gpu module: refactored devmem2d.hpp (atomic bomb)

diff --git a/modules/gpu/include/opencv2/gpu/devmem2d.hpp b/modules/gpu/include/opencv2/gpu/devmem2d.hpp
index 232783f..e454f00 100644 (file)
--- a/modules/gpu/include/opencv2/gpu/devmem2d.hpp
+++ b/modules/gpu/include/opencv2/gpu/devmem2d.hpp
@@ -40,12 +40,8 @@
 //\r
 //M*/\r
 \r
-#ifndef __OPENCV_GPU_DEVMEM2D_HPP__\r
-#define __OPENCV_GPU_DEVMEM2D_HPP__\r
-\r
-#if defined(__DEVCLASES_ADD_THRUST_BEGIN_END__)\r
-    #include "thrust/device_ptr.h"\r
-#endif\r
+#ifndef __OPENCV_GPU_DevMem2D_HPP__\r
+#define __OPENCV_GPU_DevMem2D_HPP__\r
 \r
 \r
 namespace cv\r
@@ -63,97 +59,103 @@ namespace cv
 \r
         template <bool expr> struct StaticAssert;\r
         template <> struct StaticAssert<true> {static __CV_GPU_HOST_DEVICE__ void check(){}};        \r
-        \r
-        template <typename T> struct DevMem2D_\r
-        {            \r
-            typedef T elem_type;\r
-            typedef int index_type;\r
 \r
-            int cols;\r
-            int rows;\r
-            T* data;\r
-            size_t step;\r
-            \r
-            DevMem2D_() : cols(0), rows(0), data(0), step(0) {}\r
-            \r
-            DevMem2D_(int rows_, int cols_, T *data_, size_t step_)\r
-                : cols(cols_), rows(rows_), data(data_), step(step_) {}\r
-            \r
-            template <typename U>            \r
-            explicit DevMem2D_(const DevMem2D_<U>& d)\r
-                : cols(d.cols), rows(d.rows), data((T*)d.data), step(d.step) {}\r
+               template<typename T> struct DevPtr\r
+               {\r
+                       typedef T elem_type;\r
+                       typedef int index_type;\r
 \r
-            enum { elem_size = sizeof(elem_type) };\r
+                       enum { elem_size = sizeof(elem_type) };\r
 \r
-            __CV_GPU_HOST_DEVICE__ size_t elemSize() const { return elem_size; }\r
-            __CV_GPU_HOST_DEVICE__ T* ptr(int y = 0) { return (T*)( (char*)data + y * step ); }\r
-            __CV_GPU_HOST_DEVICE__ const T* ptr(int y = 0) const { return (const T*)( (const char*)data + y * step ); }            \r
+                       T* data;\r
 \r
-                       __CV_GPU_HOST_DEVICE__ operator T*() const { return data; }\r
+                       __CV_GPU_HOST_DEVICE__ DevPtr() : data(0) {}\r
+                       __CV_GPU_HOST_DEVICE__ DevPtr(T* data_) : data(data_) {}\r
 \r
-            __CV_GPU_HOST_DEVICE__ T& operator ()(int y, int x) { return ptr(y)[x]; }\r
-            __CV_GPU_HOST_DEVICE__ const T& operator ()(int y, int x) const { return ptr(y)[x]; }\r
+                       __CV_GPU_HOST_DEVICE__ size_t elemSize() const { return elem_size; }\r
+                       __CV_GPU_HOST_DEVICE__ operator       T*()       { return data; }\r
+                       __CV_GPU_HOST_DEVICE__ operator const T*() const { return data; }\r
+               };\r
+               \r
+               template<typename T> struct PtrSz : public DevPtr<T>\r
+        {                     \r
+            __CV_GPU_HOST_DEVICE__ PtrSz() : size(0) {}\r
+            __CV_GPU_HOST_DEVICE__ PtrSz(T* data_, size_t size_) : DevPtr<T>(data_), size(size_) {}\r
 \r
-#if defined(__DEVCLASES_ADD_THRUST_BEGIN_END__)    \r
-            thrust::device_ptr<T> begin() const { return thrust::device_ptr<T>(data); }\r
-            thrust::device_ptr<T> end() const { return thrust::device_ptr<T>(data) + cols * rows; }\r
-#endif\r
+            size_t size;\r
         };\r
- \r
-        template<typename T> struct PtrStep_\r
-        {\r
-            typedef T elem_type;\r
-            typedef int index_type;\r
-\r
-            T* data;\r
-            size_t step;\r
 \r
-            PtrStep_() : data(0), step(0) {}            \r
-            PtrStep_(const DevMem2D_<T>& mem) : data(mem.data), step(mem.step) {}\r
+               template<typename T> struct PtrStep : public DevPtr<T>\r
+        {   \r
+            __CV_GPU_HOST_DEVICE__ PtrStep() : step(0) {}\r
+                       __CV_GPU_HOST_DEVICE__ PtrStep(T* data_, size_t step_) : DevPtr<T>(data_), step(step_) {}\r
 \r
-            enum { elem_size = sizeof(elem_type) };\r
+            /** \brief stride between two consecutive rows in bytes. Step is stored always and everywhere in bytes!!! */\r
+            size_t step;            \r
 \r
-            __CV_GPU_HOST_DEVICE__ size_t elemSize() const { return elem_size; }\r
-            __CV_GPU_HOST_DEVICE__ T* ptr(int y = 0) { return (T*)( (char*)data + y * step); }\r
-            __CV_GPU_HOST_DEVICE__ const T* ptr(int y = 0) const { return (const T*)( (const char*)data + y * step); }\r
+                       __CV_GPU_HOST_DEVICE__       T* ptr(int y = 0)       { return (      T*)( (      char*)DevPtr<T>::data + y * step); }\r
+            __CV_GPU_HOST_DEVICE__ const T* ptr(int y = 0) const { return (const T*)( (const char*)DevPtr<T>::data + y * step); }\r
 \r
-            __CV_GPU_HOST_DEVICE__ T& operator ()(int y, int x) { return ptr(y)[x]; }\r
+                       __CV_GPU_HOST_DEVICE__       T& operator ()(int y, int x)       { return ptr(y)[x]; }\r
             __CV_GPU_HOST_DEVICE__ const T& operator ()(int y, int x) const { return ptr(y)[x]; }\r
+        };\r
 \r
-#if defined(__DEVCLASES_ADD_THRUST_BEGIN_END__)    \r
-            thrust::device_ptr<T> begin() const { return thrust::device_ptr<T>(data); }\r
-#endif\r
+               template <typename T> struct PtrStepSz : public PtrStep<T>\r
+        {   \r
+            __CV_GPU_HOST_DEVICE__ PtrStepSz() : cols(0), rows(0) {}\r
+            __CV_GPU_HOST_DEVICE__ PtrStepSz(int rows_, int cols_, T* data_, size_t step_) \r
+                : PtrStep<T>(data_, step_), cols(cols_), rows(rows_) {}\r
+\r
+            int cols;\r
+            int rows;                                                                              \r
         };\r
-               \r
-        template<typename T> struct PtrElemStep_ : public PtrStep_<T>\r
+\r
+               template <typename T> struct DevMem2D_ : public PtrStepSz<T>\r
+        {            \r
+            DevMem2D_() {}\r
+                       DevMem2D_(int rows_, int cols_, T *data_, size_t step_) : PtrStepSz<T>(rows_, cols_, data_, step_) {}\r
+                            \r
+            template <typename U>            \r
+                       explicit DevMem2D_(const DevMem2D_<U>& d) : PtrStepSz<T>(d.rows, d.cols, (T*)d.data, d.step) {}                                                                \r
+        };\r
+                              \r
+        template<typename T> struct PtrElemStep_ : public PtrStep<T>\r
         {                   \r
-            PtrElemStep_(const DevMem2D_<T>& mem) : PtrStep_<T>(mem) \r
+            PtrElemStep_(const DevMem2D_<T>& mem) : PtrStep<T>(mem.data, mem.step) \r
             {\r
                 StaticAssert<256 % sizeof(T) == 0>::check();\r
 \r
-                PtrStep_<T>::step /= PtrStep_<T>::elem_size;             \r
+                PtrStep<T>::step /= PtrStep<T>::elem_size;             \r
             }\r
-            __CV_GPU_HOST_DEVICE__ T* ptr(int y = 0) { return PtrStep_<T>::data + y * PtrStep_<T>::step; }\r
-            __CV_GPU_HOST_DEVICE__ const T* ptr(int y = 0) const { return PtrStep_<T>::data + y * PtrStep_<T>::step; }  \r
+            __CV_GPU_HOST_DEVICE__ T* ptr(int y = 0) { return PtrStep<T>::data + y * PtrStep<T>::step; }\r
+            __CV_GPU_HOST_DEVICE__ const T* ptr(int y = 0) const { return PtrStep<T>::data + y * PtrStep<T>::step; }  \r
 \r
             __CV_GPU_HOST_DEVICE__ T& operator ()(int y, int x) { return ptr(y)[x]; }\r
             __CV_GPU_HOST_DEVICE__ const T& operator ()(int y, int x) const { return ptr(y)[x]; }                  \r
         };\r
 \r
-        typedef DevMem2D_<unsigned char> DevMem2D;\r
+               template<typename T> struct PtrStep_ : public PtrStep<T>\r
+        {            \r
+            PtrStep_() {}\r
+            PtrStep_(const DevMem2D_<T>& mem) : PtrStep<T>(mem.data, mem.step) {}                        \r
+        };\r
+\r
+#undef __CV_GPU_HOST_DEVICE__\r
+\r
+\r
+        typedef DevMem2D_<unsigned char> DevMem2Db;\r
+               typedef DevMem2Db DevMem2D;\r
         typedef DevMem2D_<float> DevMem2Df;\r
         typedef DevMem2D_<int> DevMem2Di;\r
 \r
-        typedef PtrStep_<unsigned char> PtrStep;\r
-        typedef PtrStep_<float> PtrStepf;\r
-        typedef PtrStep_<int> PtrStepi;\r
+        typedef PtrStep<unsigned char> PtrStepb;\r
+        typedef PtrStep<float> PtrStepf;\r
+        typedef PtrStep<int> PtrStepi;\r
 \r
         typedef PtrElemStep_<unsigned char> PtrElemStep;\r
         typedef PtrElemStep_<float> PtrElemStepf;\r
-        typedef PtrElemStep_<int> PtrElemStepi;\r
-\r
-#undef __CV_GPU_HOST_DEVICE__\r
+        typedef PtrElemStep_<int> PtrElemStepi;                \r
     }    \r
 }\r
 \r
-#endif /* __OPENCV_GPU_DEVMEM2D_HPP__ */\r
+#endif /* __OPENCV_GPU_DevMem2D_HPP__ */\r

diff --git a/modules/gpu/include/opencv2/gpu/gpumat.hpp b/modules/gpu/include/opencv2/gpu/gpumat.hpp
index 8ce456a..c3dc39b 100644 (file)
--- a/modules/gpu/include/opencv2/gpu/gpumat.hpp
+++ b/modules/gpu/include/opencv2/gpu/gpumat.hpp
@@ -89,6 +89,7 @@ namespace cv { namespace gpu
         // Contains just image size, data ptr and step.\r
         template <class T> operator DevMem2D_<T>() const;\r
         template <class T> operator PtrStep_<T>() const;\r
+               template <class T> operator PtrStep<T>() const;\r
 \r
         //! pefroms blocking upload data to GpuMat.\r
         void upload(const cv::Mat& m);\r
@@ -238,6 +239,7 @@ namespace cv { namespace gpu
 \r
     template <class T> inline GpuMat::operator DevMem2D_<T>() const { return DevMem2D_<T>(rows, cols, (T*)data, step); }\r
     template <class T> inline GpuMat::operator PtrStep_<T>() const { return PtrStep_<T>(static_cast< DevMem2D_<T> >(*this)); }\r
+       template <class T> inline GpuMat::operator PtrStep<T>() const { return PtrStep<T>((T*)data, step); }    \r
 \r
     inline GpuMat GpuMat::clone() const\r
     {\r

diff --git a/modules/gpu/src/bilateral_filter.cpp b/modules/gpu/src/bilateral_filter.cpp
index 79ff0f0..bc2bec2 100644 (file)
--- a/modules/gpu/src/bilateral_filter.cpp
+++ b/modules/gpu/src/bilateral_filter.cpp
@@ -59,8 +59,8 @@ namespace cv { namespace gpu { namespace bf
 {\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
+    void bilateral_filter_gpu(const DevMem2Db& disp, const DevMem2Db& img, int channels, int iters, cudaStream_t stream);\r
+    void bilateral_filter_gpu(const DevMem2D_<short>& disp, const DevMem2Db& img, int channels, int iters, cudaStream_t stream);\r
 }}}\r
 \r
 namespace\r

diff --git a/modules/gpu/src/blend.cpp b/modules/gpu/src/blend.cpp
index 8eb578b..54345f6 100644 (file)
--- a/modules/gpu/src/blend.cpp
+++ b/modules/gpu/src/blend.cpp
@@ -55,11 +55,11 @@ void cv::gpu::blendLinear(const GpuMat&, const GpuMat&, const GpuMat&, const Gpu
 namespace cv { namespace gpu \r
 {\r
     template <typename T>\r
-    void blendLinearCaller(int rows, int cols, int cn, const PtrStep_<T> img1, const PtrStep_<T> img2, \r
-                           const PtrStep_<float> weights1, const PtrStep_<float> weights2, PtrStep_<T> result, cudaStream_t stream);\r
+    void blendLinearCaller(int rows, int cols, int cn, const PtrStep<T>& img1, const PtrStep<T>& img2, \r
+                           const PtrStepf& weights1, const PtrStepf& weights2, PtrStep<T> result, cudaStream_t stream);\r
 \r
-    void blendLinearCaller8UC4(int rows, int cols, const PtrStep img1, const PtrStep img2, \r
-                               const PtrStepf weights1, const PtrStepf weights2, PtrStep result, cudaStream_t stream);\r
+    void blendLinearCaller8UC4(int rows, int cols, const PtrStepb& img1, const PtrStepb& img2, \r
+                               const PtrStepf& weights1, const PtrStepf& weights2, PtrStepb result, cudaStream_t stream);\r
 }}\r
 \r
 void cv::gpu::blendLinear(const GpuMat& img1, const GpuMat& img2, const GpuMat& weights1, const GpuMat& weights2, \r

diff --git a/modules/gpu/src/brute_force_matcher.cpp b/modules/gpu/src/brute_force_matcher.cpp
index c109d7f..4f9bd5c 100644 (file)
--- a/modules/gpu/src/brute_force_matcher.cpp
+++ b/modules/gpu/src/brute_force_matcher.cpp
@@ -84,71 +84,71 @@ void cv::gpu::BruteForceMatcher_GPU_base::radiusMatch(const GpuMat&, vector< vec
 \r
 namespace cv { namespace gpu { namespace bf_match\r
 {\r
-    template <typename T> void matchL1_gpu(const DevMem2D& query, const DevMem2D& train, const DevMem2D& mask, \r
+    template <typename T> void matchL1_gpu(const DevMem2Db& query, const DevMem2Db& train, const DevMem2Db& mask, \r
         const DevMem2Di& trainIdx, const DevMem2Df& distance, \r
         int cc, cudaStream_t stream);\r
-    template <typename T> void matchL2_gpu(const DevMem2D& query, const DevMem2D& train, const DevMem2D& mask, \r
+    template <typename T> void matchL2_gpu(const DevMem2Db& query, const DevMem2Db& train, const DevMem2Db& mask, \r
         const DevMem2Di& trainIdx, const DevMem2Df& distance, \r
         int cc, cudaStream_t stream);\r
-    template <typename T> void matchHamming_gpu(const DevMem2D& query, const DevMem2D& train, const DevMem2D& mask, \r
+    template <typename T> void matchHamming_gpu(const DevMem2Db& query, const DevMem2Db& train, const DevMem2Db& mask, \r
         const DevMem2Di& trainIdx, const DevMem2Df& distance, \r
         int cc, cudaStream_t stream);\r
 \r
-    template <typename T> void matchL1_gpu(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, \r
+    template <typename T> void matchL1_gpu(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, \r
         const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, \r
         int cc, cudaStream_t stream);\r
-    template <typename T> void matchL2_gpu(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, \r
+    template <typename T> void matchL2_gpu(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, \r
         const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, \r
         int cc, cudaStream_t stream);\r
-    template <typename T> void matchHamming_gpu(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, \r
+    template <typename T> void matchHamming_gpu(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, \r
         const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance,\r
         int cc, cudaStream_t stream);\r
 }}}\r
 \r
 namespace cv { namespace gpu { namespace bf_knnmatch\r
 {\r
-    template <typename T> void matchL1_gpu(const DevMem2D& query, const DevMem2D& train, int k, const DevMem2D& mask, \r
-        const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, \r
+    template <typename T> void matchL1_gpu(const DevMem2Db& query, const DevMem2Db& train, int k, const DevMem2Db& mask, \r
+        const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, \r
         int cc, cudaStream_t stream);\r
-    template <typename T> void matchL2_gpu(const DevMem2D& query, const DevMem2D& train, int k, const DevMem2D& mask, \r
-        const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, \r
+    template <typename T> void matchL2_gpu(const DevMem2Db& query, const DevMem2Db& train, int k, const DevMem2Db& mask, \r
+        const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, \r
         int cc, cudaStream_t stream);\r
-    template <typename T> void matchHamming_gpu(const DevMem2D& query, const DevMem2D& train, int k, const DevMem2D& mask, \r
-        const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, \r
+    template <typename T> void matchHamming_gpu(const DevMem2Db& query, const DevMem2Db& train, int k, const DevMem2Db& mask, \r
+        const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, \r
         int cc, cudaStream_t stream);\r
 \r
-    template <typename T> void match2L1_gpu(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, \r
-        const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, \r
+    template <typename T> void match2L1_gpu(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, \r
+        const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, \r
         int cc, cudaStream_t stream);\r
-    template <typename T> void match2L2_gpu(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, \r
-        const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, \r
+    template <typename T> void match2L2_gpu(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, \r
+        const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, \r
         int cc, cudaStream_t stream);\r
-    template <typename T> void match2Hamming_gpu(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, \r
-        const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, \r
+    template <typename T> void match2Hamming_gpu(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, \r
+        const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, \r
         int cc, cudaStream_t stream);\r
 }}}\r
 \r
 namespace cv { namespace gpu { namespace bf_radius_match \r
 {\r
-    template <typename T> void matchL1_gpu(const DevMem2D& query, const DevMem2D& train, float maxDistance, const DevMem2D& mask, \r
+    template <typename T> void matchL1_gpu(const DevMem2Db& query, const DevMem2Db& train, float maxDistance, const DevMem2Db& mask, \r
         const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, \r
         int cc, cudaStream_t stream);\r
-    template <typename T> void matchL2_gpu(const DevMem2D& query, const DevMem2D& train, float maxDistance, const DevMem2D& mask, \r
+    template <typename T> void matchL2_gpu(const DevMem2Db& query, const DevMem2Db& train, float maxDistance, const DevMem2Db& mask, \r
         const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, \r
         int cc, cudaStream_t stream);\r
-    template <typename T> void matchHamming_gpu(const DevMem2D& query, const DevMem2D& train, float maxDistance, const DevMem2D& mask, \r
+    template <typename T> void matchHamming_gpu(const DevMem2Db& query, const DevMem2Db& train, float maxDistance, const DevMem2Db& mask, \r
         const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, \r
         int cc, cudaStream_t stream);\r
 \r
-    template <typename T> void matchL1_gpu(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, \r
+    template <typename T> void matchL1_gpu(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, \r
         const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, \r
         int cc, cudaStream_t stream);\r
 \r
-    template <typename T> void matchL2_gpu(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, \r
+    template <typename T> void matchL2_gpu(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, \r
         const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, \r
         int cc, cudaStream_t stream);\r
 \r
-    template <typename T> void matchHamming_gpu(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, \r
+    template <typename T> void matchHamming_gpu(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, \r
         const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, \r
         int cc, cudaStream_t stream);\r
 }}}\r
@@ -197,7 +197,7 @@ void cv::gpu::BruteForceMatcher_GPU_base::matchSingle(const GpuMat& query, const
 \r
     using namespace cv::gpu::bf_match;\r
 \r
-    typedef void (*caller_t)(const DevMem2D& query, const DevMem2D& train, const DevMem2D& mask, \r
+    typedef void (*caller_t)(const DevMem2Db& query, const DevMem2Db& train, const DevMem2Db& mask, \r
                              const DevMem2Di& trainIdx, const DevMem2Df& distance,\r
                              int cc, cudaStream_t stream);\r
 \r
@@ -294,9 +294,9 @@ void cv::gpu::BruteForceMatcher_GPU_base::makeGpuCollection(GpuMat& trainCollect
 \r
     if (masks.empty())\r
     {\r
-        Mat trainCollectionCPU(1, static_cast<int>(trainDescCollection.size()), CV_8UC(sizeof(DevMem2D)));\r
+        Mat trainCollectionCPU(1, static_cast<int>(trainDescCollection.size()), CV_8UC(sizeof(DevMem2Db)));\r
 \r
-        DevMem2D* trainCollectionCPU_ptr = trainCollectionCPU.ptr<DevMem2D>();\r
+        DevMem2Db* trainCollectionCPU_ptr = trainCollectionCPU.ptr<DevMem2Db>();\r
 \r
         for (size_t i = 0, size = trainDescCollection.size(); i < size; ++i, ++trainCollectionCPU_ptr)\r
             *trainCollectionCPU_ptr = trainDescCollection[i];\r
@@ -308,11 +308,11 @@ void cv::gpu::BruteForceMatcher_GPU_base::makeGpuCollection(GpuMat& trainCollect
     {\r
         CV_Assert(masks.size() == trainDescCollection.size());\r
 \r
-        Mat trainCollectionCPU(1, static_cast<int>(trainDescCollection.size()), CV_8UC(sizeof(DevMem2D)));\r
-        Mat maskCollectionCPU(1, static_cast<int>(trainDescCollection.size()), CV_8UC(sizeof(PtrStep)));\r
+        Mat trainCollectionCPU(1, static_cast<int>(trainDescCollection.size()), CV_8UC(sizeof(DevMem2Db)));\r
+        Mat maskCollectionCPU(1, static_cast<int>(trainDescCollection.size()), CV_8UC(sizeof(PtrStepb)));\r
 \r
-        DevMem2D* trainCollectionCPU_ptr = trainCollectionCPU.ptr<DevMem2D>();\r
-        PtrStep* maskCollectionCPU_ptr = maskCollectionCPU.ptr<PtrStep>();\r
+        DevMem2Db* trainCollectionCPU_ptr = trainCollectionCPU.ptr<DevMem2Db>();\r
+        PtrStepb* maskCollectionCPU_ptr = maskCollectionCPU.ptr<PtrStepb>();\r
 \r
         for (size_t i = 0, size = trainDescCollection.size(); i < size; ++i, ++trainCollectionCPU_ptr, ++maskCollectionCPU_ptr)\r
         {\r
@@ -339,7 +339,7 @@ void cv::gpu::BruteForceMatcher_GPU_base::matchCollection(const GpuMat& query, c
 \r
     using namespace cv::gpu::bf_match;\r
 \r
-    typedef void (*caller_t)(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, \r
+    typedef void (*caller_t)(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, \r
                              const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, \r
                              int cc, cudaStream_t stream);\r
 \r
@@ -450,8 +450,8 @@ void cv::gpu::BruteForceMatcher_GPU_base::knnMatchSingle(const GpuMat& query, co
 \r
     using namespace cv::gpu::bf_knnmatch;\r
 \r
-    typedef void (*caller_t)(const DevMem2D& query, const DevMem2D& train, int k, const DevMem2D& mask, \r
-                             const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, \r
+    typedef void (*caller_t)(const DevMem2Db& query, const DevMem2Db& train, int k, const DevMem2Db& mask, \r
+                             const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, \r
                              int cc, cudaStream_t stream);\r
 \r
     static const caller_t callers[3][6] =\r
@@ -579,8 +579,8 @@ void cv::gpu::BruteForceMatcher_GPU_base::knnMatch2Collection(const GpuMat& quer
 \r
     using namespace cv::gpu::bf_knnmatch;\r
 \r
-    typedef void (*caller_t)(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, \r
-                             const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, \r
+    typedef void (*caller_t)(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, \r
+                             const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, \r
                              int cc, cudaStream_t stream);\r
 \r
     static const caller_t callers[3][6] =\r
@@ -760,7 +760,7 @@ void cv::gpu::BruteForceMatcher_GPU_base::radiusMatchSingle(const GpuMat& query,
 \r
     using namespace cv::gpu::bf_radius_match;\r
 \r
-    typedef void (*caller_t)(const DevMem2D& query, const DevMem2D& train, float maxDistance, const DevMem2D& mask, \r
+    typedef void (*caller_t)(const DevMem2Db& query, const DevMem2Db& train, float maxDistance, const DevMem2Db& mask, \r
                              const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, \r
                              int cc, cudaStream_t stream);\r
 \r
@@ -891,7 +891,7 @@ void cv::gpu::BruteForceMatcher_GPU_base::radiusMatchCollection(const GpuMat& qu
 \r
     using namespace cv::gpu::bf_radius_match;\r
 \r
-    typedef void (*caller_t)(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, \r
+    typedef void (*caller_t)(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, \r
                              const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, \r
                              int cc, cudaStream_t stream);\r
 \r
@@ -940,8 +940,8 @@ void cv::gpu::BruteForceMatcher_GPU_base::radiusMatchCollection(const GpuMat& qu
     caller_t func = callers[distType][query.depth()];\r
     CV_Assert(func != 0);\r
 \r
-    vector<DevMem2D> trains_(trainDescCollection.begin(), trainDescCollection.end());\r
-    vector<DevMem2D> masks_(masks.begin(), masks.end());\r
+    vector<DevMem2Db> trains_(trainDescCollection.begin(), trainDescCollection.end());\r
+    vector<DevMem2Db> masks_(masks.begin(), masks.end());\r
 \r
     func(query, &trains_[0], static_cast<int>(trains_.size()), maxDistance, masks_.size() == 0 ? 0 : &masks_[0], \r
         trainIdx, imgIdx, distance, nMatches, cc, StreamAccessor::getStream(stream));\r

