--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////\r
+//\r
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.\r
+//\r
+// By downloading, copying, installing or using the software you agree to this license.\r
+// If you do not agree to this license, do not download, install,\r
+// copy or use the software.\r
+//\r
+//\r
+// License Agreement\r
+// For Open Source Computer Vision Library\r
+//\r
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.\r
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.\r
+// Third party copyrights are property of their respective owners.\r
+//\r
+// Redistribution and use in source and binary forms, with or without modification,\r
+// are permitted provided that the following conditions are met:\r
+//\r
+// * Redistribution's of source code must retain the above copyright notice,\r
+// this list of conditions and the following disclaimer.\r
+//\r
+// * Redistribution's in binary form must reproduce the above copyright notice,\r
+// this list of conditions and the following disclaimer in the documentation\r
+// and/or other GpuMaterials provided with the distribution.\r
+//\r
+// * The name of the copyright holders may not be used to endorse or promote products\r
+// derived from this software without specific prior written permission.\r
+//\r
+// This software is provided by the copyright holders and contributors "as is" and\r
+// any express or implied warranties, including, but not limited to, the implied\r
+// warranties of merchantability and fitness for a particular purpose are disclaimed.\r
+// In no event shall the Intel Corporation or contributors be liable for any direct,\r
+// indirect, incidental, special, exemplary, or consequential damages\r
+// (including, but not limited to, procurement of substitute goods or services;\r
+// loss of use, data, or profits; or business interruption) however caused\r
+// and on any theory of liability, whether in contract, strict liability,\r
+// or tort (including negligence or otherwise) arising in any way out of\r
+// the use of this software, even if advised of the possibility of such damage.\r
+//\r
+//M*/\r
+\r
+#include "precomp.hpp"\r
+\r
+using namespace cv;\r
+using namespace cv::gpu;\r
+using namespace std;\r
+\r
+#if !defined (HAVE_CUDA)\r
+\r
+cv::gpu::DisparityBilateralFilter::DisparityBilateralFilter(int, int, int) { throw_nogpu(); }\r
+cv::gpu::DisparityBilateralFilter::DisparityBilateralFilter(int, int, int, float, float, float) { throw_nogpu(); }\r
+\r
+void cv::gpu::DisparityBilateralFilter::operator()(const GpuMat&, const GpuMat&, GpuMat&) { throw_nogpu(); }\r
+void cv::gpu::DisparityBilateralFilter::operator()(const GpuMat&, const GpuMat&, GpuMat&, Stream&) { throw_nogpu(); }\r
+\r
+#else /* !defined (HAVE_CUDA) */\r
+\r
+namespace cv { namespace gpu { namespace bf \r
+{\r
+ void calc_space_weighted_filter_gpu(const DevMem2Df& table_space, int half, float dist_space, cudaStream_t stream);\r
+ void load_constants(float* table_color, const DevMem2Df& table_space, int ndisp, int radius, short edge_disc, short max_disc);\r
+\r
+ void bilateral_filter_gpu(const DevMem2D& disp, const DevMem2D& img, int channels, int iters, cudaStream_t stream);\r
