/core/utility.hpp>
#include <opencv2/imgproc.hpp>// Image processing methods for the CPU
#include <opencv2/highgui.hpp>// Read images
-#include <opencv2/gpu.hpp> // GPU structures and methods
+
+// GPU structures and methods
+#include <opencv2/gpuarithm.hpp>
+#include <opencv2/gpufilters.hpp>
using namespace std;
using namespace cv;
outlierRejector = tblor;
}
-#ifdef HAVE_OPENCV_GPU
+#if defined(HAVE_OPENCV_GPUIMGPROC) && defined(HAVE_OPENCV_GPU) && defined(HAVE_OPENCV_GPUOPTFLOW)
if (gpu)
{
KeypointBasedMotionEstimatorGpu *kbest = new KeypointBasedMotionEstimatorGpu(est);
outlierRejector = tblor;
}
-#ifdef HAVE_OPENCV_GPU
+#if defined(HAVE_OPENCV_GPUIMGPROC) && defined(HAVE_OPENCV_GPU) && defined(HAVE_OPENCV_GPUOPTFLOW)
if (gpu)
{
KeypointBasedMotionEstimatorGpu *kbest = new KeypointBasedMotionEstimatorGpu(est);
SET(OPENCV_GPU_SAMPLES_REQUIRED_DEPS opencv_core opencv_flann opencv_imgproc opencv_highgui
opencv_ml opencv_video opencv_objdetect opencv_features2d
opencv_calib3d opencv_legacy opencv_contrib opencv_gpu
- opencv_nonfree opencv_softcascade opencv_superres)
+ opencv_nonfree opencv_softcascade opencv_superres
+ opencv_gpuarithm opencv_gpufilters opencv_gpuwarping opencv_gpuimgproc
+ opencv_gpufeatures2d opencv_gpuoptflow opencv_gpubgsegm
+ opencv_gpustereo opencv_gpulegacy)
ocv_check_dependencies(${OPENCV_GPU_SAMPLES_REQUIRED_DEPS})
ocv_include_directories("${OpenCV_SOURCE_DIR}/modules/nonfree/include")
endif()
+ if(HAVE_opencv_gpucodec)
+ ocv_include_directories("${OpenCV_SOURCE_DIR}/modules/gpucodec/include")
+ endif()
+
if(HAVE_CUDA)
ocv_include_directories(${CUDA_INCLUDE_DIRS})
endif()
if(HAVE_opencv_nonfree)
target_link_libraries(${the_target} opencv_nonfree)
endif()
+ if(HAVE_opencv_gpucodec)
+ target_link_libraries(${the_target} opencv_gpucodec)
+ endif()
set_target_properties(${the_target} PROPERTIES
OUTPUT_NAME "${project}-example-${name}"
#include "opencv2/gpu.hpp"
#include "opencv2/highgui.hpp"
+#include "opencv2/opencv_modules.hpp"
+
+#ifdef HAVE_OPENCV_NONFREE
+# include "opencv2/nonfree/gpu.hpp"
+#endif
+
using namespace std;
using namespace cv;
using namespace cv::gpu;
#include <cstdio>
#include "opencv2/gpu/gpu.hpp"
#include "opencv2/highgui/highgui.hpp"
+#include "opencv2/objdetect/objdetect.hpp"
#include "opencv2/objdetect/objdetect_c.h"
#ifdef HAVE_CUDA
-#include "NCVHaarObjectDetection.hpp"
+#include "opencv2/gpulegacy.hpp"
#endif
using namespace std;
#include <opencv2/core/utility.hpp>
#include "opencv2/gpu.hpp"
#include "opencv2/highgui.hpp"
+#include "opencv2/objdetect.hpp"
using namespace std;
using namespace cv;
#include "opencv2/highgui/highgui_c.h"
#ifdef HAVE_CUDA
-#include "NPP_staging/NPP_staging.hpp"
-#include "NCVBroxOpticalFlow.hpp"
+#include "opencv2/gpulegacy.hpp"
#endif
#if !defined(HAVE_CUDA)
#include <iostream>
+
+#include "opencv2/opencv_modules.hpp"
+
+#if defined(HAVE_OPENCV_GPUCODEC)
+
#include <string>
#include <vector>
#include <algorithm>
#include <numeric>
-#include <opencv2/core/core.hpp>
+#include <opencv2/core.hpp>
#include <opencv2/core/opengl.hpp>
-#include <opencv2/gpu/gpu.hpp>
-#include <opencv2/highgui/highgui.hpp>
-#include <opencv2/contrib/contrib.hpp>
+#include <opencv2/gpucodec.hpp>
+#include <opencv2/highgui.hpp>
+#include <opencv2/contrib.hpp>
int main(int argc, const char* argv[])
{
return 0;
}
+
+#else
+
+int main()
+{
+ std::cout << "OpenCV was built without GPU Video decoding support\n" << std::endl;
+ return 0;
+}
+
+#endif
#include <iostream>
+
+#include "opencv2/opencv_modules.hpp"
+
+#if defined(HAVE_OPENCV_GPUCODEC) && defined(WIN32)
+
#include <vector>
#include <numeric>
-#include "opencv2/core/core.hpp"
-#include "opencv2/gpu/gpu.hpp"
-#include "opencv2/highgui/highgui.hpp"
-#include "opencv2/contrib/contrib.hpp"
+#include "opencv2/core.hpp"
+#include "opencv2/gpucodec.hpp"
+#include "opencv2/highgui.hpp"
+#include "opencv2/contrib.hpp"
int main(int argc, const char* argv[])
{
return 0;
}
+
+#else
+
+int main()
+{
+ std::cout << "OpenCV was built without GPU Video encoding support\n" << std::endl;
+ return 0;
+}
+
+#endif
projects / platform / upstream / opencv.git / commitdiff