diff --git a/modules/gpu/src/color.cpp b/modules/gpu/src/color.cpp
index 0f42d4d..69b0030 100644 (file)
--- a/modules/gpu/src/color.cpp
+++ b/modules/gpu/src/color.cpp
@@ -54,7 +54,7 @@ void cv::gpu::cvtColor(const GpuMat&, GpuMat&, int, int, Stream&) { throw_nogpu(
 namespace cv { namespace gpu {  namespace device  \r
 {\r
     #define OPENCV_GPU_DECLARE_CVTCOLOR_ONE(name) \\r
-        void name(const DevMem2D& src, const DevMem2D& dst, cudaStream_t stream);\r
+        void name(const DevMem2Db& src, const DevMem2Db& dst, cudaStream_t stream);\r
 \r
     #define OPENCV_GPU_DECLARE_CVTCOLOR_ALL(name) \\r
         OPENCV_GPU_DECLARE_CVTCOLOR_ONE(name ## _8u) \\r
@@ -203,7 +203,7 @@ namespace cv { namespace gpu {  namespace device
 \r
 namespace\r
 {\r
-    typedef void (*gpu_func_t)(const DevMem2D& src, const DevMem2D& dst, cudaStream_t stream);\r
+    typedef void (*gpu_func_t)(const DevMem2Db& src, const DevMem2Db& dst, cudaStream_t stream);\r
 \r
     void bgr_to_rgb(const GpuMat& src, GpuMat& dst, int, Stream& stream)\r
     {\r

diff --git a/modules/gpu/src/cuda/bf_knnmatch.cu b/modules/gpu/src/cuda/bf_knnmatch.cu
index 443dfa6..fb7004e 100644 (file)
--- a/modules/gpu/src/cuda/bf_knnmatch.cu
+++ b/modules/gpu/src/cuda/bf_knnmatch.cu
@@ -731,7 +731,7 @@ namespace cv { namespace gpu { namespace bf_knnmatch
 \r
     template <typename Dist, typename T, typename Mask> \r
     void match2Dispatcher(const DevMem2D_<T>& query, const DevMem2D_<T>& train, const Mask& mask, \r
-                          const DevMem2D& trainIdx, const DevMem2D& distance, \r
+                          const DevMem2Db& trainIdx, const DevMem2Db& distance, \r
                           int cc, cudaStream_t stream)\r
     {\r
         if (query.cols <= 64)\r
@@ -762,7 +762,7 @@ namespace cv { namespace gpu { namespace bf_knnmatch
 \r
     template <typename Dist, typename T, typename Mask> \r
     void match2Dispatcher(const DevMem2D_<T>& query, const DevMem2D_<T>* trains, int n, const Mask& mask, \r
-                          const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, \r
+                          const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, \r
                           int cc, cudaStream_t stream)\r
     {\r
         if (query.cols <= 64)\r
@@ -1015,7 +1015,7 @@ namespace cv { namespace gpu { namespace bf_knnmatch
             cudaSafeCall( cudaDeviceSynchronize() );\r
     }\r
 \r
-    void findKnnMatchDispatcher(int k, const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream)\r
+    void findKnnMatchDispatcher(int k, const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream)\r
     {\r
         findKnnMatch<256>(k, static_cast<DevMem2Di>(trainIdx), static_cast<DevMem2Df>(distance), allDist, stream);\r
     }\r
@@ -1025,7 +1025,7 @@ namespace cv { namespace gpu { namespace bf_knnmatch
 \r
     template <typename Dist, typename T, typename Mask>\r
     void matchDispatcher(const DevMem2D_<T>& query, const DevMem2D_<T>& train, int k, const Mask& mask, \r
-        const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, \r
+        const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, \r
         int cc, cudaStream_t stream)\r
     {\r
         if (k == 2)\r
@@ -1042,8 +1042,8 @@ namespace cv { namespace gpu { namespace bf_knnmatch
     ///////////////////////////////////////////////////////////////////////////////\r
     // knn match caller\r
 \r
-    template <typename T> void matchL1_gpu(const DevMem2D& query, const DevMem2D& train, int k, const DevMem2D& mask, \r
-        const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, \r
+    template <typename T> void matchL1_gpu(const DevMem2Db& query, const DevMem2Db& train, int k, const DevMem2Db& mask, \r
+        const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, \r
         int cc, cudaStream_t stream)\r
     {\r
         if (mask.data)\r
@@ -1052,15 +1052,15 @@ namespace cv { namespace gpu { namespace bf_knnmatch
             matchDispatcher< L1Dist<T> >(static_cast< DevMem2D_<T> >(query), static_cast< DevMem2D_<T> >(train), k, WithOutMask(), trainIdx, distance, allDist, cc, stream);\r
     }\r
 \r
-    template void matchL1_gpu<uchar >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, int k, const DevMem2D& mask, const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
-    //template void matchL1_gpu<schar >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, int k, const DevMem2D& mask, const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<ushort>(const DevMem2D& queryDescs, const DevMem2D& trainDescs, int k, const DevMem2D& mask, const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<short >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, int k, const DevMem2D& mask, const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<int   >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, int k, const DevMem2D& mask, const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<float >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, int k, const DevMem2D& mask, const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<uchar >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, int k, const DevMem2Db& mask, const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
+    //template void matchL1_gpu<schar >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, int k, const DevMem2Db& mask, const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<ushort>(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, int k, const DevMem2Db& mask, const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<short >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, int k, const DevMem2Db& mask, const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<int   >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, int k, const DevMem2Db& mask, const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<float >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, int k, const DevMem2Db& mask, const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
 \r
-    template <typename T> void matchL2_gpu(const DevMem2D& query, const DevMem2D& train, int k, const DevMem2D& mask, \r
-        const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist,\r
+    template <typename T> void matchL2_gpu(const DevMem2Db& query, const DevMem2Db& train, int k, const DevMem2Db& mask, \r
+        const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist,\r
         int cc, cudaStream_t stream)\r
     {\r
         if (mask.data)\r
@@ -1069,15 +1069,15 @@ namespace cv { namespace gpu { namespace bf_knnmatch
             matchDispatcher<L2Dist>(static_cast< DevMem2D_<T> >(query), static_cast< DevMem2D_<T> >(train), k, WithOutMask(), trainIdx, distance, allDist, cc, stream);\r
     }\r
 \r
-    //template void matchL2_gpu<uchar >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, int k, const DevMem2D& mask, const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<schar >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, int k, const DevMem2D& mask, const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<ushort>(const DevMem2D& queryDescs, const DevMem2D& trainDescs, int k, const DevMem2D& mask, const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<short >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, int k, const DevMem2D& mask, const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<int   >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, int k, const DevMem2D& mask, const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
-    template void matchL2_gpu<float >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, int k, const DevMem2D& mask, const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<uchar >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, int k, const DevMem2Db& mask, const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<schar >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, int k, const DevMem2Db& mask, const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<ushort>(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, int k, const DevMem2Db& mask, const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<short >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, int k, const DevMem2Db& mask, const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<int   >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, int k, const DevMem2Db& mask, const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
+    template void matchL2_gpu<float >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, int k, const DevMem2Db& mask, const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
 \r
-    template <typename T> void matchHamming_gpu(const DevMem2D& query, const DevMem2D& train, int k, const DevMem2D& mask,\r
-        const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, \r
+    template <typename T> void matchHamming_gpu(const DevMem2Db& query, const DevMem2Db& train, int k, const DevMem2Db& mask,\r
+        const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, \r
         int cc, cudaStream_t stream)\r
     {\r
         if (mask.data)\r
@@ -1086,14 +1086,14 @@ namespace cv { namespace gpu { namespace bf_knnmatch
             matchDispatcher<HammingDist>(static_cast< DevMem2D_<T> >(query), static_cast< DevMem2D_<T> >(train), k, WithOutMask(), trainIdx, distance, allDist, cc, stream);\r
     }\r
 \r
-    template void matchHamming_gpu<uchar >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, int k, const DevMem2D& mask, const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
-    //template void matchHamming_gpu<schar >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, int k, const DevMem2D& mask, const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
-    template void matchHamming_gpu<ushort>(const DevMem2D& queryDescs, const DevMem2D& trainDescs, int k, const DevMem2D& mask, const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
-    //template void matchHamming_gpu<short >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, int k, const DevMem2D& mask, const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
-    template void matchHamming_gpu<int   >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, int k, const DevMem2D& mask, const DevMem2D& trainIdx, const DevMem2D& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
+    template void matchHamming_gpu<uchar >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, int k, const DevMem2Db& mask, const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
+    //template void matchHamming_gpu<schar >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, int k, const DevMem2Db& mask, const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
+    template void matchHamming_gpu<ushort>(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, int k, const DevMem2Db& mask, const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
+    //template void matchHamming_gpu<short >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, int k, const DevMem2Db& mask, const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
+    template void matchHamming_gpu<int   >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, int k, const DevMem2Db& mask, const DevMem2Db& trainIdx, const DevMem2Db& distance, const DevMem2Df& allDist, int cc, cudaStream_t stream);\r
 \r
-    template <typename T> void match2L1_gpu(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, \r
-        const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, \r
+    template <typename T> void match2L1_gpu(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, \r
+        const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, \r
         int cc, cudaStream_t stream)\r
     {\r
         if (masks.data)\r
@@ -1102,15 +1102,15 @@ namespace cv { namespace gpu { namespace bf_knnmatch
             match2Dispatcher< L1Dist<T> >(static_cast< DevMem2D_<T> >(query), (const DevMem2D_<T>*)trains.ptr(), trains.cols, WithOutMask(), trainIdx, imgIdx, distance, cc, stream);\r
     }\r
     \r
-    template void match2L1_gpu<uchar >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, int cc, cudaStream_t stream);\r
-    //template void match2L1_gpu<schar >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, int cc, cudaStream_t stream);\r
-    template void match2L1_gpu<ushort>(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, int cc, cudaStream_t stream);\r
-    template void match2L1_gpu<short >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, int cc, cudaStream_t stream);\r
-    template void match2L1_gpu<int   >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, int cc, cudaStream_t stream);\r
-    template void match2L1_gpu<float >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, int cc, cudaStream_t stream);\r
-\r
-    template <typename T> void match2L2_gpu(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, \r
-        const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, \r
+    template void match2L1_gpu<uchar >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, int cc, cudaStream_t stream);\r
+    //template void match2L1_gpu<schar >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, int cc, cudaStream_t stream);\r
+    template void match2L1_gpu<ushort>(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, int cc, cudaStream_t stream);\r
+    template void match2L1_gpu<short >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, int cc, cudaStream_t stream);\r
+    template void match2L1_gpu<int   >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, int cc, cudaStream_t stream);\r
+    template void match2L1_gpu<float >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, int cc, cudaStream_t stream);\r
+\r
+    template <typename T> void match2L2_gpu(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, \r
+        const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, \r
         int cc, cudaStream_t stream)\r
     {\r
         if (masks.data)\r
@@ -1119,15 +1119,15 @@ namespace cv { namespace gpu { namespace bf_knnmatch
             match2Dispatcher<L2Dist>(static_cast< DevMem2D_<T> >(query), (const DevMem2D_<T>*)trains.ptr(), trains.cols, WithOutMask(), trainIdx, imgIdx, distance, cc, stream);\r
     }\r
     \r
-    //template void match2L2_gpu<uchar >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, int cc, cudaStream_t stream);\r
-    //template void match2L2_gpu<schar >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, int cc, cudaStream_t stream);\r
-    //template void match2L2_gpu<ushort>(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, int cc, cudaStream_t stream);\r
-    //template void match2L2_gpu<short >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, int cc, cudaStream_t stream);\r
-    //template void match2L2_gpu<int   >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2D& trainIdx, const DevMem2Di& imgIdx, const DevMem2D& distance, int cc, cudaStream_t stream);\r
-    template void match2L2_gpu<float >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, int cc, cudaStream_t stream);\r
+    //template void match2L2_gpu<uchar >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, int cc, cudaStream_t stream);\r
+    //template void match2L2_gpu<schar >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, int cc, cudaStream_t stream);\r
+    //template void match2L2_gpu<ushort>(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, int cc, cudaStream_t stream);\r
+    //template void match2L2_gpu<short >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, int cc, cudaStream_t stream);\r
+    //template void match2L2_gpu<int   >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Db& trainIdx, const DevMem2Di& imgIdx, const DevMem2Db& distance, int cc, cudaStream_t stream);\r
+    template void match2L2_gpu<float >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, int cc, cudaStream_t stream);\r
     \r
-    template <typename T> void match2Hamming_gpu(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, \r
-        const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, \r
+    template <typename T> void match2Hamming_gpu(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, \r
+        const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, \r
         int cc, cudaStream_t stream)\r
     {\r
         if (masks.data)\r
@@ -1136,9 +1136,9 @@ namespace cv { namespace gpu { namespace bf_knnmatch
             match2Dispatcher<HammingDist>(static_cast< DevMem2D_<T> >(query), (const DevMem2D_<T>*)trains.ptr(), trains.cols, WithOutMask(), trainIdx, imgIdx, distance, cc, stream);\r
     }\r
     \r
-    template void match2Hamming_gpu<uchar >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, int cc, cudaStream_t stream);\r
-    //template void match2Hamming_gpu<schar >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, int cc, cudaStream_t stream);\r
-    template void match2Hamming_gpu<ushort>(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, int cc, cudaStream_t stream);\r
-    //template void match2Hamming_gpu<short >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, int cc, cudaStream_t stream);\r
-    template void match2Hamming_gpu<int   >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2D& trainIdx, const DevMem2D& imgIdx, const DevMem2D& distance, int cc, cudaStream_t stream);\r
+    template void match2Hamming_gpu<uchar >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, int cc, cudaStream_t stream);\r
+    //template void match2Hamming_gpu<schar >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, int cc, cudaStream_t stream);\r
+    template void match2Hamming_gpu<ushort>(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, int cc, cudaStream_t stream);\r
+    //template void match2Hamming_gpu<short >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, int cc, cudaStream_t stream);\r
+    template void match2Hamming_gpu<int   >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Db& trainIdx, const DevMem2Db& imgIdx, const DevMem2Db& distance, int cc, cudaStream_t stream);\r
 }}}\r

diff --git a/modules/gpu/src/cuda/bf_match.cu b/modules/gpu/src/cuda/bf_match.cu
index e3372b3..2c7f74a 100644 (file)
--- a/modules/gpu/src/cuda/bf_match.cu
+++ b/modules/gpu/src/cuda/bf_match.cu
@@ -613,7 +613,7 @@ namespace cv { namespace gpu { namespace bf_match
     ///////////////////////////////////////////////////////////////////////////////\r
     // Match caller\r
 \r
-    template <typename T> void matchL1_gpu(const DevMem2D& query, const DevMem2D& train, const DevMem2D& mask, \r
+    template <typename T> void matchL1_gpu(const DevMem2Db& query, const DevMem2Db& train, const DevMem2Db& mask, \r
                                            const DevMem2Di& trainIdx, const DevMem2Df& distance,\r
                                            int cc, cudaStream_t stream)\r
     {\r
@@ -631,14 +631,14 @@ namespace cv { namespace gpu { namespace bf_match
         }\r
     }\r
 \r
-    template void matchL1_gpu<uchar >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    //template void matchL1_gpu<schar >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<ushort>(const DevMem2D& queryDescs, const DevMem2D& trainDescs, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<short >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<int   >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<float >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<uchar >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    //template void matchL1_gpu<schar >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<ushort>(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<short >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<int   >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<float >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
 \r
-    template <typename T> void matchL2_gpu(const DevMem2D& query, const DevMem2D& train, const DevMem2D& mask, \r
+    template <typename T> void matchL2_gpu(const DevMem2Db& query, const DevMem2Db& train, const DevMem2Db& mask, \r
                                            const DevMem2Di& trainIdx, const DevMem2Df& distance, \r
                                            int cc, cudaStream_t stream)\r
     {\r
@@ -656,14 +656,14 @@ namespace cv { namespace gpu { namespace bf_match
         }\r
     }\r
 \r
-    //template void matchL2_gpu<uchar >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<schar >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<ushort>(const DevMem2D& queryDescs, const DevMem2D& trainDescs, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<short >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<int   >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    template void matchL2_gpu<float >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<uchar >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<schar >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<ushort>(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<short >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<int   >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    template void matchL2_gpu<float >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
 \r
-    template <typename T> void matchHamming_gpu(const DevMem2D& query, const DevMem2D& train, const DevMem2D& mask, \r
+    template <typename T> void matchHamming_gpu(const DevMem2Db& query, const DevMem2Db& train, const DevMem2Db& mask, \r
                                                 const DevMem2Di& trainIdx, const DevMem2Df& distance, \r
                                                 int cc, cudaStream_t stream)\r
     {\r
@@ -681,13 +681,13 @@ namespace cv { namespace gpu { namespace bf_match
         }\r
     }\r
 \r
-    template void matchHamming_gpu<uchar >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    //template void matchHamming_gpu<schar >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    template void matchHamming_gpu<ushort>(const DevMem2D& queryDescs, const DevMem2D& trainDescs, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    //template void matchHamming_gpu<short >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    template void matchHamming_gpu<int   >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    template void matchHamming_gpu<uchar >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    //template void matchHamming_gpu<schar >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    template void matchHamming_gpu<ushort>(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    //template void matchHamming_gpu<short >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    template void matchHamming_gpu<int   >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
 \r
-    template <typename T> void matchL1_gpu(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, \r
+    template <typename T> void matchL1_gpu(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, \r
                                            const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, \r
                                            int cc, cudaStream_t stream)\r
     {\r
@@ -705,14 +705,14 @@ namespace cv { namespace gpu { namespace bf_match
         }\r
     }\r
 \r
-    template void matchL1_gpu<uchar >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    //template void matchL1_gpu<schar >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<ushort>(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<short >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<int   >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<float >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<uchar >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    //template void matchL1_gpu<schar >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<ushort>(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<short >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<int   >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<float >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
 \r
-    template <typename T> void matchL2_gpu(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, \r
+    template <typename T> void matchL2_gpu(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, \r
                                            const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, \r
                                            int cc, cudaStream_t stream)\r
     {\r
@@ -730,14 +730,14 @@ namespace cv { namespace gpu { namespace bf_match
         }\r
     }\r
 \r
-    //template void matchL2_gpu<uchar >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<schar >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<ushort>(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<short >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<int   >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    template void matchL2_gpu<float >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& maskCollection, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<uchar >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<schar >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<ushort>(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<short >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<int   >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    template void matchL2_gpu<float >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& maskCollection, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
 \r
-    template <typename T> void matchHamming_gpu(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, \r
+    template <typename T> void matchHamming_gpu(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, \r
                                                 const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, \r
                                                 int cc, cudaStream_t stream)\r
     {\r
@@ -755,9 +755,9 @@ namespace cv { namespace gpu { namespace bf_match
         }\r
     }\r
 \r
-    template void matchHamming_gpu<uchar >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    //template void matchHamming_gpu<schar >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    template void matchHamming_gpu<ushort>(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    //template void matchHamming_gpu<short >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
-    template void matchHamming_gpu<int   >(const DevMem2D& query, const DevMem2D& trains, const DevMem2D_<PtrStep>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    template void matchHamming_gpu<uchar >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    //template void matchHamming_gpu<schar >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    template void matchHamming_gpu<ushort>(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    //template void matchHamming_gpu<short >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
+    template void matchHamming_gpu<int   >(const DevMem2Db& query, const DevMem2Db& trains, const DevMem2D_<PtrStepb>& masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, int cc, cudaStream_t stream);\r
 }}}\r