+ void bilateral_filter_gpu(const DevMem2D_<short>& disp, const DevMem2D& img, int channels, int iters, cudaStream_t stream);\r
+}}}\r
+\r
+namespace\r
+{\r
+ const float DEFAULT_EDGE_THRESHOLD = 0.1f;\r
+ const float DEFAULT_MAX_DISC_THRESHOLD = 0.2f;\r
+ const float DEFAULT_SIGMA_RANGE = 10.0f;\r
+\r
+ inline void calc_color_weighted_table(vector<float>& table_color, float sigma_range, int len)\r
+ {\r
+ float* color_table_x;\r
+ \r
+ table_color.resize(len);\r
+ color_table_x = &table_color[0];\r
+\r
+ for(int y = 0; y < len; y++) \r
+ table_color[y] = static_cast<float>(std::exp(-double(y * y) / (2 * sigma_range * sigma_range)));\r
+ }\r
+\r
+ inline void calc_space_weighted_filter(GpuMat& table_space, int win_size, float dist_space, cudaStream_t stream)\r
+ {\r
+ int half = (win_size >> 1);\r
+ table_space.create(half + 1, half + 1, CV_32F);\r
+\r
+ bf::calc_space_weighted_filter_gpu(table_space, half, dist_space, stream);\r
+ }\r
+\r
+ template <typename T>\r
+ void bilateral_filter_operator(DisparityBilateralFilter& rthis, vector<float>& table_color, GpuMat& table_space, \r
+ const GpuMat& disp, const GpuMat& img, GpuMat& dst, cudaStream_t stream)\r
+ {\r
+ calc_color_weighted_table(table_color, rthis.sigma_range, 255);\r
+ calc_space_weighted_filter(table_space, rthis.radius * 2 + 1, rthis.radius + 1.0f, stream);\r
+\r
+ short edge_disc = max<short>(short(1), short(rthis.ndisp * rthis.edge_threshold + 0.5));\r
+ short max_disc = short(rthis.ndisp * rthis.max_disc_threshold + 0.5);\r
+\r
+ bf::load_constants(&table_color[0], table_space, rthis.ndisp, rthis.radius, edge_disc, max_disc);\r
+\r
+ if (&dst != &disp)\r
+ disp.copyTo(dst);\r
+\r
+ bf::bilateral_filter_gpu((DevMem2D_<T>)dst, img, img.channels(), rthis.iters, stream);\r
+ }\r
+\r
+ typedef void (*bilateral_filter_operator_t)(DisparityBilateralFilter& rthis, vector<float>& table_color, GpuMat& table_space, \r
+ const GpuMat& disp, const GpuMat& img, GpuMat& dst, cudaStream_t stream);\r
+ \r
+ const bilateral_filter_operator_t operators[] = \r
+ {bilateral_filter_operator<unsigned char>, 0, 0, bilateral_filter_operator<short>, 0, 0, 0, 0};\r
+}\r
+\r
+cv::gpu::DisparityBilateralFilter::DisparityBilateralFilter(int ndisp_, int radius_, int iters_)\r
+ : ndisp(ndisp_), radius(radius_), iters(iters_), edge_threshold(DEFAULT_EDGE_THRESHOLD), max_disc_threshold(DEFAULT_MAX_DISC_THRESHOLD),\r
+ sigma_range(DEFAULT_SIGMA_RANGE)\r
+{\r
+}\r
+\r
+cv::gpu::DisparityBilateralFilter::DisparityBilateralFilter(int ndisp_, int radius_, int iters_, float edge_threshold_, \r
+ float max_disc_threshold_, float sigma_range_)\r
+ : ndisp(ndisp_), radius(radius_), iters(iters_), edge_threshold(edge_threshold_), max_disc_threshold(max_disc_threshold_), \r
+ sigma_range(sigma_range_)\r
+{\r
+}\r
+\r
+void cv::gpu::DisparityBilateralFilter::operator()(const GpuMat& disp, const GpuMat& img, GpuMat& dst)\r
+{\r
+ CV_DbgAssert(0 < ndisp && 0 < radius && 0 < iters);\r
+ CV_Assert(disp.rows == img.rows && disp.cols == img.cols && (disp.type() == CV_8U || disp.type() == CV_16S) && (img.type() == CV_8UC1 || img.type() == CV_8UC3));\r
+ operators[disp.type()](*this, table_color, table_space, disp, img, dst, 0);\r