diff --git a/modules/gpu/src/cuda/bf_radius_match.cu b/modules/gpu/src/cuda/bf_radius_match.cu
index 6dbaf85..4cef2bc 100644 (file)
--- a/modules/gpu/src/cuda/bf_radius_match.cu
+++ b/modules/gpu/src/cuda/bf_radius_match.cu
@@ -127,7 +127,7 @@ namespace cv { namespace gpu { namespace bf_radius_match
     }   \r
 \r
     template <int BLOCK_SIZE, int MAX_DESC_LEN, typename Dist, typename T> \r
-    void matchUnrolled(const DevMem2D_<T>& query, const DevMem2D_<T>* trains, int n, float maxDistance, const DevMem2D* masks, \r
+    void matchUnrolled(const DevMem2D_<T>& query, const DevMem2D_<T>* trains, int n, float maxDistance, const DevMem2Db* masks, \r
         const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, \r
         cudaStream_t stream)\r
     {\r
@@ -236,7 +236,7 @@ namespace cv { namespace gpu { namespace bf_radius_match
     }\r
 \r
     template <int BLOCK_SIZE, typename Dist, typename T> \r
-    void match(const DevMem2D_<T>& query, const DevMem2D_<T>* trains, int n, float maxDistance, const DevMem2D* masks, \r
+    void match(const DevMem2D_<T>& query, const DevMem2D_<T>* trains, int n, float maxDistance, const DevMem2Db* masks, \r
         const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, \r
         cudaStream_t stream)\r
     {\r
@@ -302,7 +302,7 @@ namespace cv { namespace gpu { namespace bf_radius_match
     }\r
 \r
     template <typename Dist, typename T> \r
-    void matchDispatcher(const DevMem2D_<T>& query, const DevMem2D_<T>* trains, int n, float maxDistance, const DevMem2D* masks, \r
+    void matchDispatcher(const DevMem2D_<T>& query, const DevMem2D_<T>* trains, int n, float maxDistance, const DevMem2Db* masks, \r
                          const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, \r
                          int cc, cudaStream_t stream)\r
     {\r
@@ -335,7 +335,7 @@ namespace cv { namespace gpu { namespace bf_radius_match
     ///////////////////////////////////////////////////////////////////////////////\r
     // Radius Match caller\r
 \r
-    template <typename T> void matchL1_gpu(const DevMem2D& query, const DevMem2D& train, float maxDistance, const DevMem2D& mask, \r
+    template <typename T> void matchL1_gpu(const DevMem2Db& query, const DevMem2Db& train, float maxDistance, const DevMem2Db& mask, \r
         const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, \r
         int cc, cudaStream_t stream)\r
     {\r
@@ -353,14 +353,14 @@ namespace cv { namespace gpu { namespace bf_radius_match
         }\r
     }\r
 \r
-    template void matchL1_gpu<uchar >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, float maxDistance, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    //template void matchL1_gpu<schar >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, float maxDistance, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<ushort>(const DevMem2D& queryDescs, const DevMem2D& trainDescs, float maxDistance, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<short >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, float maxDistance, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<int   >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, float maxDistance, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<float >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, float maxDistance, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<uchar >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, float maxDistance, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    //template void matchL1_gpu<schar >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, float maxDistance, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<ushort>(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, float maxDistance, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<short >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, float maxDistance, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<int   >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, float maxDistance, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<float >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, float maxDistance, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
 \r
-    template <typename T> void matchL2_gpu(const DevMem2D& query, const DevMem2D& train, float maxDistance, const DevMem2D& mask, \r
+    template <typename T> void matchL2_gpu(const DevMem2Db& query, const DevMem2Db& train, float maxDistance, const DevMem2Db& mask, \r
         const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, \r
         int cc, cudaStream_t stream)\r
     {\r
@@ -378,14 +378,14 @@ namespace cv { namespace gpu { namespace bf_radius_match
         }\r
     }\r
 \r
-    //template void matchL2_gpu<uchar >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, float maxDistance, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<schar >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, float maxDistance, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<ushort>(const DevMem2D& queryDescs, const DevMem2D& trainDescs, float maxDistance, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<short >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, float maxDistance, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<int   >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, float maxDistance, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    template void matchL2_gpu<float >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, float maxDistance, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<uchar >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, float maxDistance, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<schar >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, float maxDistance, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<ushort>(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, float maxDistance, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<short >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, float maxDistance, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<int   >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, float maxDistance, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    template void matchL2_gpu<float >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, float maxDistance, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
 \r
-    template <typename T> void matchHamming_gpu(const DevMem2D& query, const DevMem2D& train, float maxDistance, const DevMem2D& mask, \r
+    template <typename T> void matchHamming_gpu(const DevMem2Db& query, const DevMem2Db& train, float maxDistance, const DevMem2Db& mask, \r
         const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, \r
         int cc, cudaStream_t stream)\r
     {\r
@@ -403,13 +403,13 @@ namespace cv { namespace gpu { namespace bf_radius_match
         }\r
     }\r
 \r
-    template void matchHamming_gpu<uchar >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, float maxDistance, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    //template void matchHamming_gpu<schar >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, float maxDistance, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    template void matchHamming_gpu<ushort>(const DevMem2D& queryDescs, const DevMem2D& trainDescs, float maxDistance, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    //template void matchHamming_gpu<short >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, float maxDistance, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    template void matchHamming_gpu<int   >(const DevMem2D& queryDescs, const DevMem2D& trainDescs, float maxDistance, const DevMem2D& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    template void matchHamming_gpu<uchar >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, float maxDistance, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    //template void matchHamming_gpu<schar >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, float maxDistance, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    template void matchHamming_gpu<ushort>(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, float maxDistance, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    //template void matchHamming_gpu<short >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, float maxDistance, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    template void matchHamming_gpu<int   >(const DevMem2Db& queryDescs, const DevMem2Db& trainDescs, float maxDistance, const DevMem2Db& mask, const DevMem2Di& trainIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
 \r
-    template <typename T> void matchL1_gpu(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, \r
+    template <typename T> void matchL1_gpu(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, \r
         const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, \r
         int cc, cudaStream_t stream)\r
     {\r
@@ -418,14 +418,14 @@ namespace cv { namespace gpu { namespace bf_radius_match
             cc, stream);\r
     }\r
 \r
-    template void matchL1_gpu<uchar >(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    //template void matchL1_gpu<schar >(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<ushort>(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<short >(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<int   >(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    template void matchL1_gpu<float >(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<uchar >(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    //template void matchL1_gpu<schar >(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<ushort>(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<short >(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<int   >(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    template void matchL1_gpu<float >(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
 \r
-    template <typename T> void matchL2_gpu(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, \r
+    template <typename T> void matchL2_gpu(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, \r
         const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, \r
         int cc, cudaStream_t stream)\r
     {\r
@@ -434,14 +434,14 @@ namespace cv { namespace gpu { namespace bf_radius_match
             cc, stream);\r
     }\r
 \r
-    //template void matchL2_gpu<uchar >(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<schar >(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<ushort>(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<short >(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    //template void matchL2_gpu<int   >(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    template void matchL2_gpu<float >(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<uchar >(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<schar >(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<ushort>(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<short >(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    //template void matchL2_gpu<int   >(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    template void matchL2_gpu<float >(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
 \r
-    template <typename T> void matchHamming_gpu(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, \r
+    template <typename T> void matchHamming_gpu(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, \r
         const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, \r
         int cc, cudaStream_t stream)\r
     {\r
@@ -450,9 +450,9 @@ namespace cv { namespace gpu { namespace bf_radius_match
             cc, stream);\r
     }\r
 \r
-    template void matchHamming_gpu<uchar >(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    //template void matchHamming_gpu<schar >(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    template void matchHamming_gpu<ushort>(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    //template void matchHamming_gpu<short >(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
-    template void matchHamming_gpu<int   >(const DevMem2D& query, const DevMem2D* trains, int n, float maxDistance, const DevMem2D* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    template void matchHamming_gpu<uchar >(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    //template void matchHamming_gpu<schar >(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    template void matchHamming_gpu<ushort>(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    //template void matchHamming_gpu<short >(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
+    template void matchHamming_gpu<int   >(const DevMem2Db& query, const DevMem2Db* trains, int n, float maxDistance, const DevMem2Db* masks, const DevMem2Di& trainIdx, const DevMem2Di& imgIdx, const DevMem2Df& distance, const DevMem2D_<unsigned int>& nMatches, int cc, cudaStream_t stream);\r
 }}}\r

diff --git a/modules/gpu/src/cuda/bilateral_filter.cu b/modules/gpu/src/cuda/bilateral_filter.cu
index ae5d801..173c156 100644 (file)
--- a/modules/gpu/src/cuda/bilateral_filter.cu
+++ b/modules/gpu/src/cuda/bilateral_filter.cu
@@ -186,7 +186,7 @@ namespace bf_krnls
 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
+    void bilateral_filter_caller(const DevMem2D_<T>& disp, const DevMem2Db& img, int channels, int iters, cudaStream_t stream)\r
     {\r
         dim3 threads(32, 8, 1);\r
         dim3 grid(1, 1, 1);\r
@@ -221,12 +221,12 @@ namespace cv { namespace gpu { namespace bf
             cudaSafeCall( cudaDeviceSynchronize() );\r
     }\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 DevMem2Db& disp, const DevMem2Db& 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
+    void bilateral_filter_gpu(const DevMem2D_<short>& disp, const DevMem2Db& img, int channels, int iters, cudaStream_t stream)\r
     {\r
         bilateral_filter_caller(disp, img, channels, iters, stream);\r
     }\r

diff --git a/modules/gpu/src/cuda/blend.cu b/modules/gpu/src/cuda/blend.cu
index bed0ee0..4b29a70 100644 (file)
--- a/modules/gpu/src/cuda/blend.cu
+++ b/modules/gpu/src/cuda/blend.cu
@@ -48,8 +48,8 @@ namespace cv { namespace gpu
 {\r
 \r
     template <typename T>\r
-    __global__ void blendLinearKernel(int rows, int cols, int cn, const PtrStep_<T> img1, const PtrStep_<T> img2,\r
-                                      const PtrStepf weights1, const PtrStepf weights2, PtrStep_<T> result)\r
+    __global__ void blendLinearKernel(int rows, int cols, int cn, const PtrStep<T> img1, const PtrStep<T> img2,\r
+                                      const PtrStepf weights1, const PtrStepf weights2, PtrStep<T> result)\r
     {\r
         int x = blockIdx.x * blockDim.x + threadIdx.x;\r
         int y = blockIdx.y * blockDim.y + threadIdx.y;\r
@@ -63,12 +63,11 @@ namespace cv { namespace gpu
             T p2 = img2.ptr(y)[x];\r
             result.ptr(y)[x] = (p1 * w1 + p2 * w2) / (w1 + w2 + 1e-5f);\r
         }\r
-    }\r
-\r
+    }  \r
 \r
     template <typename T>\r
-    void blendLinearCaller(int rows, int cols, int cn, const PtrStep_<T> img1, const PtrStep_<T> img2, \r
-                           const PtrStepf weights1, const PtrStepf weights2, PtrStep_<T> result, cudaStream_t stream)\r
+    void blendLinearCaller(int rows, int cols, int cn, const PtrStep<T>& img1, const PtrStep<T>& img2, \r
+                           const PtrStepf& weights1, const PtrStepf& weights2, PtrStep<T> result, cudaStream_t stream)\r
     {\r
         dim3 threads(16, 16);\r
         dim3 grid(divUp(cols * cn, threads.x), divUp(rows, threads.y));\r
@@ -80,14 +79,14 @@ namespace cv { namespace gpu
             cudaSafeCall(cudaDeviceSynchronize());\r
     }\r
 \r
-    template void blendLinearCaller<uchar>(int, int, int, const PtrStep, const PtrStep, \r
-                                           const PtrStepf, const PtrStepf, PtrStep, cudaStream_t stream);\r
-    template void blendLinearCaller<float>(int, int, int, const PtrStepf, const PtrStepf, \r
-                                           const PtrStepf, const PtrStepf, PtrStepf, cudaStream_t stream);\r
+    template void blendLinearCaller<uchar>(int, int, int, const PtrStep<uchar>&, const PtrStep<uchar>&, \r
+                                           const PtrStepf&, const PtrStepf&, PtrStep<uchar>, cudaStream_t stream);\r
+    template void blendLinearCaller<float>(int, int, int, const PtrStep<float>&, const PtrStep<float>&, \r
+                                           const PtrStepf&, const PtrStepf&, PtrStep<float>, cudaStream_t stream);\r
 \r
 \r
-    __global__ void blendLinearKernel8UC4(int rows, int cols, const PtrStep img1, const PtrStep img2,\r
-                                          const PtrStepf weights1, const PtrStepf weights2, PtrStep result)\r
+    __global__ void blendLinearKernel8UC4(int rows, int cols, const PtrStepb img1, const PtrStepb img2,\r
+                                          const PtrStepf weights1, const PtrStepf weights2, PtrStepb result)\r
     {\r
         int x = blockIdx.x * blockDim.x + threadIdx.x;\r
         int y = blockIdx.y * blockDim.y + threadIdx.y;\r
@@ -107,8 +106,8 @@ namespace cv { namespace gpu
     }\r
 \r
 \r
-    void blendLinearCaller8UC4(int rows, int cols, const PtrStep img1, const PtrStep img2, \r
-                               const PtrStepf weights1, const PtrStepf weights2, PtrStep result, cudaStream_t stream)\r
+    void blendLinearCaller8UC4(int rows, int cols, const PtrStepb& img1, const PtrStepb& img2, \r
+                               const PtrStepf& weights1, const PtrStepf& weights2, PtrStepb result, cudaStream_t stream)\r
     {\r
         dim3 threads(16, 16);\r
         dim3 grid(divUp(cols, threads.x), divUp(rows, threads.y));\r

diff --git a/modules/gpu/src/cuda/canny.cu b/modules/gpu/src/cuda/canny.cu
index 95d1b98..4ea26fc 100644 (file)
--- a/modules/gpu/src/cuda/canny.cu
+++ b/modules/gpu/src/cuda/canny.cu
@@ -48,7 +48,7 @@ using namespace cv::gpu;
 \r
 namespace cv { namespace gpu { namespace canny\r
 {\r
-    __global__ void calcSobelRowPass(const PtrStep src, PtrStepi dx_buf, PtrStepi dy_buf, int rows, int cols)\r
+    __global__ void calcSobelRowPass(const PtrStepb src, PtrStepi dx_buf, PtrStepi dy_buf, int rows, int cols)\r
     {\r
         __shared__ int smem[16][18];\r
 \r
@@ -73,7 +73,7 @@ namespace cv { namespace gpu { namespace canny
         }\r
     }\r
 \r
-    void calcSobelRowPass_gpu(PtrStep src, PtrStepi dx_buf, PtrStepi dy_buf, int rows, int cols)\r
+    void calcSobelRowPass_gpu(PtrStepb src, PtrStepi dx_buf, PtrStepi dy_buf, int rows, int cols)\r
     {\r
         dim3 block(16, 16, 1);\r
         dim3 grid(divUp(cols, block.x), divUp(rows, block.y), 1);\r
@@ -468,7 +468,7 @@ namespace cv { namespace gpu { namespace canny
         }\r
     }\r
 \r
-    __global__ void getEdges(PtrStepi map, PtrStep dst, int rows, int cols)\r
+    __global__ void getEdges(PtrStepi map, PtrStepb dst, int rows, int cols)\r
     {\r
         const int j = blockIdx.x * 16 + threadIdx.x;\r
         const int i = blockIdx.y * 16 + threadIdx.y;\r
@@ -477,7 +477,7 @@ namespace cv { namespace gpu { namespace canny
             dst.ptr(i)[j] = (uchar)(-(map.ptr(i + 1)[j + 1] >> 1));\r
     }\r
 \r
-    void getEdges_gpu(PtrStepi map, PtrStep dst, int rows, int cols)\r
+    void getEdges_gpu(PtrStepi map, PtrStepb dst, int rows, int cols)\r
     {\r
         dim3 block(16, 16, 1);\r
         dim3 grid(divUp(cols, block.x), divUp(rows, block.y), 1);\r

diff --git a/modules/gpu/src/cuda/color.cu b/modules/gpu/src/cuda/color.cu
index 71f84e3..dfa7429 100644 (file)
--- a/modules/gpu/src/cuda/color.cu
+++ b/modules/gpu/src/cuda/color.cu
@@ -221,7 +221,7 @@ namespace cv { namespace gpu { namespace device
     };\r
 \r
     #define OPENCV_GPU_IMPLEMENT_CVTCOLOR(name, traits) \\r
-        void name(const DevMem2D& src, const DevMem2D& dst, cudaStream_t stream) \\r
+        void name(const DevMem2Db& src, const DevMem2Db& dst, cudaStream_t stream) \\r
         { \\r
             traits::functor_type functor = traits::create_functor(); \\r
             typedef typename traits::functor_type::argument_type src_t; \\r

diff --git a/modules/gpu/src/cuda/column_filter.cu b/modules/gpu/src/cuda/column_filter.cu
index 14146df..29b828e 100644 (file)
--- a/modules/gpu/src/cuda/column_filter.cu
+++ b/modules/gpu/src/cuda/column_filter.cu
@@ -66,7 +66,7 @@ namespace filter_column
     }\r
 \r
     template <int KERNEL_SIZE, typename T, typename D, typename B>\r
-    __global__ void linearColumnFilter(const DevMem2D_<T> src, PtrStep_<D> dst, int anchor, const B b)\r
+    __global__ void linearColumnFilter(const DevMem2D_<T> src, PtrStep<D> dst, int anchor, const B b)\r
     {\r
         typedef typename TypeVec<float, VecTraits<T>::cn>::vec_type sum_t;\r
 \r
@@ -133,7 +133,7 @@ namespace cv { namespace gpu { namespace filters
     }\r
 \r
     template <typename T, typename D>\r
-    void linearColumnFilter_gpu(const DevMem2D& src, const DevMem2D& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream)\r
+    void linearColumnFilter_gpu(const DevMem2Db& src, const DevMem2Db& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream)\r
     {\r
         typedef void (*caller_t)(const DevMem2D_<T>& src, const DevMem2D_<D>& dst, int anchor, cudaStream_t stream);\r
         static const caller_t callers[5][17] = \r
@@ -240,11 +240,11 @@ namespace cv { namespace gpu { namespace filters
         callers[brd_type][ksize]((DevMem2D_<T>)src, (DevMem2D_<D>)dst, anchor, stream);\r
     }\r
 \r
-    template void linearColumnFilter_gpu<float , uchar >(const DevMem2D& src, const DevMem2D& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
-    template void linearColumnFilter_gpu<float4, uchar4>(const DevMem2D& src, const DevMem2D& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
-    //template void linearColumnFilter_gpu<float , short >(const DevMem2D& src, const DevMem2D& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
-    //template void linearColumnFilter_gpu<float2, short2>(const DevMem2D& src, const DevMem2D& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
-    template void linearColumnFilter_gpu<float3, short3>(const DevMem2D& src, const DevMem2D& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
-    template void linearColumnFilter_gpu<float , int   >(const DevMem2D& src, const DevMem2D& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
-    template void linearColumnFilter_gpu<float , float >(const DevMem2D& src, const DevMem2D& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
+    template void linearColumnFilter_gpu<float , uchar >(const DevMem2Db& src, const DevMem2Db& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
+    template void linearColumnFilter_gpu<float4, uchar4>(const DevMem2Db& src, const DevMem2Db& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
+    //template void linearColumnFilter_gpu<float , short >(const DevMem2Db& src, const DevMem2Db& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
+    //template void linearColumnFilter_gpu<float2, short2>(const DevMem2Db& src, const DevMem2Db& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
+    template void linearColumnFilter_gpu<float3, short3>(const DevMem2Db& src, const DevMem2Db& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
+    template void linearColumnFilter_gpu<float , int   >(const DevMem2Db& src, const DevMem2Db& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
+    template void linearColumnFilter_gpu<float , float >(const DevMem2Db& src, const DevMem2Db& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
 }}}\r

diff --git a/modules/gpu/src/cuda/copy_make_border.cu b/modules/gpu/src/cuda/copy_make_border.cu
index 8d70393..89af4e0 100644 (file)
--- a/modules/gpu/src/cuda/copy_make_border.cu
+++ b/modules/gpu/src/cuda/copy_make_border.cu
@@ -66,7 +66,7 @@ namespace cv { namespace gpu { namespace imgproc
             dim3 grid(divUp(dst.cols, block.x), divUp(dst.rows, block.y));\r
 \r
             B<T> brd(src.rows, src.cols, VecTraits<T>::make(borderValue));\r
-            BorderReader< PtrStep_<T>, B<T> > brdSrc(src, brd);\r
+            BorderReader< PtrStep<T>, B<T> > brdSrc(src, brd);\r
 \r
             copyMakeBorder<<<grid, block, 0, stream>>>(brdSrc, dst, top, left);\r
             cudaSafeCall( cudaGetLastError() );\r
@@ -76,7 +76,7 @@ namespace cv { namespace gpu { namespace imgproc
         }\r
     };\r
 \r
-    template <typename T, int cn> void copyMakeBorder_gpu(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, \r
+    template <typename T, int cn> void copyMakeBorder_gpu(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, \r
         const T* borderValue, cudaStream_t stream)\r
     {\r
         typedef typename TypeVec<T, cn>::vec_type vec_type;\r
@@ -95,33 +95,33 @@ namespace cv { namespace gpu { namespace imgproc
         callers[borderMode](DevMem2D_<vec_type>(src), DevMem2D_<vec_type>(dst), top, left, borderValue, stream);\r
     }\r
 \r
-    template void copyMakeBorder_gpu<uchar, 1>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const uchar* borderValue, cudaStream_t stream);\r
-    //template void copyMakeBorder_gpu<uchar, 2>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const uchar* borderValue, cudaStream_t stream);\r
-    template void copyMakeBorder_gpu<uchar, 3>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const uchar* borderValue, cudaStream_t stream);\r
-    template void copyMakeBorder_gpu<uchar, 4>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const uchar* borderValue, cudaStream_t stream);\r
+    template void copyMakeBorder_gpu<uchar, 1>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const uchar* borderValue, cudaStream_t stream);\r
+    //template void copyMakeBorder_gpu<uchar, 2>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const uchar* borderValue, cudaStream_t stream);\r
+    template void copyMakeBorder_gpu<uchar, 3>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const uchar* borderValue, cudaStream_t stream);\r
+    template void copyMakeBorder_gpu<uchar, 4>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const uchar* borderValue, cudaStream_t stream);\r
     \r
-    //template void copyMakeBorder_gpu<schar, 1>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const schar* borderValue, cudaStream_t stream);\r
-    //template void copyMakeBorder_gpu<schar, 2>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const schar* borderValue, cudaStream_t stream);\r
-    //template void copyMakeBorder_gpu<schar, 3>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const schar* borderValue, cudaStream_t stream);\r
-    //template void copyMakeBorder_gpu<schar, 4>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const schar* borderValue, cudaStream_t stream);\r
+    //template void copyMakeBorder_gpu<schar, 1>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const schar* borderValue, cudaStream_t stream);\r
+    //template void copyMakeBorder_gpu<schar, 2>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const schar* borderValue, cudaStream_t stream);\r
+    //template void copyMakeBorder_gpu<schar, 3>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const schar* borderValue, cudaStream_t stream);\r
+    //template void copyMakeBorder_gpu<schar, 4>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const schar* borderValue, cudaStream_t stream);\r
     \r
-    template void copyMakeBorder_gpu<ushort, 1>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const ushort* borderValue, cudaStream_t stream);\r
-    //template void copyMakeBorder_gpu<ushort, 2>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const ushort* borderValue, cudaStream_t stream);\r
-    template void copyMakeBorder_gpu<ushort, 3>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const ushort* borderValue, cudaStream_t stream);\r
-    template void copyMakeBorder_gpu<ushort, 4>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const ushort* borderValue, cudaStream_t stream);\r
+    template void copyMakeBorder_gpu<ushort, 1>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const ushort* borderValue, cudaStream_t stream);\r
+    //template void copyMakeBorder_gpu<ushort, 2>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const ushort* borderValue, cudaStream_t stream);\r
+    template void copyMakeBorder_gpu<ushort, 3>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const ushort* borderValue, cudaStream_t stream);\r
+    template void copyMakeBorder_gpu<ushort, 4>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const ushort* borderValue, cudaStream_t stream);\r
     \r
-    template void copyMakeBorder_gpu<short, 1>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const short* borderValue, cudaStream_t stream);\r
-    //template void copyMakeBorder_gpu<short, 2>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const short* borderValue, cudaStream_t stream);\r
-    template void copyMakeBorder_gpu<short, 3>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const short* borderValue, cudaStream_t stream);\r
-    template void copyMakeBorder_gpu<short, 4>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const short* borderValue, cudaStream_t stream);\r
+    template void copyMakeBorder_gpu<short, 1>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const short* borderValue, cudaStream_t stream);\r
+    //template void copyMakeBorder_gpu<short, 2>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const short* borderValue, cudaStream_t stream);\r
+    template void copyMakeBorder_gpu<short, 3>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const short* borderValue, cudaStream_t stream);\r
+    template void copyMakeBorder_gpu<short, 4>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const short* borderValue, cudaStream_t stream);\r
     \r
-    //template void copyMakeBorder_gpu<int, 1>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const int* borderValue, cudaStream_t stream);\r
-    //template void copyMakeBorder_gpu<int, 2>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const int* borderValue, cudaStream_t stream);\r
-    //template void copyMakeBorder_gpu<int, 3>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const int* borderValue, cudaStream_t stream);\r
-    //template void copyMakeBorder_gpu<int, 4>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const int* borderValue, cudaStream_t stream);\r
+    //template void copyMakeBorder_gpu<int, 1>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const int* borderValue, cudaStream_t stream);\r
+    //template void copyMakeBorder_gpu<int, 2>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const int* borderValue, cudaStream_t stream);\r
+    //template void copyMakeBorder_gpu<int, 3>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const int* borderValue, cudaStream_t stream);\r
+    //template void copyMakeBorder_gpu<int, 4>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const int* borderValue, cudaStream_t stream);\r
     \r
-    template void copyMakeBorder_gpu<float, 1>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    //template void copyMakeBorder_gpu<float, 2>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    template void copyMakeBorder_gpu<float, 3>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    template void copyMakeBorder_gpu<float, 4>(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    template void copyMakeBorder_gpu<float, 1>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    //template void copyMakeBorder_gpu<float, 2>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    template void copyMakeBorder_gpu<float, 3>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    template void copyMakeBorder_gpu<float, 4>(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const float* borderValue, cudaStream_t stream);\r
 }}}\r

diff --git a/modules/gpu/src/cuda/element_operations.cu b/modules/gpu/src/cuda/element_operations.cu
index ce9281c..cd72468 100644 (file)
--- a/modules/gpu/src/cuda/element_operations.cu
+++ b/modules/gpu/src/cuda/element_operations.cu
@@ -81,7 +81,7 @@ namespace cv { namespace gpu { namespace device
         enum { smart_shift = 4 };\r
     };\r
 \r
-    template <typename T, typename D> void add_gpu(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream)\r
+    template <typename T, typename D> void add_gpu(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream)\r
     {\r
         if (mask.data)\r
             transform((DevMem2D_<T>)src1, (DevMem2D_<T>)src2, (DevMem2D_<D>)dst, mask, Add<T, D>(), stream);\r
@@ -89,61 +89,61 @@ namespace cv { namespace gpu { namespace device
             transform((DevMem2D_<T>)src1, (DevMem2D_<T>)src2, (DevMem2D_<D>)dst, Add<T, D>(), stream);\r
     }\r
 \r
-    template void add_gpu<uchar, uchar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<uchar, schar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<uchar, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<uchar, short>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<uchar, int>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<uchar, float>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<uchar, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void add_gpu<schar, uchar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<schar, schar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<schar, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<schar, short>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<schar, int>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<schar, float>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<schar, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void add_gpu<ushort, uchar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<ushort, schar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<ushort, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<ushort, short>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<ushort, int>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<ushort, float>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<ushort, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void add_gpu<short, uchar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<short, schar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<short, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<short, short>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<short, int>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<short, float>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<short, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void add_gpu<int, uchar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<int, schar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<int, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<int, short>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<int, int>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<int, float>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<int, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void add_gpu<float, uchar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<float, schar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<float, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<float, short>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<float, int>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<float, float>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<float, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void add_gpu<double, uchar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<double, schar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<double, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<double, short>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<double, int>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<double, float>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<double, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
+    template void add_gpu<uchar, uchar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<uchar, schar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<uchar, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<uchar, short>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<uchar, int>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<uchar, float>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<uchar, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void add_gpu<schar, uchar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<schar, schar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<schar, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<schar, short>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<schar, int>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<schar, float>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<schar, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void add_gpu<ushort, uchar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<ushort, schar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<ushort, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<ushort, short>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<ushort, int>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<ushort, float>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<ushort, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void add_gpu<short, uchar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<short, schar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<short, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<short, short>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<short, int>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<short, float>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<short, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void add_gpu<int, uchar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<int, schar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<int, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<int, short>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<int, int>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<int, float>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<int, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void add_gpu<float, uchar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<float, schar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<float, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<float, short>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<float, int>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<float, float>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<float, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void add_gpu<double, uchar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<double, schar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<double, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<double, short>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<double, int>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<double, float>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<double, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
 \r
     template <typename T, typename D> struct AddScalar : unary_function<T, D>\r
     {\r
@@ -176,7 +176,7 @@ namespace cv { namespace gpu { namespace device
         enum { smart_shift = 4 };\r
     };\r
 \r
-    template <typename T, typename D> void add_gpu(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream)\r
+    template <typename T, typename D> void add_gpu(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream)\r
     {\r
         cudaSafeCall( cudaSetDoubleForDevice(&val) );\r
         AddScalar<T, D> op(val);\r
@@ -186,61 +186,61 @@ namespace cv { namespace gpu { namespace device
             transform((DevMem2D_<T>)src1, (DevMem2D_<D>)dst, op, stream);\r
     }\r
 \r
-    template void add_gpu<uchar, uchar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<uchar, schar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<uchar, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<uchar, short >(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<uchar, int   >(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<uchar, float >(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<uchar, double>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void add_gpu<schar, uchar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<schar, schar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<schar, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<schar, short>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<schar, int>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<schar, float>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<schar, double>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void add_gpu<ushort, uchar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<ushort, schar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<ushort, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<ushort, short>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<ushort, int>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<ushort, float>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<ushort, double>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void add_gpu<short, uchar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<short, schar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<short, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<short, short>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<short, int>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<short, float>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<short, double>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void add_gpu<int, uchar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<int, schar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<int, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<int, short>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<int, int>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<int, float>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<int, double>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void add_gpu<float, uchar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<float, schar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<float, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<float, short>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<float, int>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<float, float>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<float, double>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void add_gpu<double, uchar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<double, schar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<double, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<double, short>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<double, int>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void add_gpu<double, float>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void add_gpu<double, double>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
+    template void add_gpu<uchar, uchar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<uchar, schar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<uchar, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<uchar, short >(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<uchar, int   >(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<uchar, float >(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<uchar, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void add_gpu<schar, uchar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<schar, schar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<schar, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<schar, short>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<schar, int>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<schar, float>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<schar, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void add_gpu<ushort, uchar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<ushort, schar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<ushort, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<ushort, short>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<ushort, int>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<ushort, float>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<ushort, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void add_gpu<short, uchar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<short, schar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<short, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<short, short>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<short, int>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<short, float>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<short, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void add_gpu<int, uchar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<int, schar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<int, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<int, short>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<int, int>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<int, float>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<int, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void add_gpu<float, uchar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<float, schar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<float, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<float, short>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<float, int>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<float, float>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<float, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void add_gpu<double, uchar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<double, schar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<double, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<double, short>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<double, int>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void add_gpu<double, float>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void add_gpu<double, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
 \r
     //////////////////////////////////////////////////////////////////////////\r
     // subtract\r
@@ -274,7 +274,7 @@ namespace cv { namespace gpu { namespace device
         enum { smart_shift = 4 };\r
     };\r
 \r
-    template <typename T, typename D> void subtract_gpu(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream)\r
+    template <typename T, typename D> void subtract_gpu(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream)\r
     {\r
         if (mask.data)\r
             transform((DevMem2D_<T>)src1, (DevMem2D_<T>)src2, (DevMem2D_<D>)dst, mask, Subtract<T, D>(), stream);\r
@@ -282,61 +282,61 @@ namespace cv { namespace gpu { namespace device
             transform((DevMem2D_<T>)src1, (DevMem2D_<T>)src2, (DevMem2D_<D>)dst, Subtract<T, D>(), stream);\r
     }\r
 \r
-    template void subtract_gpu<uchar, uchar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<uchar, schar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<uchar, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<uchar, short>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<uchar, int>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<uchar, float>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<uchar, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void subtract_gpu<schar, uchar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<schar, schar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<schar, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<schar, short>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<schar, int>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<schar, float>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<schar, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void subtract_gpu<ushort, uchar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<ushort, schar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<ushort, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<ushort, short>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<ushort, int>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<ushort, float>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<ushort, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void subtract_gpu<short, uchar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<short, schar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<short, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<short, short>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<short, int>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<short, float>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<short, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void subtract_gpu<int, uchar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<int, schar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<int, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<int, short>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<int, int>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<int, float>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<int, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void subtract_gpu<float, uchar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<float, schar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<float, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<float, short>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<float, int>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<float, float>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<float, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void subtract_gpu<double, uchar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<double, schar>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<double, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<double, short>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<double, int>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<double, float>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<double, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
+    template void subtract_gpu<uchar, uchar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<uchar, schar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<uchar, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<uchar, short>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<uchar, int>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<uchar, float>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<uchar, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void subtract_gpu<schar, uchar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<schar, schar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<schar, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<schar, short>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<schar, int>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<schar, float>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<schar, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void subtract_gpu<ushort, uchar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<ushort, schar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<ushort, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<ushort, short>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<ushort, int>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<ushort, float>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<ushort, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void subtract_gpu<short, uchar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<short, schar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<short, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<short, short>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<short, int>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<short, float>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<short, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void subtract_gpu<int, uchar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<int, schar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<int, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<int, short>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<int, int>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<int, float>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<int, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void subtract_gpu<float, uchar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<float, schar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<float, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<float, short>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<float, int>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<float, float>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<float, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void subtract_gpu<double, uchar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<double, schar>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<double, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<double, short>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<double, int>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<double, float>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<double, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
 \r
     template <typename T, typename D> struct SubtractScalar : unary_function<T, D>\r
     {\r
@@ -369,7 +369,7 @@ namespace cv { namespace gpu { namespace device
         enum { smart_shift = 4 };\r
     };\r
 \r
-    template <typename T, typename D> void subtract_gpu(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream)\r
+    template <typename T, typename D> void subtract_gpu(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream)\r
     {\r
         cudaSafeCall( cudaSetDoubleForDevice(&val) );\r
         SubtractScalar<T, D> op(val);\r
@@ -379,61 +379,61 @@ namespace cv { namespace gpu { namespace device
             transform((DevMem2D_<T>)src1, (DevMem2D_<D>)dst, op, stream);\r
     }\r
 \r
-    template void subtract_gpu<uchar, uchar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<uchar, schar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<uchar, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<uchar, short >(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<uchar, int   >(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<uchar, float >(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<uchar, double>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void subtract_gpu<schar, uchar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<schar, schar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<schar, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<schar, short>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<schar, int>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<schar, float>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<schar, double>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void subtract_gpu<ushort, uchar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<ushort, schar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<ushort, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<ushort, short>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<ushort, int>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<ushort, float>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<ushort, double>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void subtract_gpu<short, uchar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<short, schar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<short, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<short, short>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<short, int>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<short, float>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<short, double>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void subtract_gpu<int, uchar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<int, schar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<int, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<int, short>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<int, int>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<int, float>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<int, double>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void subtract_gpu<float, uchar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<float, schar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<float, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<float, short>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<float, int>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<float, float>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<float, double>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-\r
-    //template void subtract_gpu<double, uchar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<double, schar>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<double, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<double, short>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<double, int>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    //template void subtract_gpu<double, float>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
-    template void subtract_gpu<double, double>(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
+    template void subtract_gpu<uchar, uchar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<uchar, schar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<uchar, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<uchar, short >(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<uchar, int   >(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<uchar, float >(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<uchar, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void subtract_gpu<schar, uchar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<schar, schar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<schar, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<schar, short>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<schar, int>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<schar, float>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<schar, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void subtract_gpu<ushort, uchar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<ushort, schar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<ushort, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<ushort, short>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<ushort, int>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<ushort, float>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<ushort, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void subtract_gpu<short, uchar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<short, schar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<short, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<short, short>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<short, int>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<short, float>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<short, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void subtract_gpu<int, uchar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<int, schar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<int, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<int, short>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<int, int>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<int, float>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<int, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void subtract_gpu<float, uchar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<float, schar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<float, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<float, short>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<float, int>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<float, float>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<float, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+\r
+    //template void subtract_gpu<double, uchar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<double, schar>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<double, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<double, short>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<double, int>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    //template void subtract_gpu<double, float>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
+    template void subtract_gpu<double, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
 \r
     //////////////////////////////////////////////////////////////////////////\r
     // multiply\r
@@ -518,68 +518,68 @@ namespace cv { namespace gpu { namespace device
         enum { smart_shift = 4 };\r
     };\r
 \r
-    template <typename T, typename D> void multiply_gpu(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream)\r
+    template <typename T, typename D> void multiply_gpu(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream)\r
     {\r
         cudaSafeCall( cudaSetDoubleForDevice(&scale) );\r
         Multiply<T, D> op(scale);\r
         transform((DevMem2D_<T>)src1, (DevMem2D_<T>)src2, (DevMem2D_<D>)dst, op, stream);\r
     }\r
 \r
-    template void multiply_gpu<uchar, uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<uchar, schar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<uchar, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<uchar, short >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<uchar, int   >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<uchar, float >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<uchar, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void multiply_gpu<schar, uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<schar, schar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<schar, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<schar, short >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<schar, int   >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<schar, float >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<schar, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void multiply_gpu<ushort, uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<ushort, schar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<ushort, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<ushort, short >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<ushort, int   >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<ushort, float >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<ushort, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void multiply_gpu<short, uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<short, schar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<short, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<short, short >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<short, int   >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<short, float >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<short, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void multiply_gpu<int, uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<int, schar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<int, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<int, short >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<int, int   >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<int, float >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<int, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void multiply_gpu<float, uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<float, schar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<float, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<float, short >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<float, int   >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<float, float >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<float, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void multiply_gpu<double, uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<double, schar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<double, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<double, short >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<double, int   >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<double, float >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<double, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<uchar, uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<uchar, schar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<uchar, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<uchar, short >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<uchar, int   >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<uchar, float >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<uchar, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void multiply_gpu<schar, uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<schar, schar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<schar, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<schar, short >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<schar, int   >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<schar, float >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<schar, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void multiply_gpu<ushort, uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<ushort, schar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<ushort, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<ushort, short >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<ushort, int   >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<ushort, float >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<ushort, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void multiply_gpu<short, uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<short, schar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<short, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<short, short >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<short, int   >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<short, float >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<short, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void multiply_gpu<int, uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<int, schar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<int, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<int, short >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<int, int   >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<int, float >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<int, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void multiply_gpu<float, uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<float, schar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<float, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<float, short >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<float, int   >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<float, float >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<float, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void multiply_gpu<double, uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<double, schar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<double, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<double, short >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<double, int   >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<double, float >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<double, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
     \r
     template <typename T, typename D> struct MultiplyScalar : unary_function<T, D>\r
     {\r
@@ -613,7 +613,7 @@ namespace cv { namespace gpu { namespace device
         enum { smart_shift = 4 };\r
     };\r
 \r
-    template <typename T, typename D> void multiply_gpu(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream)\r
+    template <typename T, typename D> void multiply_gpu(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream)\r
     {\r
         cudaSafeCall( cudaSetDoubleForDevice(&val) );\r
         cudaSafeCall( cudaSetDoubleForDevice(&scale) );\r
@@ -621,61 +621,61 @@ namespace cv { namespace gpu { namespace device
         transform((DevMem2D_<T>)src1, (DevMem2D_<D>)dst, op, stream);\r
     }\r
 \r
-    template void multiply_gpu<uchar, uchar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<uchar, schar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<uchar, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<uchar, short >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<uchar, int   >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<uchar, float >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<uchar, double>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void multiply_gpu<schar, uchar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<schar, schar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<schar, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<schar, short >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<schar, int   >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<schar, float >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<schar, double>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void multiply_gpu<ushort, uchar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<ushort, schar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<ushort, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<ushort, short >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<ushort, int   >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<ushort, float >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<ushort, double>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void multiply_gpu<short, uchar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<short, schar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<short, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<short, short >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<short, int   >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<short, float >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<short, double>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void multiply_gpu<int, uchar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<int, schar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<int, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<int, short >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<int, int   >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<int, float >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<int, double>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void multiply_gpu<float, uchar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<float, schar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<float, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<float, short >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<float, int   >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<float, float >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<float, double>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void multiply_gpu<double, uchar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<double, schar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<double, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<double, short >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<double, int   >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void multiply_gpu<double, float >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void multiply_gpu<double, double>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<uchar, uchar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<uchar, schar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<uchar, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<uchar, short >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<uchar, int   >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<uchar, float >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<uchar, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void multiply_gpu<schar, uchar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<schar, schar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<schar, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<schar, short >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<schar, int   >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<schar, float >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<schar, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void multiply_gpu<ushort, uchar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<ushort, schar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<ushort, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<ushort, short >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<ushort, int   >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<ushort, float >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<ushort, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void multiply_gpu<short, uchar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<short, schar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<short, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<short, short >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<short, int   >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<short, float >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<short, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void multiply_gpu<int, uchar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<int, schar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<int, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<int, short >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<int, int   >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<int, float >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<int, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void multiply_gpu<float, uchar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<float, schar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<float, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<float, short >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<float, int   >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<float, float >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<float, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void multiply_gpu<double, uchar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<double, schar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<double, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<double, short >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<double, int   >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void multiply_gpu<double, float >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void multiply_gpu<double, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
 \r
     //////////////////////////////////////////////////////////////////////////\r
     // divide\r
@@ -756,68 +756,68 @@ namespace cv { namespace gpu { namespace device
         enum { smart_shift = 4 };\r
     };\r
 \r
-    template <typename T, typename D> void divide_gpu(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream)\r
+    template <typename T, typename D> void divide_gpu(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream)\r
     {\r
         cudaSafeCall( cudaSetDoubleForDevice(&scale) );\r
         Divide<T, D> op(scale);\r
         transform((DevMem2D_<T>)src1, (DevMem2D_<T>)src2, (DevMem2D_<D>)dst, op, stream);\r
     }\r
 \r
-    template void divide_gpu<uchar, uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<uchar, schar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<uchar, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<uchar, short >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<uchar, int   >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<uchar, float >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<uchar, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void divide_gpu<schar, uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<schar, schar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<schar, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<schar, short >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<schar, int   >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<schar, float >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<schar, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void divide_gpu<ushort, uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<ushort, schar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<ushort, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<ushort, short >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<ushort, int   >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<ushort, float >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<ushort, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void divide_gpu<short, uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<short, schar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<short, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<short, short >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<short, int   >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<short, float >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<short, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void divide_gpu<int, uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<int, schar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<int, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<int, short >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<int, int   >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<int, float >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<int, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void divide_gpu<float, uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<float, schar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<float, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<float, short >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<float, int   >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<float, float >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<float, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void divide_gpu<double, uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<double, schar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<double, ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<double, short >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<double, int   >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<double, float >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<double, double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<uchar, uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<uchar, schar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<uchar, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<uchar, short >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<uchar, int   >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<uchar, float >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<uchar, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void divide_gpu<schar, uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<schar, schar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<schar, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<schar, short >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<schar, int   >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<schar, float >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<schar, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void divide_gpu<ushort, uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<ushort, schar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<ushort, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<ushort, short >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<ushort, int   >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<ushort, float >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<ushort, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void divide_gpu<short, uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<short, schar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<short, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<short, short >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<short, int   >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<short, float >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<short, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void divide_gpu<int, uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<int, schar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<int, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<int, short >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<int, int   >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<int, float >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<int, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void divide_gpu<float, uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<float, schar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<float, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<float, short >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<float, int   >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<float, float >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<float, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void divide_gpu<double, uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<double, schar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<double, ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<double, short >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<double, int   >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<double, float >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<double, double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
     \r
     template <typename T, typename D> struct DivideScalar : unary_function<T, D>\r
     {\r
@@ -851,7 +851,7 @@ namespace cv { namespace gpu { namespace device
         enum { smart_shift = 4 };\r
     };\r
 \r
-    template <typename T, typename D> void divide_gpu(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream)\r
+    template <typename T, typename D> void divide_gpu(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream)\r
     {\r
         cudaSafeCall( cudaSetDoubleForDevice(&val) );\r
         cudaSafeCall( cudaSetDoubleForDevice(&scale) );\r
@@ -859,61 +859,61 @@ namespace cv { namespace gpu { namespace device
         transform((DevMem2D_<T>)src1, (DevMem2D_<D>)dst, op, stream);\r
     }\r
 \r
-    template void divide_gpu<uchar, uchar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<uchar, schar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<uchar, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<uchar, short >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<uchar, int   >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<uchar, float >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<uchar, double>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void divide_gpu<schar, uchar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<schar, schar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<schar, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<schar, short >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<schar, int   >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<schar, float >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<schar, double>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void divide_gpu<ushort, uchar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<ushort, schar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<ushort, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<ushort, short >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<ushort, int   >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<ushort, float >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<ushort, double>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void divide_gpu<short, uchar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<short, schar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<short, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<short, short >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<short, int   >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<short, float >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<short, double>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void divide_gpu<int, uchar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<int, schar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<int, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<int, short >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<int, int   >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<int, float >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<int, double>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void divide_gpu<float, uchar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<float, schar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<float, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<float, short >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<float, int   >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<float, float >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<float, double>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-\r
-    //template void divide_gpu<double, uchar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<double, schar >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<double, ushort>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<double, short >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<double, int   >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    //template void divide_gpu<double, float >(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
-    template void divide_gpu<double, double>(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<uchar, uchar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<uchar, schar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<uchar, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<uchar, short >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<uchar, int   >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<uchar, float >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<uchar, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void divide_gpu<schar, uchar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<schar, schar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<schar, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<schar, short >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<schar, int   >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<schar, float >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<schar, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void divide_gpu<ushort, uchar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<ushort, schar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<ushort, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<ushort, short >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<ushort, int   >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<ushort, float >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<ushort, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void divide_gpu<short, uchar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<short, schar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<short, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<short, short >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<short, int   >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<short, float >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<short, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void divide_gpu<int, uchar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<int, schar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<int, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<int, short >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<int, int   >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<int, float >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<int, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void divide_gpu<float, uchar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<float, schar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<float, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<float, short >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<float, int   >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<float, float >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<float, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+\r
+    //template void divide_gpu<double, uchar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<double, schar >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<double, ushort>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<double, short >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<double, int   >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    //template void divide_gpu<double, float >(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
+    template void divide_gpu<double, double>(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
 \r
     template <typename T, typename D> struct Reciprocal : unary_function<T, D>\r
     {\r
@@ -946,68 +946,68 @@ namespace cv { namespace gpu { namespace device
         enum { smart_shift = 4 };\r
     };\r
 \r
-    template <typename T, typename D> void divide_gpu(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream)\r
+    template <typename T, typename D> void divide_gpu(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream)\r
     {\r
         cudaSafeCall( cudaSetDoubleForDevice(&scalar) );\r
         Reciprocal<T, D> op(scalar);\r
         transform((DevMem2D_<T>)src2, (DevMem2D_<D>)dst, op, stream);\r
     }\r
 \r
-    template void divide_gpu<uchar, uchar >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<uchar, schar >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void divide_gpu<uchar, ushort>(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void divide_gpu<uchar, short >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void divide_gpu<uchar, int   >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void divide_gpu<uchar, float >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void divide_gpu<uchar, double>(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    //template void divide_gpu<schar, uchar >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<schar, schar >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<schar, ushort>(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<schar, short >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<schar, int   >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<schar, float >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<schar, double>(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    //template void divide_gpu<ushort, uchar >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<ushort, schar >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void divide_gpu<ushort, ushort>(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<ushort, short >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void divide_gpu<ushort, int   >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void divide_gpu<ushort, float >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void divide_gpu<ushort, double>(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    //template void divide_gpu<short, uchar >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<short, schar >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<short, ushort>(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void divide_gpu<short, short >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void divide_gpu<short, int   >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void divide_gpu<short, float >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void divide_gpu<short, double>(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    //template void divide_gpu<int, uchar >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<int, schar >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<int, ushort>(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<int, short >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void divide_gpu<int, int   >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void divide_gpu<int, float >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void divide_gpu<int, double>(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    //template void divide_gpu<float, uchar >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<float, schar >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<float, ushort>(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<float, short >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<float, int   >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void divide_gpu<float, float >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void divide_gpu<float, double>(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    //template void divide_gpu<double, uchar >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<double, schar >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<double, ushort>(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<double, short >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<double, int   >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void divide_gpu<double, float >(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void divide_gpu<double, double>(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
+    template void divide_gpu<uchar, uchar >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<uchar, schar >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void divide_gpu<uchar, ushort>(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void divide_gpu<uchar, short >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void divide_gpu<uchar, int   >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void divide_gpu<uchar, float >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void divide_gpu<uchar, double>(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    //template void divide_gpu<schar, uchar >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<schar, schar >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<schar, ushort>(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<schar, short >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<schar, int   >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<schar, float >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<schar, double>(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    //template void divide_gpu<ushort, uchar >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<ushort, schar >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void divide_gpu<ushort, ushort>(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<ushort, short >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void divide_gpu<ushort, int   >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void divide_gpu<ushort, float >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void divide_gpu<ushort, double>(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    //template void divide_gpu<short, uchar >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<short, schar >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<short, ushort>(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void divide_gpu<short, short >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void divide_gpu<short, int   >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void divide_gpu<short, float >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void divide_gpu<short, double>(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    //template void divide_gpu<int, uchar >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<int, schar >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<int, ushort>(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<int, short >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void divide_gpu<int, int   >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void divide_gpu<int, float >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void divide_gpu<int, double>(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    //template void divide_gpu<float, uchar >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<float, schar >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<float, ushort>(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<float, short >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<float, int   >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void divide_gpu<float, float >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void divide_gpu<float, double>(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    //template void divide_gpu<double, uchar >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<double, schar >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<double, ushort>(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<double, short >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<double, int   >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void divide_gpu<double, float >(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void divide_gpu<double, double>(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
 \r
     //////////////////////////////////////////////////////////////////////////\r
     // absdiff\r
@@ -1054,18 +1054,18 @@ namespace cv { namespace gpu { namespace device
         enum { smart_shift = 4 };\r
     };\r
 \r
-    template <typename T> void absdiff_gpu(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream)\r
+    template <typename T> void absdiff_gpu(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream)\r
     {\r
         transform((DevMem2D_<T>)src1, (DevMem2D_<T>)src2, (DevMem2D_<T>)dst, Absdiff<T>(), stream);\r
     }\r
 \r
-    //template void absdiff_gpu<uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void absdiff_gpu<schar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void absdiff_gpu<ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void absdiff_gpu<short >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void absdiff_gpu<int   >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    //template void absdiff_gpu<float >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void absdiff_gpu<double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
+    //template void absdiff_gpu<uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void absdiff_gpu<schar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void absdiff_gpu<ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void absdiff_gpu<short >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void absdiff_gpu<int   >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    //template void absdiff_gpu<float >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void absdiff_gpu<double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
 \r
     template <typename T> struct AbsdiffScalar : unary_function<T, T>\r
     {\r
@@ -1098,20 +1098,20 @@ namespace cv { namespace gpu { namespace device
         enum { smart_shift = 4 };\r
     };\r
 \r
-    template <typename T> void absdiff_gpu(const DevMem2D& src1, double val, const DevMem2D& dst, cudaStream_t stream)\r
+    template <typename T> void absdiff_gpu(const DevMem2Db& src1, double val, const DevMem2Db& dst, cudaStream_t stream)\r
     {\r
         cudaSafeCall( cudaSetDoubleForDevice(&val) );\r
         AbsdiffScalar<T> op(val);\r
         transform((DevMem2D_<T>)src1, (DevMem2D_<T>)dst, op, stream);\r
     }\r
 \r
-    template void absdiff_gpu<uchar >(const DevMem2D& src1, double src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void absdiff_gpu<schar >(const DevMem2D& src1, double src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void absdiff_gpu<ushort>(const DevMem2D& src1, double src2, const DevMem2D& dst, cudaStream_t stream);                  \r
-    template void absdiff_gpu<short >(const DevMem2D& src1, double src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void absdiff_gpu<int   >(const DevMem2D& src1, double src2, const DevMem2D& dst, cudaStream_t stream);                  \r
-    //template void absdiff_gpu<float >(const DevMem2D& src1, double src2, const DevMem2D& dst, cudaStream_t stream);                  \r
-    template void absdiff_gpu<double>(const DevMem2D& src1, double src2, const DevMem2D& dst, cudaStream_t stream);\r
+    template void absdiff_gpu<uchar >(const DevMem2Db& src1, double src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void absdiff_gpu<schar >(const DevMem2Db& src1, double src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void absdiff_gpu<ushort>(const DevMem2Db& src1, double src2, const DevMem2Db& dst, cudaStream_t stream);                  \r
+    template void absdiff_gpu<short >(const DevMem2Db& src1, double src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void absdiff_gpu<int   >(const DevMem2Db& src1, double src2, const DevMem2Db& dst, cudaStream_t stream);                  \r
+    //template void absdiff_gpu<float >(const DevMem2Db& src1, double src2, const DevMem2Db& dst, cudaStream_t stream);                  \r
+    template void absdiff_gpu<double>(const DevMem2Db& src1, double src2, const DevMem2Db& dst, cudaStream_t stream);\r
 \r
     //////////////////////////////////////////////////////////////////////////////////////\r
     // Compare\r
@@ -1186,60 +1186,60 @@ namespace cv { namespace gpu { namespace device
         enum { smart_shift = 4 };\r
     };\r
 \r
-    template <template <typename> class Op, typename T> void compare(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream)\r
+    template <template <typename> class Op, typename T> void compare(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream)\r
     {\r
         Op<T> op;\r
         transform(static_cast< DevMem2D_<T> >(src1), static_cast< DevMem2D_<T> >(src2), dst, op, stream);\r
     }\r
 \r
-    template <typename T> void compare_eq(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream)\r
+    template <typename T> void compare_eq(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream)\r
     {\r
         compare<Equal, T>(src1, src2, dst, stream);\r
     }\r
-    template <typename T> void compare_ne(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream)\r
+    template <typename T> void compare_ne(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream)\r
     {\r
         compare<NotEqual, T>(src1, src2, dst, stream);\r
     }\r
-    template <typename T> void compare_lt(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream)\r
+    template <typename T> void compare_lt(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream)\r
     {\r
         compare<Less, T>(src1, src2, dst, stream);\r
     }\r
-    template <typename T> void compare_le(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream)\r
+    template <typename T> void compare_le(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream)\r
     {\r
         compare<LessEqual, T>(src1, src2, dst, stream);\r
     }\r
     \r
-    template void compare_eq<uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_eq<schar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_eq<ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_eq<short >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_eq<int   >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_eq<float >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_eq<double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
+    template void compare_eq<uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_eq<schar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_eq<ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_eq<short >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_eq<int   >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_eq<float >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_eq<double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
     \r
-    template void compare_ne<uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_ne<schar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_ne<ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_ne<short >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_ne<int   >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_ne<float >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_ne<double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
+    template void compare_ne<uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_ne<schar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_ne<ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_ne<short >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_ne<int   >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_ne<float >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_ne<double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
     \r
-    template void compare_lt<uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_lt<schar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_lt<ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_lt<short >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_lt<int   >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_lt<float >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_lt<double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
+    template void compare_lt<uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_lt<schar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_lt<ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_lt<short >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_lt<int   >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_lt<float >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_lt<double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
     \r
-    template void compare_le<uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_le<schar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_le<ushort>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_le<short >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_le<int   >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_le<float >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template void compare_le<double>(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
+    template void compare_le<uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_le<schar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_le<ushort>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_le<short >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_le<int   >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_le<float >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void compare_le<double>(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
 \r
 \r
     //////////////////////////////////////////////////////////////////////////\r
@@ -1258,7 +1258,7 @@ namespace cv { namespace gpu { namespace device
 \r
 \r
     template <int opid>\r
-    __global__ void bitwiseUnOpKernel(int rows, int width, const PtrStep src, PtrStep dst)\r
+    __global__ void bitwiseUnOpKernel(int rows, int width, const PtrStepb src, PtrStepb dst)\r
     {\r
         const int x = (blockDim.x * blockIdx.x + threadIdx.x) * 4;\r
         const int y = blockDim.y * blockIdx.y + threadIdx.y;\r
@@ -1284,7 +1284,7 @@ namespace cv { namespace gpu { namespace device
 \r
 \r
     template <int opid>\r
-    void bitwiseUnOp(int rows, int width, const PtrStep src, PtrStep dst, \r
+    void bitwiseUnOp(int rows, int width, const PtrStepb src, PtrStepb dst, \r
                      cudaStream_t stream)\r
     {\r
         dim3 threads(16, 16);\r
@@ -1300,8 +1300,8 @@ namespace cv { namespace gpu { namespace device
 \r
 \r
     template <typename T, int opid>\r
-    __global__ void bitwiseUnOpKernel(int rows, int cols, int cn, const PtrStep src, \r
-                                      const PtrStep mask, PtrStep dst)\r
+    __global__ void bitwiseUnOpKernel(int rows, int cols, int cn, const PtrStepb src, \r
+                                      const PtrStepb mask, PtrStepb dst)\r
     {\r
         const int x = blockDim.x * blockIdx.x + threadIdx.x;\r
         const int y = blockDim.y * blockIdx.y + threadIdx.y;\r
@@ -1317,8 +1317,8 @@ namespace cv { namespace gpu { namespace device
 \r
 \r
     template <typename T, int opid>\r
-    void bitwiseUnOp(int rows, int cols, int cn, const PtrStep src, \r
-                     const PtrStep mask, PtrStep dst, cudaStream_t stream)\r
+    void bitwiseUnOp(int rows, int cols, int cn, const PtrStepb src, \r
+                     const PtrStepb mask, PtrStepb dst, cudaStream_t stream)\r
     {\r
         dim3 threads(16, 16);\r
         dim3 grid(divUp(cols, threads.x), divUp(rows, threads.y));\r
@@ -1332,22 +1332,22 @@ namespace cv { namespace gpu { namespace device
 \r
 \r
     void bitwiseNotCaller(int rows, int cols, size_t elem_size1, int cn, \r
-                          const PtrStep src, PtrStep dst, cudaStream_t stream)\r
+                          const PtrStepb src, PtrStepb dst, cudaStream_t stream)\r
     {\r
         bitwiseUnOp<UN_OP_NOT>(rows, static_cast<int>(cols * elem_size1 * cn), src, dst, stream);\r
     }\r
 \r
 \r
     template <typename T>\r
-    void bitwiseMaskNotCaller(int rows, int cols, int cn, const PtrStep src, \r
-                              const PtrStep mask, PtrStep dst, cudaStream_t stream)\r
+    void bitwiseMaskNotCaller(int rows, int cols, int cn, const PtrStepb src, \r
+                              const PtrStepb mask, PtrStepb dst, cudaStream_t stream)\r
     {\r
         bitwiseUnOp<T, UN_OP_NOT>(rows, cols * cn, cn, src, mask, dst, stream);\r
     }\r
 \r
-    template void bitwiseMaskNotCaller<uchar>(int, int, int, const PtrStep, const PtrStep, PtrStep, cudaStream_t);\r
-    template void bitwiseMaskNotCaller<ushort>(int, int, int, const PtrStep, const PtrStep, PtrStep, cudaStream_t);\r
-    template void bitwiseMaskNotCaller<uint>(int, int, int, const PtrStep, const PtrStep, PtrStep, cudaStream_t);\r
+    template void bitwiseMaskNotCaller<uchar>(int, int, int, const PtrStepb, const PtrStepb, PtrStepb, cudaStream_t);\r
+    template void bitwiseMaskNotCaller<ushort>(int, int, int, const PtrStepb, const PtrStepb, PtrStepb, cudaStream_t);\r
+    template void bitwiseMaskNotCaller<uint>(int, int, int, const PtrStepb, const PtrStepb, PtrStepb, cudaStream_t);\r
 \r
 \r
     //////////////////////////////////////////////////////////////////////////\r
@@ -1379,8 +1379,8 @@ namespace cv { namespace gpu { namespace device
 \r
 \r
     template <int opid>\r
-    __global__ void bitwiseBinOpKernel(int rows, int width, const PtrStep src1, \r
-                                       const PtrStep src2, PtrStep dst)\r
+    __global__ void bitwiseBinOpKernel(int rows, int width, const PtrStepb src1, \r
+                                       const PtrStepb src2, PtrStepb dst)\r
     {\r
         const int x = (blockDim.x * blockIdx.x + threadIdx.x) * 4;\r
         const int y = blockDim.y * blockIdx.y + threadIdx.y;\r
@@ -1408,8 +1408,8 @@ namespace cv { namespace gpu { namespace device
 \r
 \r
     template <int opid>\r
-    void bitwiseBinOp(int rows, int width, const PtrStep src1, const PtrStep src2, \r
-                      PtrStep dst, cudaStream_t stream)\r
+    void bitwiseBinOp(int rows, int width, const PtrStepb src1, const PtrStepb src2, \r
+                      PtrStepb dst, cudaStream_t stream)\r
     {\r
         dim3 threads(16, 16);\r
         dim3 grid(divUp(width, threads.x * sizeof(uint)), divUp(rows, threads.y));\r
@@ -1424,8 +1424,8 @@ namespace cv { namespace gpu { namespace device
 \r
     template <typename T, int opid>\r
     __global__ void bitwiseBinOpKernel(\r
-            int rows, int cols, int cn, const PtrStep src1, const PtrStep src2, \r
-            const PtrStep mask, PtrStep dst)\r
+            int rows, int cols, int cn, const PtrStepb src1, const PtrStepb src2, \r
+            const PtrStepb mask, PtrStepb dst)\r
     {\r
         const int x = blockDim.x * blockIdx.x + threadIdx.x;\r
         const int y = blockDim.y * blockIdx.y + threadIdx.y;\r
@@ -1442,8 +1442,8 @@ namespace cv { namespace gpu { namespace device
 \r
 \r
     template <typename T, int opid>\r
-    void bitwiseBinOp(int rows, int cols, int cn, const PtrStep src1, const PtrStep src2, \r
-                        const PtrStep mask, PtrStep dst, cudaStream_t stream)\r
+    void bitwiseBinOp(int rows, int cols, int cn, const PtrStepb src1, const PtrStepb src2, \r
+                        const PtrStepb mask, PtrStepb dst, cudaStream_t stream)\r
     {\r
         dim3 threads(16, 16);\r
         dim3 grid(divUp(cols, threads.x), divUp(rows, threads.y));\r
@@ -1456,61 +1456,61 @@ namespace cv { namespace gpu { namespace device
     }\r
 \r
 \r
-    void bitwiseOrCaller(int rows, int cols, size_t elem_size1, int cn, const PtrStep src1, \r
-                         const PtrStep src2, PtrStep dst, cudaStream_t stream)\r
+    void bitwiseOrCaller(int rows, int cols, size_t elem_size1, int cn, const PtrStepb src1, \r
+                         const PtrStepb src2, PtrStepb dst, cudaStream_t stream)\r
     {\r
         bitwiseBinOp<BIN_OP_OR>(rows, static_cast<int>(cols * elem_size1 * cn), src1, src2, dst, stream);\r
     }\r
 \r
 \r
     template <typename T>\r
-    void bitwiseMaskOrCaller(int rows, int cols, int cn, const PtrStep src1, const PtrStep src2, \r
-                             const PtrStep mask, PtrStep dst, cudaStream_t stream)\r
+    void bitwiseMaskOrCaller(int rows, int cols, int cn, const PtrStepb src1, const PtrStepb src2, \r
+                             const PtrStepb mask, PtrStepb dst, cudaStream_t stream)\r
     {\r
         bitwiseBinOp<T, BIN_OP_OR>(rows, cols * cn, cn, src1, src2, mask, dst, stream);\r
     }\r
 \r
-    template void bitwiseMaskOrCaller<uchar>(int, int, int, const PtrStep, const PtrStep, const PtrStep, PtrStep, cudaStream_t);\r
-    template void bitwiseMaskOrCaller<ushort>(int, int, int, const PtrStep, const PtrStep, const PtrStep, PtrStep, cudaStream_t);\r
-    template void bitwiseMaskOrCaller<uint>(int, int, int, const PtrStep, const PtrStep, const PtrStep, PtrStep, cudaStream_t);\r
+    template void bitwiseMaskOrCaller<uchar>(int, int, int, const PtrStepb, const PtrStepb, const PtrStepb, PtrStepb, cudaStream_t);\r
+    template void bitwiseMaskOrCaller<ushort>(int, int, int, const PtrStepb, const PtrStepb, const PtrStepb, PtrStepb, cudaStream_t);\r
+    template void bitwiseMaskOrCaller<uint>(int, int, int, const PtrStepb, const PtrStepb, const PtrStepb, PtrStepb, cudaStream_t);\r
 \r
 \r
-    void bitwiseAndCaller(int rows, int cols, size_t elem_size1, int cn, const PtrStep src1, \r
-                          const PtrStep src2, PtrStep dst, cudaStream_t stream)\r
+    void bitwiseAndCaller(int rows, int cols, size_t elem_size1, int cn, const PtrStepb src1, \r
+                          const PtrStepb src2, PtrStepb dst, cudaStream_t stream)\r
     {\r
         bitwiseBinOp<BIN_OP_AND>(rows, static_cast<int>(cols * elem_size1 * cn), src1, src2, dst, stream);\r
     }\r
 \r
 \r
     template <typename T>\r
-    void bitwiseMaskAndCaller(int rows, int cols, int cn, const PtrStep src1, const PtrStep src2, \r
-                              const PtrStep mask, PtrStep dst, cudaStream_t stream)\r
+    void bitwiseMaskAndCaller(int rows, int cols, int cn, const PtrStepb src1, const PtrStepb src2, \r
+                              const PtrStepb mask, PtrStepb dst, cudaStream_t stream)\r
     {\r
         bitwiseBinOp<T, BIN_OP_AND>(rows, cols * cn, cn, src1, src2, mask, dst, stream);\r
     }\r
 \r
-    template void bitwiseMaskAndCaller<uchar>(int, int, int, const PtrStep, const PtrStep, const PtrStep, PtrStep, cudaStream_t);\r
-    template void bitwiseMaskAndCaller<ushort>(int, int, int, const PtrStep, const PtrStep, const PtrStep, PtrStep, cudaStream_t);\r
-    template void bitwiseMaskAndCaller<uint>(int, int, int, const PtrStep, const PtrStep, const PtrStep, PtrStep, cudaStream_t);\r
+    template void bitwiseMaskAndCaller<uchar>(int, int, int, const PtrStepb, const PtrStepb, const PtrStepb, PtrStepb, cudaStream_t);\r
+    template void bitwiseMaskAndCaller<ushort>(int, int, int, const PtrStepb, const PtrStepb, const PtrStepb, PtrStepb, cudaStream_t);\r
+    template void bitwiseMaskAndCaller<uint>(int, int, int, const PtrStepb, const PtrStepb, const PtrStepb, PtrStepb, cudaStream_t);\r
 \r
 \r
-    void bitwiseXorCaller(int rows, int cols, size_t elem_size1, int cn, const PtrStep src1, \r
-                          const PtrStep src2, PtrStep dst, cudaStream_t stream)\r
+    void bitwiseXorCaller(int rows, int cols, size_t elem_size1, int cn, const PtrStepb src1, \r
+                          const PtrStepb src2, PtrStepb dst, cudaStream_t stream)\r
     {\r
         bitwiseBinOp<BIN_OP_XOR>(rows, static_cast<int>(cols * elem_size1 * cn), src1, src2, dst, stream);\r
     }\r
 \r
 \r
     template <typename T>\r
-    void bitwiseMaskXorCaller(int rows, int cols, int cn, const PtrStep src1, const PtrStep src2, \r
-                              const PtrStep mask, PtrStep dst, cudaStream_t stream)\r
+    void bitwiseMaskXorCaller(int rows, int cols, int cn, const PtrStepb src1, const PtrStepb src2, \r
+                              const PtrStepb mask, PtrStepb dst, cudaStream_t stream)\r
     {\r
         bitwiseBinOp<T, BIN_OP_XOR>(rows, cols * cn, cn, src1, src2, mask, dst, stream);\r
     }\r
 \r
-    template void bitwiseMaskXorCaller<uchar>(int, int, int, const PtrStep, const PtrStep, const PtrStep, PtrStep, cudaStream_t);\r
-    template void bitwiseMaskXorCaller<ushort>(int, int, int, const PtrStep, const PtrStep, const PtrStep, PtrStep, cudaStream_t);\r
-    template void bitwiseMaskXorCaller<uint>(int, int, int, const PtrStep, const PtrStep, const PtrStep, PtrStep, cudaStream_t);\r
+    template void bitwiseMaskXorCaller<uchar>(int, int, int, const PtrStepb, const PtrStepb, const PtrStepb, PtrStepb, cudaStream_t);\r
+    template void bitwiseMaskXorCaller<ushort>(int, int, int, const PtrStepb, const PtrStepb, const PtrStepb, PtrStepb, cudaStream_t);\r
+    template void bitwiseMaskXorCaller<uint>(int, int, int, const PtrStepb, const PtrStepb, const PtrStepb, PtrStepb, cudaStream_t);\r
 \r
 \r
     //////////////////////////////////////////////////////////////////////////\r
@@ -1551,7 +1551,7 @@ namespace cv { namespace gpu { namespace device
         transform(src1, src2, dst, minimum<T>(), stream);    \r
     }\r
 \r
-    template void min_gpu<uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
+    template void min_gpu<uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
     template void min_gpu<schar >(const DevMem2D_<schar>& src1, const DevMem2D_<schar>& src2, const DevMem2D_<schar>& dst, cudaStream_t stream);\r
     template void min_gpu<ushort>(const DevMem2D_<ushort>& src1, const DevMem2D_<ushort>& src2, const DevMem2D_<ushort>& dst, cudaStream_t stream);\r
     template void min_gpu<short >(const DevMem2D_<short>& src1, const DevMem2D_<short>& src2, const DevMem2D_<short>& dst, cudaStream_t stream);\r
@@ -1565,7 +1565,7 @@ namespace cv { namespace gpu { namespace device
         transform(src1, src2, dst, maximum<T>(), stream);    \r
     }\r
     \r
-    template void max_gpu<uchar >(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
+    template void max_gpu<uchar >(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
     template void max_gpu<schar >(const DevMem2D_<schar>& src1, const DevMem2D_<schar>& src2, const DevMem2D_<schar>& dst, cudaStream_t stream);\r
     template void max_gpu<ushort>(const DevMem2D_<ushort>& src1, const DevMem2D_<ushort>& src2, const DevMem2D_<ushort>& dst, cudaStream_t stream);\r
     template void max_gpu<short >(const DevMem2D_<short>& src1, const DevMem2D_<short>& src2, const DevMem2D_<short>& dst, cudaStream_t stream);\r
@@ -1579,7 +1579,7 @@ namespace cv { namespace gpu { namespace device
         transform(src1, dst, device::bind2nd(minimum<T>(), src2), stream);    \r
     }\r
 \r
-    template void min_gpu<uchar >(const DevMem2D& src1, uchar src2, const DevMem2D& dst, cudaStream_t stream);\r
+    template void min_gpu<uchar >(const DevMem2Db& src1, uchar src2, const DevMem2Db& dst, cudaStream_t stream);\r
     template void min_gpu<schar >(const DevMem2D_<schar>& src1, schar src2, const DevMem2D_<schar>& dst, cudaStream_t stream);\r
     template void min_gpu<ushort>(const DevMem2D_<ushort>& src1, ushort src2, const DevMem2D_<ushort>& dst, cudaStream_t stream);\r
     template void min_gpu<short >(const DevMem2D_<short>& src1, short src2, const DevMem2D_<short>& dst, cudaStream_t stream);\r
@@ -1593,7 +1593,7 @@ namespace cv { namespace gpu { namespace device
         transform(src1, dst, device::bind2nd(maximum<T>(), src2), stream);    \r
     }\r
 \r
-    template void max_gpu<uchar >(const DevMem2D& src1, uchar src2, const DevMem2D& dst, cudaStream_t stream);\r
+    template void max_gpu<uchar >(const DevMem2Db& src1, uchar src2, const DevMem2Db& dst, cudaStream_t stream);\r
     template void max_gpu<schar >(const DevMem2D_<schar>& src1, schar src2, const DevMem2D_<schar>& dst, cudaStream_t stream);\r
     template void max_gpu<ushort>(const DevMem2D_<ushort>& src1, ushort src2, const DevMem2D_<ushort>& dst, cudaStream_t stream);\r
     template void max_gpu<short >(const DevMem2D_<short>& src1, short src2, const DevMem2D_<short>& dst, cudaStream_t stream);\r
@@ -1646,7 +1646,7 @@ namespace cv { namespace gpu { namespace device
     }\r
 \r
     template <typename T>\r
-    void threshold_gpu(const DevMem2D& src, const DevMem2D& dst, T thresh, T maxVal, int type,\r
+    void threshold_gpu(const DevMem2Db& src, const DevMem2Db& dst, T thresh, T maxVal, int type,\r
         cudaStream_t stream)\r
     {\r
         typedef void (*caller_t)(const DevMem2D_<T>& src, const DevMem2D_<T>& dst, T thresh, T maxVal, \r
@@ -1664,13 +1664,13 @@ namespace cv { namespace gpu { namespace device
         callers[type]((DevMem2D_<T>)src, (DevMem2D_<T>)dst, thresh, maxVal, stream);\r
     }\r
 \r
-    template void threshold_gpu<uchar>(const DevMem2D& src, const DevMem2D& dst, uchar thresh, uchar maxVal, int type, cudaStream_t stream);\r
-    template void threshold_gpu<schar>(const DevMem2D& src, const DevMem2D& dst, schar thresh, schar maxVal, int type, cudaStream_t stream);\r
-    template void threshold_gpu<ushort>(const DevMem2D& src, const DevMem2D& dst, ushort thresh, ushort maxVal, int type, cudaStream_t stream);\r
-    template void threshold_gpu<short>(const DevMem2D& src, const DevMem2D& dst, short thresh, short maxVal, int type, cudaStream_t stream);\r
-    template void threshold_gpu<int>(const DevMem2D& src, const DevMem2D& dst, int thresh, int maxVal, int type, cudaStream_t stream);\r
-    template void threshold_gpu<float>(const DevMem2D& src, const DevMem2D& dst, float thresh, float maxVal, int type, cudaStream_t stream);\r
-    template void threshold_gpu<double>(const DevMem2D& src, const DevMem2D& dst, double thresh, double maxVal, int type, cudaStream_t stream);\r
+    template void threshold_gpu<uchar>(const DevMem2Db& src, const DevMem2Db& dst, uchar thresh, uchar maxVal, int type, cudaStream_t stream);\r
+    template void threshold_gpu<schar>(const DevMem2Db& src, const DevMem2Db& dst, schar thresh, schar maxVal, int type, cudaStream_t stream);\r
+    template void threshold_gpu<ushort>(const DevMem2Db& src, const DevMem2Db& dst, ushort thresh, ushort maxVal, int type, cudaStream_t stream);\r
+    template void threshold_gpu<short>(const DevMem2Db& src, const DevMem2Db& dst, short thresh, short maxVal, int type, cudaStream_t stream);\r
+    template void threshold_gpu<int>(const DevMem2Db& src, const DevMem2Db& dst, int thresh, int maxVal, int type, cudaStream_t stream);\r
+    template void threshold_gpu<float>(const DevMem2Db& src, const DevMem2Db& dst, float thresh, float maxVal, int type, cudaStream_t stream);\r
+    template void threshold_gpu<double>(const DevMem2Db& src, const DevMem2Db& dst, double thresh, double maxVal, int type, cudaStream_t stream);\r
 \r
 \r
 \r
@@ -1741,17 +1741,17 @@ namespace cv { namespace gpu { namespace device
     };\r
 \r
     template<typename T>\r
-    void pow_caller(const DevMem2D& src, float power, DevMem2D dst, cudaStream_t stream)\r
+    void pow_caller(const DevMem2Db& src, float power, DevMem2Db dst, cudaStream_t stream)\r
     {\r
         transform((DevMem2D_<T>)src, (DevMem2D_<T>)dst, PowOp<T>(power), stream);\r
     }   \r
 \r
-    template void pow_caller<uchar>(const DevMem2D& src, float power, DevMem2D dst, cudaStream_t stream);\r
-    template void pow_caller<schar>(const DevMem2D& src, float power, DevMem2D dst, cudaStream_t stream);\r
-    template void pow_caller<short>(const DevMem2D& src, float power, DevMem2D dst, cudaStream_t stream);\r
-    template void pow_caller<ushort>(const DevMem2D& src, float power, DevMem2D dst, cudaStream_t stream);\r
-    template void pow_caller<int>(const DevMem2D& src, float power, DevMem2D dst, cudaStream_t stream);\r
-    template void pow_caller<float>(const DevMem2D& src, float power, DevMem2D dst, cudaStream_t stream);\r
+    template void pow_caller<uchar>(const DevMem2Db& src, float power, DevMem2Db dst, cudaStream_t stream);\r
+    template void pow_caller<schar>(const DevMem2Db& src, float power, DevMem2Db dst, cudaStream_t stream);\r
+    template void pow_caller<short>(const DevMem2Db& src, float power, DevMem2Db dst, cudaStream_t stream);\r
+    template void pow_caller<ushort>(const DevMem2Db& src, float power, DevMem2Db dst, cudaStream_t stream);\r
+    template void pow_caller<int>(const DevMem2Db& src, float power, DevMem2Db dst, cudaStream_t stream);\r
+    template void pow_caller<float>(const DevMem2Db& src, float power, DevMem2Db dst, cudaStream_t stream);\r
 \r
 \r
     \r
@@ -1830,7 +1830,7 @@ namespace cv { namespace gpu { namespace device
     };\r
 \r
     template <typename T1, typename T2, typename D>\r
-    void addWeighted_gpu(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream)\r
+    void addWeighted_gpu(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream)\r
     {\r
         cudaSafeCall( cudaSetDoubleForDevice(&alpha) );\r
         cudaSafeCall( cudaSetDoubleForDevice(&beta) );\r
@@ -1841,239 +1841,239 @@ namespace cv { namespace gpu { namespace device
         transform(static_cast< DevMem2D_<T1> >(src1), static_cast< DevMem2D_<T2> >(src2), static_cast< DevMem2D_<D> >(dst), op, stream);\r
     }\r
 \r
-    template void addWeighted_gpu<uchar, uchar, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, uchar, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, uchar, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, uchar, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, uchar, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, uchar, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, uchar, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<uchar, schar, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, schar, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, schar, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, schar, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, schar, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, schar, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, schar, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<uchar, ushort, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, ushort, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, ushort, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, ushort, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, ushort, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, ushort, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, ushort, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<uchar, short, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, short, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, short, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, short, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, short, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, short, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, short, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<uchar, int, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, int, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, int, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, int, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, int, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, int, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, int, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<uchar, float, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, float, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, float, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, float, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, float, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, float, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, float, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<uchar, double, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, double, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, double, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, double, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, double, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, double, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<uchar, double, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-\r
-\r
-    template void addWeighted_gpu<schar, schar, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, schar, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, schar, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, schar, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, schar, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, schar, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, schar, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<schar, ushort, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, ushort, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, ushort, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, ushort, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, ushort, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, ushort, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, ushort, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<schar, short, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, short, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, short, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, short, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, short, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, short, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, short, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<schar, int, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, int, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, int, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, int, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, int, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, int, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, int, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<schar, float, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, float, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, float, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, float, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, float, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, float, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, float, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<schar, double, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, double, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, double, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, double, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, double, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, double, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<schar, double, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-\r
-\r
-    template void addWeighted_gpu<ushort, ushort, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, ushort, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, ushort, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, ushort, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, ushort, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, ushort, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, ushort, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<ushort, short, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, short, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, short, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, short, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, short, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, short, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, short, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<ushort, int, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, int, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, int, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, int, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, int, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, int, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, int, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<ushort, float, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, float, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, float, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, float, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, float, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, float, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, float, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<ushort, double, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, double, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, double, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, double, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, double, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, double, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<ushort, double, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-\r
-\r
-    template void addWeighted_gpu<short, short, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, short, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, short, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, short, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, short, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, short, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, short, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<short, int, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, int, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, int, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, int, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, int, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, int, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, int, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<short, float, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, float, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, float, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, float, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, float, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, float, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, float, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<short, double, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, double, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, double, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, double, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, double, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, double, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<short, double, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, uchar, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, uchar, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, uchar, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, uchar, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, uchar, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, uchar, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, uchar, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<uchar, schar, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, schar, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, schar, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, schar, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, schar, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, schar, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, schar, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<uchar, ushort, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, ushort, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, ushort, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, ushort, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, ushort, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, ushort, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, ushort, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<uchar, short, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, short, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, short, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, short, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, short, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, short, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, short, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<uchar, int, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, int, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, int, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, int, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, int, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, int, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, int, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<uchar, float, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, float, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, float, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, float, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, float, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, float, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, float, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<uchar, double, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, double, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, double, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, double, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, double, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, double, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<uchar, double, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+\r
+\r
+    template void addWeighted_gpu<schar, schar, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, schar, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, schar, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, schar, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, schar, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, schar, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, schar, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<schar, ushort, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, ushort, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, ushort, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, ushort, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, ushort, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, ushort, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, ushort, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<schar, short, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, short, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, short, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, short, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, short, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, short, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, short, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<schar, int, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, int, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, int, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, int, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, int, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, int, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, int, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<schar, float, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, float, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, float, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, float, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, float, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, float, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, float, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<schar, double, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, double, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, double, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, double, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, double, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, double, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<schar, double, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+\r
+\r
+    template void addWeighted_gpu<ushort, ushort, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, ushort, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, ushort, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, ushort, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, ushort, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, ushort, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, ushort, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<ushort, short, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, short, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, short, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, short, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, short, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, short, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, short, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<ushort, int, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, int, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, int, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, int, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, int, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, int, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, int, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<ushort, float, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, float, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, float, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, float, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, float, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, float, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, float, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<ushort, double, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, double, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, double, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, double, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, double, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, double, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<ushort, double, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+\r
+\r
+    template void addWeighted_gpu<short, short, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, short, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, short, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, short, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, short, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, short, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, short, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<short, int, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, int, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, int, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, int, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, int, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, int, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, int, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<short, float, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, float, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, float, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, float, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, float, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, float, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, float, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<short, double, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, double, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, double, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, double, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, double, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, double, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<short, double, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
 \r
     \r
 \r
-    template void addWeighted_gpu<int, int, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<int, int, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<int, int, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<int, int, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<int, int, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<int, int, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<int, int, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<int, float, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<int, float, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<int, float, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<int, float, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<int, float, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<int, float, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<int, float, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<int, double, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<int, double, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<int, double, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<int, double, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<int, double, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<int, double, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<int, double, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<int, int, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<int, int, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<int, int, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<int, int, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<int, int, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<int, int, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<int, int, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<int, float, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<int, float, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<int, float, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<int, float, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<int, float, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<int, float, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<int, float, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<int, double, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<int, double, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<int, double, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<int, double, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<int, double, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<int, double, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<int, double, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
 \r
     \r
 \r
-    template void addWeighted_gpu<float, float, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<float, float, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<float, float, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<float, float, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<float, float, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<float, float, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<float, float, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-\r
-    template void addWeighted_gpu<float, double, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<float, double, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<float, double, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<float, double, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<float, double, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<float, double, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<float, double, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<float, float, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<float, float, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<float, float, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<float, float, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<float, float, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<float, float, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<float, float, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+\r
+    template void addWeighted_gpu<float, double, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<float, double, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<float, double, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<float, double, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<float, double, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<float, double, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<float, double, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
 \r
     \r
 \r
-    template void addWeighted_gpu<double, double, uchar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<double, double, schar>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<double, double, ushort>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<double, double, short>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<double, double, int>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<double, double, float>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
-    template void addWeighted_gpu<double, double, double>(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<double, double, uchar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<double, double, schar>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<double, double, ushort>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<double, double, short>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<double, double, int>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<double, double, float>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void addWeighted_gpu<double, double, double>(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
 }}}\r