+}\r
+\r
+void cv::gpu::DisparityBilateralFilter::operator()(const GpuMat& disp, const GpuMat& img, GpuMat& dst, Stream& stream)\r
+{\r
+ CV_DbgAssert(0 < ndisp && 0 < radius && 0 < iters);\r
+ CV_Assert(disp.rows == img.rows && disp.cols == img.cols && (disp.type() == CV_8U || disp.type() == CV_16S) && (img.type() == CV_8UC1 || img.type() == CV_8UC3));\r
+ operators[disp.type()](*this, table_color, table_space, disp, img, dst, StreamAccessor::getStream(stream));\r
+}\r
+\r
+#endif /* !defined (HAVE_CUDA) */\r
--- /dev/null
+/*M///////////////////////////////////////////////////////////////////////////////////////\r
+//\r
+// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.\r
+//\r
+// By downloading, copying, installing or using the software you agree to this license.\r
+// If you do not agree to this license, do not download, install,\r
+// copy or use the software.\r
+//\r
+//\r
+// License Agreement\r
+// For Open Source Computer Vision Library\r
+//\r
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.\r
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.\r
+// Third party copyrights are property of their respective owners.\r
+//\r
+// Redistribution and use in source and binary forms, with or without modification,\r
+// are permitted provided that the following conditions are met:\r
+//\r
+// * Redistribution's of source code must retain the above copyright notice,\r
+// this list of conditions and the following disclaimer.\r
+//\r
+// * Redistribution's in binary form must reproduce the above copyright notice,\r
+// this list of conditions and the following disclaimer in the documentation\r
+// and/or other materials provided with the distribution.\r
+//\r
+// * The name of the copyright holders may not be used to endorse or promote products\r
+// derived from this software without specific prior written permission.\r
+//\r
+// This software is provided by the copyright holders and contributors "as is" and\r
+// any express or implied warranties, including, but not limited to, the implied\r
+// warranties of merchantability and fitness for a particular purpose are disclaimed.\r
+// In no event shall the Intel Corporation or contributors be liable for any direct,\r
+// indirect, incidental, special, exemplary, or consequential damages\r
+// (including, but not limited to, procurement of substitute goods or services;\r
+// loss of use, data, or profits; or business interruption) however caused\r
+// and on any theory of liability, whether in contract, strict liability,\r
+// or tort (including negligence or otherwise) arising in any way out of\r
+// the use of this software, even if advised of the possibility of such damage.\r
+//\r
+//M*/\r
+\r
+#include "opencv2/gpu/devmem2d.hpp"\r
+#include "saturate_cast.hpp"\r
+#include "safe_call.hpp"\r
+\r
+using namespace cv::gpu;\r
+using namespace cv::gpu::impl;\r
+\r
+#ifndef FLT_MAX\r
+#define FLT_MAX 3.402823466e+30F\r
+#endif\r
+\r
+namespace bf_krnls\r
+{\r
+ __global__ void calc_space_weighted_filter(float* table_space, size_t step, int half, float dist_space)\r
+ {\r
+ int x = blockIdx.x * blockDim.x + threadIdx.x;\r
+ int y = blockIdx.y * blockDim.y + threadIdx.y;\r