diff --git a/modules/gpu/src/cuda/hist.cu b/modules/gpu/src/cuda/hist.cu
index cafcb09..c420c97 100644 (file)
--- a/modules/gpu/src/cuda/hist.cu
+++ b/modules/gpu/src/cuda/hist.cu
@@ -105,7 +105,7 @@ namespace cv { namespace gpu { namespace histograms
         if (x + 3 < cols) addByte(s_WarpHist, (data >> 24) & 0xFFU, tag);\r
     }\r
 \r
-    __global__ void histogram256(const PtrStep_<uint> d_Data, uint* d_PartialHistograms, uint dataCount, uint cols)\r
+    __global__ void histogram256(const PtrStep<uint> d_Data, uint* d_PartialHistograms, uint dataCount, uint cols)\r
     {\r
         //Per-warp subhistogram storage\r
         __shared__ uint s_Hist[HISTOGRAM256_THREADBLOCK_MEMORY];\r
@@ -171,7 +171,7 @@ namespace cv { namespace gpu { namespace histograms
             d_Histogram[blockIdx.x] = saturate_cast<int>(data[0]);\r
     }\r
 \r
-    void histogram256_gpu(DevMem2D src, int* hist, uint* buf, cudaStream_t stream)\r
+    void histogram256_gpu(DevMem2Db src, int* hist, uint* buf, cudaStream_t stream)\r
     {\r
         histogram256<<<PARTIAL_HISTOGRAM256_COUNT, HISTOGRAM256_THREADBLOCK_SIZE, 0, stream>>>(\r
             DevMem2D_<uint>(src),\r
@@ -191,7 +191,7 @@ namespace cv { namespace gpu { namespace histograms
 \r
     __constant__ int c_lut[256];\r
 \r
-    __global__ void equalizeHist(const DevMem2D src, PtrStep dst)\r
+    __global__ void equalizeHist(const DevMem2Db src, PtrStepb dst)\r
     {\r
         const int x = blockIdx.x * blockDim.x + threadIdx.x;\r
         const int y = blockIdx.y * blockDim.y + threadIdx.y;\r
@@ -204,7 +204,7 @@ namespace cv { namespace gpu { namespace histograms
         }\r
     }\r
 \r
-    void equalizeHist_gpu(DevMem2D src, DevMem2D dst, const int* lut, cudaStream_t stream)\r
+    void equalizeHist_gpu(DevMem2Db src, DevMem2Db dst, const int* lut, cudaStream_t stream)\r
     {\r
         dim3 block(16, 16);\r
         dim3 grid(divUp(src.cols, block.x), divUp(src.rows, block.y));\r

diff --git a/modules/gpu/src/cuda/hog.cu b/modules/gpu/src/cuda/hog.cu
index 6ef4c4c..8ac6196 100644 (file)
--- a/modules/gpu/src/cuda/hog.cu
+++ b/modules/gpu/src/cuda/hog.cu
@@ -187,7 +187,7 @@ __global__ void compute_hists_kernel_many_blocks(const int img_block_width, cons
 \r
 void compute_hists(int nbins, int block_stride_x, int block_stride_y, \r
                    int height, int width, const DevMem2Df& grad, \r
-                   const DevMem2D& qangle, float sigma, float* block_hists)                             \r
+                   const DevMem2Db& qangle, float sigma, float* block_hists)                             \r
 {\r
     const int nblocks = 1;\r
 \r
@@ -614,8 +614,8 @@ __global__ void compute_gradients_8UC4_kernel(int height, int width, const PtrEl
 }\r
 \r
 \r
-void compute_gradients_8UC4(int nbins, int height, int width, const DevMem2D& img, \r
-                            float angle_scale, DevMem2Df grad, DevMem2D qangle, bool correct_gamma)\r
+void compute_gradients_8UC4(int nbins, int height, int width, const DevMem2Db& img, \r
+                            float angle_scale, DevMem2Df grad, DevMem2Db qangle, bool correct_gamma)\r
 {\r
     const int nthreads = 256;\r
 \r
@@ -686,8 +686,8 @@ __global__ void compute_gradients_8UC1_kernel(int height, int width, const PtrEl
 }\r
 \r
 \r
-void compute_gradients_8UC1(int nbins, int height, int width, const DevMem2D& img, \r
-                            float angle_scale, DevMem2Df grad, DevMem2D qangle, bool correct_gamma)\r
+void compute_gradients_8UC1(int nbins, int height, int width, const DevMem2Db& img, \r
+                            float angle_scale, DevMem2Df grad, DevMem2Db qangle, bool correct_gamma)\r
 {\r
     const int nthreads = 256;\r
 \r
@@ -734,7 +734,7 @@ __global__ void resize_for_hog_kernel(float sx, float sy, DevMem2D_<uchar4> dst,
 }\r
 \r
 template<class T, class TEX> \r
-static void resize_for_hog(const DevMem2D& src, DevMem2D dst, TEX& tex)\r
+static void resize_for_hog(const DevMem2Db& src, DevMem2Db dst, TEX& tex)\r
 {\r
     tex.filterMode = cudaFilterModeLinear;\r
 \r
@@ -765,7 +765,7 @@ static void resize_for_hog(const DevMem2D& src, DevMem2D dst, TEX& tex)
     cudaSafeCall( cudaUnbindTexture(tex) );\r
 }\r
 \r
-void resize_8UC1(const DevMem2D& src, DevMem2D dst) { resize_for_hog<uchar> (src, dst, resize8UC1_tex); }\r
-void resize_8UC4(const DevMem2D& src, DevMem2D dst) { resize_for_hog<uchar4>(src, dst, resize8UC4_tex); }\r
+void resize_8UC1(const DevMem2Db& src, DevMem2Db dst) { resize_for_hog<uchar> (src, dst, resize8UC1_tex); }\r
+void resize_8UC4(const DevMem2Db& src, DevMem2Db dst) { resize_for_hog<uchar4>(src, dst, resize8UC4_tex); }\r
 \r
 }}}\r