+\r
+ if (y <= half && x <= half)\r
+ *(table_space + y * step + x) = expf(-sqrtf(float(y * y) + float(x * x)) / dist_space);\r
+ }\r
+}\r
+\r
+namespace cv { namespace gpu { namespace bf \r
+{\r
+ void calc_space_weighted_filter_gpu(const DevMem2Df& table_space, int half, float dist_space, cudaStream_t stream)\r
+ {\r
+ dim3 threads(32, 8, 1);\r
+ dim3 grid(1, 1, 1);\r
+ grid.x = divUp(half + 1, threads.x);\r
+ grid.x = divUp(half + 1, threads.y);\r
+\r
+ bf_krnls::calc_space_weighted_filter<<<grid, threads, 0, stream>>>(table_space.ptr, table_space.step/sizeof(float), half, dist_space);\r
+\r
+ if (stream != 0)\r
+ cudaSafeCall( cudaThreadSynchronize() );\r
+ }\r
+}}}\r
+\r
+namespace bf_krnls\r
+{\r
+ __constant__ float* ctable_color;\r
+ __constant__ float* ctable_space;\r
+ __constant__ size_t ctable_space_step;\r
+\r
+ __constant__ int cndisp;\r
+ __constant__ int cradius;\r
+\r
+ __constant__ short cedge_disc;\r
+ __constant__ short cmax_disc;\r
+}\r
+\r
+namespace cv { namespace gpu { namespace bf \r
+{\r
+ void load_constants(float* table_color, const DevMem2Df& table_space, int ndisp, int radius, short edge_disc, short max_disc)\r
+ {\r
+ cudaSafeCall( cudaMemcpyToSymbol(bf_krnls::ctable_color, &table_color, sizeof(table_color)) );\r
+ cudaSafeCall( cudaMemcpyToSymbol(bf_krnls::ctable_space, &table_space.ptr, sizeof(table_space.ptr)) );\r
+ size_t table_space_step = table_space.step / sizeof(float);\r
+ cudaSafeCall( cudaMemcpyToSymbol(bf_krnls::ctable_space_step, &table_space_step, sizeof(size_t)) );\r
+ \r
+ cudaSafeCall( cudaMemcpyToSymbol(bf_krnls::cndisp, &ndisp, sizeof(int)) );\r
+ cudaSafeCall( cudaMemcpyToSymbol(bf_krnls::cradius, &radius, sizeof(int)) );\r
+ \r
+ cudaSafeCall( cudaMemcpyToSymbol(bf_krnls::cedge_disc, &edge_disc, sizeof(short)) );\r
+ cudaSafeCall( cudaMemcpyToSymbol(bf_krnls::cmax_disc, &max_disc, sizeof(short)) );\r
+ }\r
+}}}\r
+\r
+namespace bf_krnls\r
+{\r
+ template <int channels>\r
+ struct DistRgbMax\r
+ {\r
+ static __device__ uchar calc(const uchar* a, const uchar* b)\r
+ {\r
+ uchar x = abs(a[0] - b[0]);\r
+ uchar y = abs(a[1] - b[1]);\r
+ uchar z = abs(a[2] - b[2]);\r
+ return (max(max(x, y), z));\r
+ }\r
+ };\r
+\r
+ template <>\r
+ struct DistRgbMax<1>\r
+ {\r
+ static __device__ uchar calc(const uchar* a, const uchar* b)\r
+ {\r
+ return abs(a[0] - b[0]);\r
+ }\r
+ };\r
+\r
+ template <int channels, typename T>\r
+ __global__ void bilateral_filter(int t, T* disp, size_t disp_step, const uchar* img, size_t img_step, int h, int w)\r
+ {\r
+ const int y = blockIdx.y * blockDim.y + threadIdx.y;\r
+ const int x = ((blockIdx.x * blockDim.x + threadIdx.x) << 1) + ((y + t) & 1);\r
+\r
+ T dp[5];\r
+\r
+ if (y > 0 && y < h - 1 && x > 0 && x < w - 1)\r
+ {\r
+ dp[0] = *(disp + (y ) * disp_step + x + 0);\r
+ dp[1] = *(disp + (y-1) * disp_step + x + 0);\r
+ dp[2] = *(disp + (y ) * disp_step + x - 1);\r
+ dp[3] = *(disp + (y+1) * disp_step + x + 0);\r
+ dp[4] = *(disp + (y ) * disp_step + x + 1);\r
+\r
+ if(abs(dp[1] - dp[0]) >= cedge_disc || abs(dp[2] - dp[0]) >= cedge_disc || abs(dp[3] - dp[0]) >= cedge_disc || abs(dp[4] - dp[0]) >= cedge_disc) \r
+ {\r
+ const int ymin = max(0, y - cradius);\r
+ const int xmin = max(0, x - cradius);\r
+ const int ymax = min(h - 1, y + cradius);\r
+ const int xmax = min(w - 1, x + cradius);\r
+\r
+ float cost[] = {0.0f, 0.0f, 0.0f, 0.0f, 0.0f};\r
+\r
+ const uchar* ic = img + y * img_step + channels * x;\r
+\r
+ for(int yi = ymin; yi <= ymax; yi++)\r
+ {\r
+ const T* disp_y = disp + yi * disp_step;\r
+\r
+ for(int xi = xmin; xi <= xmax; xi++)\r
+ {\r
+ const uchar* in = img + yi * img_step + channels * xi;\r
+\r
+ uchar dist_rgb = DistRgbMax<channels>::calc(in, ic);\r
+\r
+ const float weight = ctable_color[dist_rgb] * (ctable_space + abs(y-yi)* ctable_space_step)[abs(x-xi)];\r
+\r
+ const T disp_reg = disp_y[xi];\r
+\r
+ cost[0] += min(cmax_disc, abs(disp_reg - dp[0])) * weight;\r
+ cost[1] += min(cmax_disc, abs(disp_reg - dp[1])) * weight;\r
+ cost[2] += min(cmax_disc, abs(disp_reg - dp[2])) * weight;\r
+ cost[3] += min(cmax_disc, abs(disp_reg - dp[3])) * weight;\r
+ cost[4] += min(cmax_disc, abs(disp_reg - dp[4])) * weight;\r
+ }\r
+ }\r
+\r
+ float minimum = FLT_MAX;\r
+ int id = 0;\r
+\r
+ if (cost[0] < minimum)\r
+ {\r
+ minimum = cost[0];\r
+ id = 0;\r
+ }\r
+ if (cost[1] < minimum)\r
+ {\r
+ minimum = cost[1];\r
+ id = 1;\r
+ }\r
+ if (cost[2] < minimum)\r
+ {\r
+ minimum = cost[2];\r
+ id = 2;\r
+ }\r
+ if (cost[3] < minimum)\r
+ {\r
+ minimum = cost[3];\r
+ id = 3;\r
+ }\r
+ if (cost[4] < minimum)\r
+ {\r
+ minimum = cost[4];\r
+ id = 4;\r
+ }\r
+\r
+ *(disp + y * disp_step + x) = dp[id];\r
+ }\r
+ }\r
+ }\r
+}\r
+\r
+namespace cv { namespace gpu { namespace bf \r
+{\r
+ template <typename T> \r
+ void bilateral_filter_caller(const DevMem2D_<T>& disp, const DevMem2D& img, int channels, int iters, cudaStream_t stream)\r
+ {\r
+ dim3 threads(32, 8, 1);\r
+ dim3 grid(1, 1, 1);\r
+ grid.x = divUp(disp.cols, threads.x << 1);\r
+ grid.y = divUp(disp.rows, threads.y);\r
+\r
+ switch (channels)\r
+ {\r
+ case 1:\r
+ for (int i = 0; i < iters; ++i)\r
+ {\r
+ bf_krnls::bilateral_filter<1><<<grid, threads, 0, stream>>>(0, disp.ptr, disp.step/sizeof(T), img.ptr, img.step, disp.rows, disp.cols);\r
+ bf_krnls::bilateral_filter<1><<<grid, threads, 0, stream>>>(1, disp.ptr, disp.step/sizeof(T), img.ptr, img.step, disp.rows, disp.cols);\r
+ }\r
+ break;\r
+ case 3:\r
+ for (int i = 0; i < iters; ++i)\r
+ {\r
+ bf_krnls::bilateral_filter<3><<<grid, threads, 0, stream>>>(0, disp.ptr, disp.step/sizeof(T), img.ptr, img.step, disp.rows, disp.cols);\r
+ bf_krnls::bilateral_filter<3><<<grid, threads, 0, stream>>>(1, disp.ptr, disp.step/sizeof(T), img.ptr, img.step, disp.rows, disp.cols);\r
+ }\r
+ break;\r
+ default:\r
+ cv::gpu::error("Unsupported channels count", __FILE__, __LINE__);\r
+ } \r
+\r
+ if (stream != 0)\r
+ cudaSafeCall( cudaThreadSynchronize() );\r
+ }\r
+\r
+ void bilateral_filter_gpu(const DevMem2D& disp, const DevMem2D& img, int channels, int iters, cudaStream_t stream)\r
+ {\r
+ bilateral_filter_caller(disp, img, channels, iters, stream);\r
+ }\r
+\r
+ void bilateral_filter_gpu(const DevMem2D_<short>& disp, const DevMem2D& img, int channels, int iters, cudaStream_t stream)\r
+ {\r
+ bilateral_filter_caller(disp, img, channels, iters, stream);\r
+ }\r
+}}}\r
projects / profile / ivi / opencv.git / commitdiff