diff --git a/modules/gpu/src/cuda/imgproc.cu b/modules/gpu/src/cuda/imgproc.cu
index d2ce30e..4d54895 100644 (file)
--- a/modules/gpu/src/cuda/imgproc.cu
+++ b/modules/gpu/src/cuda/imgproc.cu
@@ -145,7 +145,7 @@ namespace cv { namespace gpu { namespace imgproc
         }\r
     }\r
 \r
-    extern "C" void meanShiftFiltering_gpu(const DevMem2D& src, DevMem2D dst, int sp, int sr, int maxIter, float eps)\r
+    extern "C" void meanShiftFiltering_gpu(const DevMem2Db& src, DevMem2Db dst, int sp, int sr, int maxIter, float eps)\r
     {\r
         dim3 grid(1, 1, 1);\r
         dim3 threads(32, 8, 1);\r
@@ -161,7 +161,7 @@ namespace cv { namespace gpu { namespace imgproc
         cudaSafeCall( cudaDeviceSynchronize() );\r
         cudaSafeCall( cudaUnbindTexture( tex_meanshift ) );        \r
     }\r
-    extern "C" void meanShiftProc_gpu(const DevMem2D& src, DevMem2D dstr, DevMem2D dstsp, int sp, int sr, int maxIter, float eps) \r
+    extern "C" void meanShiftProc_gpu(const DevMem2Db& src, DevMem2Db dstr, DevMem2Db dstsp, int sp, int sr, int maxIter, float eps) \r
     {\r
         dim3 grid(1, 1, 1);\r
         dim3 threads(32, 8, 1);\r
@@ -281,7 +281,7 @@ namespace cv { namespace gpu { namespace imgproc
     }\r
 \r
 \r
-    void drawColorDisp_gpu(const DevMem2D& src, const DevMem2D& dst, int ndisp, const cudaStream_t& stream)\r
+    void drawColorDisp_gpu(const DevMem2Db& src, const DevMem2Db& dst, int ndisp, const cudaStream_t& stream)\r
     {\r
         dim3 threads(16, 16, 1);\r
         dim3 grid(1, 1, 1);\r
@@ -295,7 +295,7 @@ namespace cv { namespace gpu { namespace imgproc
             cudaSafeCall( cudaDeviceSynchronize() ); \r
     }\r
 \r
-    void drawColorDisp_gpu(const DevMem2D_<short>& src, const DevMem2D& dst, int ndisp, const cudaStream_t& stream)\r
+    void drawColorDisp_gpu(const DevMem2D_<short>& src, const DevMem2Db& dst, int ndisp, const cudaStream_t& stream)\r
     {\r
         dim3 threads(32, 8, 1);\r
         dim3 grid(1, 1, 1);\r
@@ -360,7 +360,7 @@ namespace cv { namespace gpu { namespace imgproc
             cudaSafeCall( cudaDeviceSynchronize() );\r
     }\r
 \r
-    void reprojectImageTo3D_gpu(const DevMem2D& disp, const DevMem2Df& xyzw, const float* q, const cudaStream_t& stream)\r
+    void reprojectImageTo3D_gpu(const DevMem2Db& disp, const DevMem2Df& xyzw, const float* q, const cudaStream_t& stream)\r
     {\r
         reprojectImageTo3D_caller(disp, xyzw, q, stream);\r
     }\r
@@ -406,7 +406,7 @@ namespace cv { namespace gpu { namespace imgproc
     texture<float, 2> harrisDyTex;\r
 \r
     __global__ void cornerHarris_kernel(const int cols, const int rows, const int block_size, const float k,\r
-                                        PtrStep dst)\r
+                                        PtrStepb dst)\r
     {\r
         const unsigned int x = blockIdx.x * blockDim.x + threadIdx.x;\r
         const unsigned int y = blockIdx.y * blockDim.y + threadIdx.y;\r
@@ -440,7 +440,7 @@ namespace cv { namespace gpu { namespace imgproc
 \r
     template <typename BR, typename BC>\r
     __global__ void cornerHarris_kernel(const int cols, const int rows, const int block_size, const float k,\r
-                                        PtrStep dst, BR border_row, BC border_col)\r
+                                        PtrStepb dst, BR border_row, BC border_col)\r
     {\r
         const unsigned int x = blockIdx.x * blockDim.x + threadIdx.x;\r
         const unsigned int y = blockIdx.y * blockDim.y + threadIdx.y;\r
@@ -474,7 +474,7 @@ namespace cv { namespace gpu { namespace imgproc
         }\r
     }\r
 \r
-    void cornerHarris_caller(const int block_size, const float k, const DevMem2D Dx, const DevMem2D Dy, DevMem2D dst, \r
+    void cornerHarris_caller(const int block_size, const float k, const DevMem2Db Dx, const DevMem2Db Dy, DevMem2Db dst, \r
                              int border_type)\r
     {\r
         const int rows = Dx.rows;\r
@@ -518,7 +518,7 @@ namespace cv { namespace gpu { namespace imgproc
     texture<float, 2> minEigenValDyTex;\r
 \r
     __global__ void cornerMinEigenVal_kernel(const int cols, const int rows, const int block_size, \r
-                                             PtrStep dst)\r
+                                             PtrStepb dst)\r
     {\r
         const unsigned int x = blockIdx.x * blockDim.x + threadIdx.x;\r
         const unsigned int y = blockIdx.y * blockDim.y + threadIdx.y;\r
@@ -555,7 +555,7 @@ namespace cv { namespace gpu { namespace imgproc
 \r
     template <typename BR, typename BC>\r
     __global__ void cornerMinEigenVal_kernel(const int cols, const int rows, const int block_size, \r
-                                             PtrStep dst, BR border_row, BC border_col)\r
+                                             PtrStepb dst, BR border_row, BC border_col)\r
     {\r
         const unsigned int x = blockIdx.x * blockDim.x + threadIdx.x;\r
         const unsigned int y = blockIdx.y * blockDim.y + threadIdx.y;\r
@@ -591,7 +591,7 @@ namespace cv { namespace gpu { namespace imgproc
         }\r
     }\r
 \r
-    void cornerMinEigenVal_caller(const int block_size, const DevMem2D Dx, const DevMem2D Dy, DevMem2D dst,\r
+    void cornerMinEigenVal_caller(const int block_size, const DevMem2Db Dx, const DevMem2Db Dy, DevMem2Db dst,\r
                                   int border_type)\r
     {\r
         const int rows = Dx.rows;\r
@@ -631,7 +631,7 @@ namespace cv { namespace gpu { namespace imgproc
 \r
 ////////////////////////////// Column Sum //////////////////////////////////////\r
 \r
-    __global__ void column_sumKernel_32F(int cols, int rows, const PtrStep src, const PtrStep dst)\r
+    __global__ void column_sumKernel_32F(int cols, int rows, const PtrStepb src, const PtrStepb dst)\r
     {\r
         int x = blockIdx.x * blockDim.x + threadIdx.x;\r
 \r
@@ -652,7 +652,7 @@ namespace cv { namespace gpu { namespace imgproc
     }\r
 \r
 \r
-    void columnSum_32F(const DevMem2D src, const DevMem2D dst)\r
+    void columnSum_32F(const DevMem2Db src, const DevMem2Db dst)\r
     {\r
         dim3 threads(256);\r
         dim3 grid(divUp(src.cols, threads.x));\r
@@ -667,7 +667,7 @@ namespace cv { namespace gpu { namespace imgproc
     //////////////////////////////////////////////////////////////////////////\r
     // mulSpectrums\r
 \r
-    __global__ void mulSpectrumsKernel(const PtrStep_<cufftComplex> a, const PtrStep_<cufftComplex> b, \r
+    __global__ void mulSpectrumsKernel(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, \r
                                        DevMem2D_<cufftComplex> c)\r
     {\r
         const int x = blockIdx.x * blockDim.x + threadIdx.x;    \r
@@ -680,7 +680,7 @@ namespace cv { namespace gpu { namespace imgproc
     }\r
 \r
 \r
-    void mulSpectrums(const PtrStep_<cufftComplex> a, const PtrStep_<cufftComplex> b, \r
+    void mulSpectrums(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, \r
                       DevMem2D_<cufftComplex> c)\r
     {\r
         dim3 threads(256);\r
@@ -697,7 +697,7 @@ namespace cv { namespace gpu { namespace imgproc
     // mulSpectrums_CONJ\r
 \r
     __global__ void mulSpectrumsKernel_CONJ(\r
-            const PtrStep_<cufftComplex> a, const PtrStep_<cufftComplex> b,\r
+            const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b,\r
             DevMem2D_<cufftComplex> c)\r
     {\r
         const int x = blockIdx.x * blockDim.x + threadIdx.x;    \r
@@ -710,7 +710,7 @@ namespace cv { namespace gpu { namespace imgproc
     }\r
 \r
 \r
-    void mulSpectrums_CONJ(const PtrStep_<cufftComplex> a, const PtrStep_<cufftComplex> b, \r
+    void mulSpectrums_CONJ(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, \r
                            DevMem2D_<cufftComplex> c)\r
     {\r
         dim3 threads(256);\r
@@ -727,7 +727,7 @@ namespace cv { namespace gpu { namespace imgproc
     // mulAndScaleSpectrums\r
 \r
     __global__ void mulAndScaleSpectrumsKernel(\r
-            const PtrStep_<cufftComplex> a, const PtrStep_<cufftComplex> b, \r
+            const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, \r
             float scale, DevMem2D_<cufftComplex> c)\r
     {\r
         const int x = blockIdx.x * blockDim.x + threadIdx.x;\r
@@ -741,7 +741,7 @@ namespace cv { namespace gpu { namespace imgproc
     }\r
 \r
 \r
-    void mulAndScaleSpectrums(const PtrStep_<cufftComplex> a, const PtrStep_<cufftComplex> b,\r
+    void mulAndScaleSpectrums(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b,\r
                               float scale, DevMem2D_<cufftComplex> c)\r
     {\r
         dim3 threads(256);\r
@@ -758,7 +758,7 @@ namespace cv { namespace gpu { namespace imgproc
     // mulAndScaleSpectrums_CONJ\r
 \r
     __global__ void mulAndScaleSpectrumsKernel_CONJ(\r
-            const PtrStep_<cufftComplex> a, const PtrStep_<cufftComplex> b,\r
+            const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b,\r
             float scale, DevMem2D_<cufftComplex> c)\r
     {\r
         const int x = blockIdx.x * blockDim.x + threadIdx.x;\r
@@ -772,7 +772,7 @@ namespace cv { namespace gpu { namespace imgproc
     }\r
 \r
 \r
-    void mulAndScaleSpectrums_CONJ(const PtrStep_<cufftComplex> a, const PtrStep_<cufftComplex> b,\r
+    void mulAndScaleSpectrums_CONJ(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b,\r
                                   float scale, DevMem2D_<cufftComplex> c)\r
     {\r
         dim3 threads(256);\r

diff --git a/modules/gpu/src/cuda/match_template.cu b/modules/gpu/src/cuda/match_template.cu
index 3f8e92a..242cf3c 100644 (file)
--- a/modules/gpu/src/cuda/match_template.cu
+++ b/modules/gpu/src/cuda/match_template.cu
@@ -81,7 +81,7 @@ __device__ __forceinline__ float4 sub(uchar4 a, uchar4 b) { return make_float4(a
 \r
 template <typename T, int cn>\r
 __global__ void matchTemplateNaiveKernel_CCORR(\r
-        int w, int h, const PtrStep image, const PtrStep templ, \r
+        int w, int h, const PtrStepb image, const PtrStepb templ, \r
         DevMem2Df result)\r
 {\r
     typedef typename TypeVec<T, cn>::vec_type Type;\r
@@ -107,7 +107,7 @@ __global__ void matchTemplateNaiveKernel_CCORR(
 }\r
 \r
 \r
-void matchTemplateNaive_CCORR_32F(const DevMem2D image, const DevMem2D templ,\r
+void matchTemplateNaive_CCORR_32F(const DevMem2Db image, const DevMem2Db templ,\r
                                   DevMem2Df result, int cn)\r
 {\r
     dim3 threads(32, 8);\r
@@ -138,7 +138,7 @@ void matchTemplateNaive_CCORR_32F(const DevMem2D image, const DevMem2D templ,
 }\r
 \r
 \r
-void matchTemplateNaive_CCORR_8U(const DevMem2D image, const DevMem2D templ,\r
+void matchTemplateNaive_CCORR_8U(const DevMem2Db image, const DevMem2Db templ,\r
                                  DevMem2Df result, int cn)\r
 {\r
     dim3 threads(32, 8);\r
@@ -171,7 +171,7 @@ void matchTemplateNaive_CCORR_8U(const DevMem2D image, const DevMem2D templ,
 \r
 template <typename T, int cn>\r
 __global__ void matchTemplateNaiveKernel_SQDIFF(\r
-        int w, int h, const PtrStep image, const PtrStep templ, \r
+        int w, int h, const PtrStepb image, const PtrStepb templ, \r
         DevMem2Df result)\r
 {\r
     typedef typename TypeVec<T, cn>::vec_type Type;\r
@@ -201,7 +201,7 @@ __global__ void matchTemplateNaiveKernel_SQDIFF(
 }\r
 \r
 \r
-void matchTemplateNaive_SQDIFF_32F(const DevMem2D image, const DevMem2D templ,\r
+void matchTemplateNaive_SQDIFF_32F(const DevMem2Db image, const DevMem2Db templ,\r
                                    DevMem2Df result, int cn)\r
 {\r
     dim3 threads(32, 8);\r
@@ -232,7 +232,7 @@ void matchTemplateNaive_SQDIFF_32F(const DevMem2D image, const DevMem2D templ,
 }\r
 \r
 \r
-void matchTemplateNaive_SQDIFF_8U(const DevMem2D image, const DevMem2D templ,\r
+void matchTemplateNaive_SQDIFF_8U(const DevMem2Db image, const DevMem2Db templ,\r
                                   DevMem2Df result, int cn)\r
 {\r
     dim3 threads(32, 8);\r
@@ -265,7 +265,7 @@ void matchTemplateNaive_SQDIFF_8U(const DevMem2D image, const DevMem2D templ,
 \r
 template <int cn>\r
 __global__ void matchTemplatePreparedKernel_SQDIFF_8U(\r
-        int w, int h, const PtrStep_<unsigned long long> image_sqsum, \r
+        int w, int h, const PtrStep<unsigned long long> image_sqsum, \r
         unsigned int templ_sqsum, DevMem2Df result)\r
 {\r
     const int x = blockIdx.x * blockDim.x + threadIdx.x;\r
@@ -338,7 +338,7 @@ __device__ float normAcc_SQDIFF(float num, float denum)
 \r
 template <int cn>\r
 __global__ void matchTemplatePreparedKernel_SQDIFF_NORMED_8U(\r
-        int w, int h, const PtrStep_<unsigned long long> image_sqsum, \r
+        int w, int h, const PtrStep<unsigned long long> image_sqsum, \r
         unsigned int templ_sqsum, DevMem2Df result)\r
 {\r
     const int x = blockIdx.x * blockDim.x + threadIdx.x;\r
@@ -389,7 +389,7 @@ void matchTemplatePrepared_SQDIFF_NORMED_8U(
 \r
 __global__ void matchTemplatePreparedKernel_CCOFF_8U(\r
         int w, int h, float templ_sum_scale, \r
-        const PtrStep_<unsigned int> image_sum, DevMem2Df result)\r
+        const PtrStep<unsigned int> image_sum, DevMem2Df result)\r
 {\r
     const int x = blockIdx.x * blockDim.x + threadIdx.x;\r
     const int y = blockIdx.y * blockDim.y + threadIdx.y;\r
@@ -421,8 +421,8 @@ void matchTemplatePrepared_CCOFF_8U(
 \r
 __global__ void matchTemplatePreparedKernel_CCOFF_8UC2(\r
         int w, int h, float templ_sum_scale_r, float templ_sum_scale_g,\r
-        const PtrStep_<unsigned int> image_sum_r,\r
-        const PtrStep_<unsigned int> image_sum_g,\r
+        const PtrStep<unsigned int> image_sum_r,\r
+        const PtrStep<unsigned int> image_sum_g,\r
         DevMem2Df result)\r
 {\r
     const int x = blockIdx.x * blockDim.x + threadIdx.x;\r
@@ -466,9 +466,9 @@ __global__ void matchTemplatePreparedKernel_CCOFF_8UC3(
         float templ_sum_scale_r,\r
         float templ_sum_scale_g,\r
         float templ_sum_scale_b,\r
-        const PtrStep_<unsigned int> image_sum_r,\r
-        const PtrStep_<unsigned int> image_sum_g,\r
-        const PtrStep_<unsigned int> image_sum_b,\r
+        const PtrStep<unsigned int> image_sum_r,\r
+        const PtrStep<unsigned int> image_sum_g,\r
+        const PtrStep<unsigned int> image_sum_b,\r
         DevMem2Df result)\r
 {\r
     const int x = blockIdx.x * blockDim.x + threadIdx.x;\r
@@ -523,10 +523,10 @@ __global__ void matchTemplatePreparedKernel_CCOFF_8UC4(
         float templ_sum_scale_g,\r
         float templ_sum_scale_b,\r
         float templ_sum_scale_a,\r
-        const PtrStep_<unsigned int> image_sum_r,\r
-        const PtrStep_<unsigned int> image_sum_g,\r
-        const PtrStep_<unsigned int> image_sum_b,\r
-        const PtrStep_<unsigned int> image_sum_a,\r
+        const PtrStep<unsigned int> image_sum_r,\r
+        const PtrStep<unsigned int> image_sum_g,\r
+        const PtrStep<unsigned int> image_sum_b,\r
+        const PtrStep<unsigned int> image_sum_a,\r
         DevMem2Df result)\r
 {\r
     const int x = blockIdx.x * blockDim.x + threadIdx.x;\r
@@ -586,8 +586,8 @@ void matchTemplatePrepared_CCOFF_8UC4(
 __global__ void matchTemplatePreparedKernel_CCOFF_NORMED_8U(\r
         int w, int h, float weight, \r
         float templ_sum_scale, float templ_sqsum_scale,\r
-        const PtrStep_<unsigned int> image_sum, \r
-        const PtrStep_<unsigned long long> image_sqsum,\r
+        const PtrStep<unsigned int> image_sum, \r
+        const PtrStep<unsigned long long> image_sqsum,\r
         DevMem2Df result)\r
 {\r
     const int x = blockIdx.x * blockDim.x + threadIdx.x;\r
@@ -633,8 +633,8 @@ __global__ void matchTemplatePreparedKernel_CCOFF_NORMED_8UC2(
         int w, int h, float weight, \r
         float templ_sum_scale_r, float templ_sum_scale_g, \r
         float templ_sqsum_scale,\r
-        const PtrStep_<unsigned int> image_sum_r, const PtrStep_<unsigned long long> image_sqsum_r,\r
-        const PtrStep_<unsigned int> image_sum_g, const PtrStep_<unsigned long long> image_sqsum_g,\r
+        const PtrStep<unsigned int> image_sum_r, const PtrStep<unsigned long long> image_sqsum_r,\r
+        const PtrStep<unsigned int> image_sum_g, const PtrStep<unsigned long long> image_sqsum_g,\r
         DevMem2Df result)\r
 {\r
     const int x = blockIdx.x * blockDim.x + threadIdx.x;\r
@@ -697,9 +697,9 @@ __global__ void matchTemplatePreparedKernel_CCOFF_NORMED_8UC3(
         int w, int h, float weight, \r
         float templ_sum_scale_r, float templ_sum_scale_g, float templ_sum_scale_b, \r
         float templ_sqsum_scale,\r
-        const PtrStep_<unsigned int> image_sum_r, const PtrStep_<unsigned long long> image_sqsum_r,\r
-        const PtrStep_<unsigned int> image_sum_g, const PtrStep_<unsigned long long> image_sqsum_g,\r
-        const PtrStep_<unsigned int> image_sum_b, const PtrStep_<unsigned long long> image_sqsum_b,\r
+        const PtrStep<unsigned int> image_sum_r, const PtrStep<unsigned long long> image_sqsum_r,\r
+        const PtrStep<unsigned int> image_sum_g, const PtrStep<unsigned long long> image_sqsum_g,\r
+        const PtrStep<unsigned int> image_sum_b, const PtrStep<unsigned long long> image_sqsum_b,\r
         DevMem2Df result)\r
 {\r
     const int x = blockIdx.x * blockDim.x + threadIdx.x;\r
@@ -775,10 +775,10 @@ __global__ void matchTemplatePreparedKernel_CCOFF_NORMED_8UC4(
         int w, int h, float weight, \r
         float templ_sum_scale_r, float templ_sum_scale_g, float templ_sum_scale_b, \r
         float templ_sum_scale_a, float templ_sqsum_scale,\r
-        const PtrStep_<unsigned int> image_sum_r, const PtrStep_<unsigned long long> image_sqsum_r,\r
-        const PtrStep_<unsigned int> image_sum_g, const PtrStep_<unsigned long long> image_sqsum_g,\r
-        const PtrStep_<unsigned int> image_sum_b, const PtrStep_<unsigned long long> image_sqsum_b,\r
-        const PtrStep_<unsigned int> image_sum_a, const PtrStep_<unsigned long long> image_sqsum_a,\r
+        const PtrStep<unsigned int> image_sum_r, const PtrStep<unsigned long long> image_sqsum_r,\r
+        const PtrStep<unsigned int> image_sum_g, const PtrStep<unsigned long long> image_sqsum_g,\r
+        const PtrStep<unsigned int> image_sum_b, const PtrStep<unsigned long long> image_sqsum_b,\r
+        const PtrStep<unsigned int> image_sum_a, const PtrStep<unsigned long long> image_sqsum_a,\r
         DevMem2Df result)\r
 {\r
     const int x = blockIdx.x * blockDim.x + threadIdx.x;\r
@@ -863,7 +863,7 @@ void matchTemplatePrepared_CCOFF_NORMED_8UC4(
 \r
 template <int cn>\r
 __global__ void normalizeKernel_8U(\r
-        int w, int h, const PtrStep_<unsigned long long> image_sqsum, \r
+        int w, int h, const PtrStep<unsigned long long> image_sqsum, \r
         unsigned int templ_sqsum, DevMem2Df result)\r
 {\r
     const int x = blockIdx.x * blockDim.x + threadIdx.x;\r
@@ -906,7 +906,7 @@ void normalize_8U(int w, int h, const DevMem2D_<unsigned long long> image_sqsum,
 \r
 \r
 template <int cn>\r
-__global__ void extractFirstChannel_32F(const PtrStep image, DevMem2Df result)\r
+__global__ void extractFirstChannel_32F(const PtrStepb image, DevMem2Df result)\r
 {\r
     typedef typename TypeVec<float, cn>::vec_type Typef;\r
 \r
@@ -921,7 +921,7 @@ __global__ void extractFirstChannel_32F(const PtrStep image, DevMem2Df result)
 }\r
 \r
 \r
-void extractFirstChannel_32F(const DevMem2D image, DevMem2Df result, int cn)\r
+void extractFirstChannel_32F(const DevMem2Db image, DevMem2Df result, int cn)\r
 {\r
     dim3 threads(32, 8);\r
     dim3 grid(divUp(result.cols, threads.x), divUp(result.rows, threads.y));\r

diff --git a/modules/gpu/src/cuda/matrix_operations.cu b/modules/gpu/src/cuda/matrix_operations.cu
index f6ba358..b2def00 100644 (file)
--- a/modules/gpu/src/cuda/matrix_operations.cu
+++ b/modules/gpu/src/cuda/matrix_operations.cu
@@ -73,10 +73,10 @@ namespace cv { namespace gpu { namespace device {
                 mat_dst[idx] = mat_src[idx];\r
             }\r
     }\r
-    typedef void (*CopyToFunc)(const DevMem2D& mat_src, const DevMem2D& mat_dst, const DevMem2D& mask, int channels, const cudaStream_t & stream);\r
+    typedef void (*CopyToFunc)(const DevMem2Db& mat_src, const DevMem2Db& mat_dst, const DevMem2Db& mask, int channels, const cudaStream_t & stream);\r
 \r
     template<typename T>\r
-    void copy_to_with_mask_run(const DevMem2D& mat_src, const DevMem2D& mat_dst, const DevMem2D& mask, int channels, const cudaStream_t & stream)\r
+    void copy_to_with_mask_run(const DevMem2Db& mat_src, const DevMem2Db& mat_dst, const DevMem2Db& mask, int channels, const cudaStream_t & stream)\r
     {\r
         dim3 threadsPerBlock(16,16, 1);\r
         dim3 numBlocks ( divUp(mat_src.cols * channels , threadsPerBlock.x) , divUp(mat_src.rows , threadsPerBlock.y), 1);\r
@@ -89,7 +89,7 @@ namespace cv { namespace gpu { namespace device {
             cudaSafeCall ( cudaDeviceSynchronize() );\r
     }\r
 \r
-    void copy_to_with_mask(const DevMem2D& mat_src, DevMem2D mat_dst, int depth, const DevMem2D& mask, int channels, const cudaStream_t & stream)\r
+    void copy_to_with_mask(const DevMem2Db& mat_src, DevMem2Db mat_dst, int depth, const DevMem2Db& mask, int channels, const cudaStream_t & stream)\r
     {\r
         static CopyToFunc tab[8] =\r
         {\r
@@ -187,7 +187,7 @@ namespace cv { namespace gpu { namespace device {
             }\r
     }\r
     template <typename T>\r
-    void set_to_gpu(const DevMem2D& mat, const T* scalar, const DevMem2D& mask, int channels, cudaStream_t stream)\r
+    void set_to_gpu(const DevMem2Db& mat, const T* scalar, const DevMem2Db& mask, int channels, cudaStream_t stream)\r
     {\r
         writeScalar(scalar);\r
 \r
@@ -201,16 +201,16 @@ namespace cv { namespace gpu { namespace device {
             cudaSafeCall ( cudaDeviceSynchronize() );\r
     }\r
 \r
-    template void set_to_gpu<uchar >(const DevMem2D& mat, const uchar* scalar, const DevMem2D& mask, int channels, cudaStream_t stream);\r
-    template void set_to_gpu<schar >(const DevMem2D& mat, const schar* scalar, const DevMem2D& mask, int channels, cudaStream_t stream);\r
-    template void set_to_gpu<ushort>(const DevMem2D& mat, const ushort* scalar, const DevMem2D& mask, int channels, cudaStream_t stream);\r
-    template void set_to_gpu<short >(const DevMem2D& mat, const short* scalar, const DevMem2D& mask, int channels, cudaStream_t stream);\r
-    template void set_to_gpu<int   >(const DevMem2D& mat, const int* scalar, const DevMem2D& mask, int channels, cudaStream_t stream);\r
-    template void set_to_gpu<float >(const DevMem2D& mat, const float* scalar, const DevMem2D& mask, int channels, cudaStream_t stream);\r
-    template void set_to_gpu<double>(const DevMem2D& mat, const double* scalar, const DevMem2D& mask, int channels, cudaStream_t stream);\r
+    template void set_to_gpu<uchar >(const DevMem2Db& mat, const uchar* scalar, const DevMem2Db& mask, int channels, cudaStream_t stream);\r
+    template void set_to_gpu<schar >(const DevMem2Db& mat, const schar* scalar, const DevMem2Db& mask, int channels, cudaStream_t stream);\r
+    template void set_to_gpu<ushort>(const DevMem2Db& mat, const ushort* scalar, const DevMem2Db& mask, int channels, cudaStream_t stream);\r
+    template void set_to_gpu<short >(const DevMem2Db& mat, const short* scalar, const DevMem2Db& mask, int channels, cudaStream_t stream);\r
+    template void set_to_gpu<int   >(const DevMem2Db& mat, const int* scalar, const DevMem2Db& mask, int channels, cudaStream_t stream);\r
+    template void set_to_gpu<float >(const DevMem2Db& mat, const float* scalar, const DevMem2Db& mask, int channels, cudaStream_t stream);\r
+    template void set_to_gpu<double>(const DevMem2Db& mat, const double* scalar, const DevMem2Db& mask, int channels, cudaStream_t stream);\r
 \r
     template <typename T>\r
-    void set_to_gpu(const DevMem2D& mat, const T* scalar, int channels, cudaStream_t stream)\r
+    void set_to_gpu(const DevMem2Db& mat, const T* scalar, int channels, cudaStream_t stream)\r
     {\r
         writeScalar(scalar);\r
 \r
@@ -224,13 +224,13 @@ namespace cv { namespace gpu { namespace device {
             cudaSafeCall ( cudaDeviceSynchronize() );\r
     }\r
 \r
-    template void set_to_gpu<uchar >(const DevMem2D& mat, const uchar* scalar, int channels, cudaStream_t stream);\r
-    template void set_to_gpu<schar >(const DevMem2D& mat, const schar* scalar, int channels, cudaStream_t stream);\r
-    template void set_to_gpu<ushort>(const DevMem2D& mat, const ushort* scalar, int channels, cudaStream_t stream);\r
-    template void set_to_gpu<short >(const DevMem2D& mat, const short* scalar, int channels, cudaStream_t stream);\r
-    template void set_to_gpu<int   >(const DevMem2D& mat, const int* scalar, int channels, cudaStream_t stream);\r
-    template void set_to_gpu<float >(const DevMem2D& mat, const float* scalar, int channels, cudaStream_t stream);\r
-    template void set_to_gpu<double>(const DevMem2D& mat, const double* scalar, int channels, cudaStream_t stream);\r
+    template void set_to_gpu<uchar >(const DevMem2Db& mat, const uchar* scalar, int channels, cudaStream_t stream);\r
+    template void set_to_gpu<schar >(const DevMem2Db& mat, const schar* scalar, int channels, cudaStream_t stream);\r
+    template void set_to_gpu<ushort>(const DevMem2Db& mat, const ushort* scalar, int channels, cudaStream_t stream);\r
+    template void set_to_gpu<short >(const DevMem2Db& mat, const short* scalar, int channels, cudaStream_t stream);\r
+    template void set_to_gpu<int   >(const DevMem2Db& mat, const int* scalar, int channels, cudaStream_t stream);\r
+    template void set_to_gpu<float >(const DevMem2Db& mat, const float* scalar, int channels, cudaStream_t stream);\r
+    template void set_to_gpu<double>(const DevMem2Db& mat, const double* scalar, int channels, cudaStream_t stream);\r
 \r
 ///////////////////////////////////////////////////////////////////////////\r
 //////////////////////////////// ConvertTo ////////////////////////////////\r
@@ -297,7 +297,7 @@ namespace cv { namespace gpu { namespace device {
     };\r
         \r
     template<typename T, typename D>\r
-    void cvt_(const DevMem2D& src, const DevMem2D& dst, double alpha, double beta, cudaStream_t stream)\r
+    void cvt_(const DevMem2Db& src, const DevMem2Db& dst, double alpha, double beta, cudaStream_t stream)\r
     {\r
         cudaSafeCall( cudaSetDoubleForDevice(&alpha) );\r
         cudaSafeCall( cudaSetDoubleForDevice(&beta) );\r
@@ -305,10 +305,10 @@ namespace cv { namespace gpu { namespace device {
         transform((DevMem2D_<T>)src, (DevMem2D_<D>)dst, op, stream);\r
     }\r
 \r
-    void convert_gpu(const DevMem2D& src, int sdepth, const DevMem2D& dst, int ddepth, double alpha, double beta, \r
+    void convert_gpu(const DevMem2Db& src, int sdepth, const DevMem2Db& dst, int ddepth, double alpha, double beta, \r
         cudaStream_t stream = 0)\r
     {\r
-        typedef void (*caller_t)(const DevMem2D& src, const DevMem2D& dst, double alpha, double beta, \r
+        typedef void (*caller_t)(const DevMem2Db& src, const DevMem2Db& dst, double alpha, double beta, \r
             cudaStream_t stream);\r
 \r
         static const caller_t tab[8][8] =\r

diff --git a/modules/gpu/src/cuda/matrix_reductions.cu b/modules/gpu/src/cuda/matrix_reductions.cu
index e5c0c56..c0ca6c8 100644 (file)
--- a/modules/gpu/src/cuda/matrix_reductions.cu
+++ b/modules/gpu/src/cuda/matrix_reductions.cu
@@ -76,14 +76,14 @@ namespace cv { namespace gpu { namespace mathfunc
 \r
     struct Mask8U\r
     {\r
-        explicit Mask8U(PtrStep mask): mask(mask) {}\r
+        explicit Mask8U(PtrStepb mask): mask(mask) {}\r
 \r
         __device__ __forceinline__ bool operator()(int y, int x) const \r
         { \r
             return mask.ptr(y)[x]; \r
         }\r
 \r
-        PtrStep mask;\r
+        PtrStepb mask;\r
     };\r
 \r
 \r
@@ -180,7 +180,7 @@ namespace cv { namespace gpu { namespace mathfunc
 \r
 \r
     template <int nthreads, typename T, typename Mask>\r
-    __global__ void minMaxKernel(const DevMem2D src, Mask mask, T* minval, T* maxval)\r
+    __global__ void minMaxKernel(const DevMem2Db src, Mask mask, T* minval, T* maxval)\r
     {\r
         typedef typename MinMaxTypeTraits<T>::best_type best_type;\r
         __shared__ best_type sminval[nthreads];\r
@@ -263,7 +263,7 @@ namespace cv { namespace gpu { namespace mathfunc
 \r
    \r
     template <typename T>\r
-    void minMaxMaskCaller(const DevMem2D src, const PtrStep mask, double* minval, double* maxval, PtrStep buf)\r
+    void minMaxMaskCaller(const DevMem2Db src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf)\r
     {\r
         dim3 threads, grid;\r
         estimateThreadCfg(src.cols, src.rows, threads, grid);\r
@@ -284,17 +284,17 @@ namespace cv { namespace gpu { namespace mathfunc
         *maxval = maxval_;\r
     }  \r
 \r
-    template void minMaxMaskCaller<uchar>(const DevMem2D, const PtrStep, double*, double*, PtrStep);\r
-    template void minMaxMaskCaller<char>(const DevMem2D, const PtrStep, double*, double*, PtrStep);\r
-    template void minMaxMaskCaller<ushort>(const DevMem2D, const PtrStep, double*, double*, PtrStep);\r
-    template void minMaxMaskCaller<short>(const DevMem2D, const PtrStep, double*, double*, PtrStep);\r
-    template void minMaxMaskCaller<int>(const DevMem2D, const PtrStep, double*, double*, PtrStep);\r
-    template void minMaxMaskCaller<float>(const DevMem2D, const PtrStep, double*, double*, PtrStep);\r
-    template void minMaxMaskCaller<double>(const DevMem2D, const PtrStep, double*, double*, PtrStep);\r
+    template void minMaxMaskCaller<uchar>(const DevMem2Db, const PtrStepb, double*, double*, PtrStepb);\r
+    template void minMaxMaskCaller<char>(const DevMem2Db, const PtrStepb, double*, double*, PtrStepb);\r
+    template void minMaxMaskCaller<ushort>(const DevMem2Db, const PtrStepb, double*, double*, PtrStepb);\r
+    template void minMaxMaskCaller<short>(const DevMem2Db, const PtrStepb, double*, double*, PtrStepb);\r
+    template void minMaxMaskCaller<int>(const DevMem2Db, const PtrStepb, double*, double*, PtrStepb);\r
+    template void minMaxMaskCaller<float>(const DevMem2Db, const PtrStepb, double*, double*, PtrStepb);\r
+    template void minMaxMaskCaller<double>(const DevMem2Db, const PtrStepb, double*, double*, PtrStepb);\r
 \r
 \r
     template <typename T>\r
-    void minMaxCaller(const DevMem2D src, double* minval, double* maxval, PtrStep buf)\r
+    void minMaxCaller(const DevMem2Db src, double* minval, double* maxval, PtrStepb buf)\r
     {\r
         dim3 threads, grid;\r
         estimateThreadCfg(src.cols, src.rows, threads, grid);\r
@@ -315,13 +315,13 @@ namespace cv { namespace gpu { namespace mathfunc
         *maxval = maxval_;\r
     }  \r
 \r
-    template void minMaxCaller<uchar>(const DevMem2D, double*, double*, PtrStep);\r
-    template void minMaxCaller<char>(const DevMem2D, double*, double*, PtrStep);\r
-    template void minMaxCaller<ushort>(const DevMem2D, double*, double*, PtrStep);\r
-    template void minMaxCaller<short>(const DevMem2D, double*, double*, PtrStep);\r
-    template void minMaxCaller<int>(const DevMem2D, double*, double*, PtrStep);\r
-    template void minMaxCaller<float>(const DevMem2D, double*,double*, PtrStep);\r
-    template void minMaxCaller<double>(const DevMem2D, double*, double*, PtrStep);\r
+    template void minMaxCaller<uchar>(const DevMem2Db, double*, double*, PtrStepb);\r
+    template void minMaxCaller<char>(const DevMem2Db, double*, double*, PtrStepb);\r
+    template void minMaxCaller<ushort>(const DevMem2Db, double*, double*, PtrStepb);\r
+    template void minMaxCaller<short>(const DevMem2Db, double*, double*, PtrStepb);\r
+    template void minMaxCaller<int>(const DevMem2Db, double*, double*, PtrStepb);\r
+    template void minMaxCaller<float>(const DevMem2Db, double*,double*, PtrStepb);\r
+    template void minMaxCaller<double>(const DevMem2Db, double*, double*, PtrStepb);\r
 \r
 \r
     template <int nthreads, typename T>\r
@@ -349,7 +349,7 @@ namespace cv { namespace gpu { namespace mathfunc
 \r
 \r
     template <typename T>\r
-    void minMaxMaskMultipassCaller(const DevMem2D src, const PtrStep mask, double* minval, double* maxval, PtrStep buf)\r
+    void minMaxMaskMultipassCaller(const DevMem2Db src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf)\r
     {\r
         dim3 threads, grid;\r
         estimateThreadCfg(src.cols, src.rows, threads, grid);\r
@@ -372,16 +372,16 @@ namespace cv { namespace gpu { namespace mathfunc
         *maxval = maxval_;\r
     }\r
 \r
-    template void minMaxMaskMultipassCaller<uchar>(const DevMem2D, const PtrStep, double*, double*, PtrStep);\r
-    template void minMaxMaskMultipassCaller<char>(const DevMem2D, const PtrStep, double*, double*, PtrStep);\r
-    template void minMaxMaskMultipassCaller<ushort>(const DevMem2D, const PtrStep, double*, double*, PtrStep);\r
-    template void minMaxMaskMultipassCaller<short>(const DevMem2D, const PtrStep, double*, double*, PtrStep);\r
-    template void minMaxMaskMultipassCaller<int>(const DevMem2D, const PtrStep, double*, double*, PtrStep);\r
-    template void minMaxMaskMultipassCaller<float>(const DevMem2D, const PtrStep, double*, double*, PtrStep);\r
+    template void minMaxMaskMultipassCaller<uchar>(const DevMem2Db, const PtrStepb, double*, double*, PtrStepb);\r
+    template void minMaxMaskMultipassCaller<char>(const DevMem2Db, const PtrStepb, double*, double*, PtrStepb);\r
+    template void minMaxMaskMultipassCaller<ushort>(const DevMem2Db, const PtrStepb, double*, double*, PtrStepb);\r
+    template void minMaxMaskMultipassCaller<short>(const DevMem2Db, const PtrStepb, double*, double*, PtrStepb);\r
+    template void minMaxMaskMultipassCaller<int>(const DevMem2Db, const PtrStepb, double*, double*, PtrStepb);\r
+    template void minMaxMaskMultipassCaller<float>(const DevMem2Db, const PtrStepb, double*, double*, PtrStepb);\r
 \r
 \r
     template <typename T>\r
-    void minMaxMultipassCaller(const DevMem2D src, double* minval, double* maxval, PtrStep buf)\r
+    void minMaxMultipassCaller(const DevMem2Db src, double* minval, double* maxval, PtrStepb buf)\r
     {\r
         dim3 threads, grid;\r
         estimateThreadCfg(src.cols, src.rows, threads, grid);\r
@@ -404,12 +404,12 @@ namespace cv { namespace gpu { namespace mathfunc
         *maxval = maxval_;\r
     }\r
 \r
-    template void minMaxMultipassCaller<uchar>(const DevMem2D, double*, double*, PtrStep);\r
-    template void minMaxMultipassCaller<char>(const DevMem2D, double*, double*, PtrStep);\r
-    template void minMaxMultipassCaller<ushort>(const DevMem2D, double*, double*, PtrStep);\r
-    template void minMaxMultipassCaller<short>(const DevMem2D, double*, double*, PtrStep);\r
-    template void minMaxMultipassCaller<int>(const DevMem2D, double*, double*, PtrStep);\r
-    template void minMaxMultipassCaller<float>(const DevMem2D, double*, double*, PtrStep);\r
+    template void minMaxMultipassCaller<uchar>(const DevMem2Db, double*, double*, PtrStepb);\r
+    template void minMaxMultipassCaller<char>(const DevMem2Db, double*, double*, PtrStepb);\r
+    template void minMaxMultipassCaller<ushort>(const DevMem2Db, double*, double*, PtrStepb);\r
+    template void minMaxMultipassCaller<short>(const DevMem2Db, double*, double*, PtrStepb);\r
+    template void minMaxMultipassCaller<int>(const DevMem2Db, double*, double*, PtrStepb);\r
+    template void minMaxMultipassCaller<float>(const DevMem2Db, double*, double*, PtrStepb);\r
 \r
     } // namespace minmax\r
 \r
@@ -499,7 +499,7 @@ namespace cv { namespace gpu { namespace mathfunc
 \r
 \r
     template <int nthreads, typename T, typename Mask>\r
-    __global__ void minMaxLocKernel(const DevMem2D src, Mask mask, T* minval, T* maxval, \r
+    __global__ void minMaxLocKernel(const DevMem2Db src, Mask mask, T* minval, T* maxval, \r
                                     uint* minloc, uint* maxloc)\r
     {\r
         typedef typename MinMaxTypeTraits<T>::best_type best_type;\r
@@ -593,8 +593,8 @@ namespace cv { namespace gpu { namespace mathfunc
 \r
 \r
     template <typename T>\r
-    void minMaxLocMaskCaller(const DevMem2D src, const PtrStep mask, double* minval, double* maxval, \r
-                             int minloc[2], int maxloc[2], PtrStep valbuf, PtrStep locbuf)\r
+    void minMaxLocMaskCaller(const DevMem2Db src, const PtrStepb mask, double* minval, double* maxval, \r
+                             int minloc[2], int maxloc[2], PtrStepb valbuf, PtrStepb locbuf)\r
     {\r
         dim3 threads, grid;\r
         estimateThreadCfg(src.cols, src.rows, threads, grid);\r
@@ -624,18 +624,18 @@ namespace cv { namespace gpu { namespace mathfunc
         maxloc[1] = maxloc_ / src.cols; maxloc[0] = maxloc_ - maxloc[1] * src.cols;\r
     }\r
 \r
-    template void minMaxLocMaskCaller<uchar>(const DevMem2D, const PtrStep, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocMaskCaller<char>(const DevMem2D, const PtrStep, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocMaskCaller<ushort>(const DevMem2D, const PtrStep, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocMaskCaller<short>(const DevMem2D, const PtrStep, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocMaskCaller<int>(const DevMem2D, const PtrStep, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocMaskCaller<float>(const DevMem2D, const PtrStep, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocMaskCaller<double>(const DevMem2D, const PtrStep, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
+    template void minMaxLocMaskCaller<uchar>(const DevMem2Db, const PtrStepb, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocMaskCaller<char>(const DevMem2Db, const PtrStepb, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocMaskCaller<ushort>(const DevMem2Db, const PtrStepb, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocMaskCaller<short>(const DevMem2Db, const PtrStepb, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocMaskCaller<int>(const DevMem2Db, const PtrStepb, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocMaskCaller<float>(const DevMem2Db, const PtrStepb, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocMaskCaller<double>(const DevMem2Db, const PtrStepb, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
 \r
 \r
     template <typename T>\r
-    void minMaxLocCaller(const DevMem2D src, double* minval, double* maxval, \r
-                         int minloc[2], int maxloc[2], PtrStep valbuf, PtrStep locbuf)\r
+    void minMaxLocCaller(const DevMem2Db src, double* minval, double* maxval, \r
+                         int minloc[2], int maxloc[2], PtrStepb valbuf, PtrStepb locbuf)\r
     {\r
         dim3 threads, grid;\r
         estimateThreadCfg(src.cols, src.rows, threads, grid);\r
@@ -665,13 +665,13 @@ namespace cv { namespace gpu { namespace mathfunc
         maxloc[1] = maxloc_ / src.cols; maxloc[0] = maxloc_ - maxloc[1] * src.cols;\r
     }\r
 \r
-    template void minMaxLocCaller<uchar>(const DevMem2D, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocCaller<char>(const DevMem2D, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocCaller<ushort>(const DevMem2D, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocCaller<short>(const DevMem2D, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocCaller<int>(const DevMem2D, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocCaller<float>(const DevMem2D, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocCaller<double>(const DevMem2D, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
+    template void minMaxLocCaller<uchar>(const DevMem2Db, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocCaller<char>(const DevMem2Db, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocCaller<ushort>(const DevMem2Db, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocCaller<short>(const DevMem2Db, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocCaller<int>(const DevMem2Db, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocCaller<float>(const DevMem2Db, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocCaller<double>(const DevMem2Db, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
 \r
 \r
     // This kernel will be used only when compute capability is 1.0\r
@@ -706,8 +706,8 @@ namespace cv { namespace gpu { namespace mathfunc
 \r
 \r
     template <typename T>\r
-    void minMaxLocMaskMultipassCaller(const DevMem2D src, const PtrStep mask, double* minval, double* maxval, \r
-                                      int minloc[2], int maxloc[2], PtrStep valbuf, PtrStep locbuf)\r
+    void minMaxLocMaskMultipassCaller(const DevMem2Db src, const PtrStepb mask, double* minval, double* maxval, \r
+                                      int minloc[2], int maxloc[2], PtrStepb valbuf, PtrStepb locbuf)\r
     {\r
         dim3 threads, grid;\r
         estimateThreadCfg(src.cols, src.rows, threads, grid);\r
@@ -739,17 +739,17 @@ namespace cv { namespace gpu { namespace mathfunc
         maxloc[1] = maxloc_ / src.cols; maxloc[0] = maxloc_ - maxloc[1] * src.cols;\r
     }\r
 \r
-    template void minMaxLocMaskMultipassCaller<uchar>(const DevMem2D, const PtrStep, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocMaskMultipassCaller<char>(const DevMem2D, const PtrStep, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocMaskMultipassCaller<ushort>(const DevMem2D, const PtrStep, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocMaskMultipassCaller<short>(const DevMem2D, const PtrStep, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocMaskMultipassCaller<int>(const DevMem2D, const PtrStep, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocMaskMultipassCaller<float>(const DevMem2D, const PtrStep, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
+    template void minMaxLocMaskMultipassCaller<uchar>(const DevMem2Db, const PtrStepb, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocMaskMultipassCaller<char>(const DevMem2Db, const PtrStepb, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocMaskMultipassCaller<ushort>(const DevMem2Db, const PtrStepb, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocMaskMultipassCaller<short>(const DevMem2Db, const PtrStepb, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocMaskMultipassCaller<int>(const DevMem2Db, const PtrStepb, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocMaskMultipassCaller<float>(const DevMem2Db, const PtrStepb, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
 \r
 \r
     template <typename T>\r
-    void minMaxLocMultipassCaller(const DevMem2D src, double* minval, double* maxval, \r
-                                  int minloc[2], int maxloc[2], PtrStep valbuf, PtrStep locbuf)\r
+    void minMaxLocMultipassCaller(const DevMem2Db src, double* minval, double* maxval, \r
+                                  int minloc[2], int maxloc[2], PtrStepb valbuf, PtrStepb locbuf)\r
     {\r
         dim3 threads, grid;\r
         estimateThreadCfg(src.cols, src.rows, threads, grid);\r
@@ -781,12 +781,12 @@ namespace cv { namespace gpu { namespace mathfunc
         maxloc[1] = maxloc_ / src.cols; maxloc[0] = maxloc_ - maxloc[1] * src.cols;\r
     }\r
 \r
-    template void minMaxLocMultipassCaller<uchar>(const DevMem2D, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocMultipassCaller<char>(const DevMem2D, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocMultipassCaller<ushort>(const DevMem2D, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocMultipassCaller<short>(const DevMem2D, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocMultipassCaller<int>(const DevMem2D, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    template void minMaxLocMultipassCaller<float>(const DevMem2D, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
+    template void minMaxLocMultipassCaller<uchar>(const DevMem2Db, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocMultipassCaller<char>(const DevMem2Db, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocMultipassCaller<ushort>(const DevMem2Db, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocMultipassCaller<short>(const DevMem2Db, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocMultipassCaller<int>(const DevMem2Db, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    template void minMaxLocMultipassCaller<float>(const DevMem2Db, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
 \r
     } // namespace minmaxloc\r
 \r
@@ -829,7 +829,7 @@ namespace cv { namespace gpu { namespace mathfunc
 \r
 \r
     template <int nthreads, typename T>\r
-    __global__ void countNonZeroKernel(const DevMem2D src, volatile uint* count)\r
+    __global__ void countNonZeroKernel(const DevMem2Db src, volatile uint* count)\r
     {\r
         __shared__ uint scount[nthreads];\r
 \r
@@ -884,7 +884,7 @@ namespace cv { namespace gpu { namespace mathfunc
 \r
    \r
     template <typename T>\r
-    int countNonZeroCaller(const DevMem2D src, PtrStep buf)\r
+    int countNonZeroCaller(const DevMem2Db src, PtrStepb buf)\r
     {\r
         dim3 threads, grid;\r
         estimateThreadCfg(src.cols, src.rows, threads, grid);\r
@@ -903,13 +903,13 @@ namespace cv { namespace gpu { namespace mathfunc
         return count;\r
     }  \r
 \r
-    template int countNonZeroCaller<uchar>(const DevMem2D, PtrStep);\r
-    template int countNonZeroCaller<char>(const DevMem2D, PtrStep);\r
-    template int countNonZeroCaller<ushort>(const DevMem2D, PtrStep);\r
-    template int countNonZeroCaller<short>(const DevMem2D, PtrStep);\r
-    template int countNonZeroCaller<int>(const DevMem2D, PtrStep);\r
-    template int countNonZeroCaller<float>(const DevMem2D, PtrStep);\r
-    template int countNonZeroCaller<double>(const DevMem2D, PtrStep);\r
+    template int countNonZeroCaller<uchar>(const DevMem2Db, PtrStepb);\r
+    template int countNonZeroCaller<char>(const DevMem2Db, PtrStepb);\r
+    template int countNonZeroCaller<ushort>(const DevMem2Db, PtrStepb);\r
+    template int countNonZeroCaller<short>(const DevMem2Db, PtrStepb);\r
+    template int countNonZeroCaller<int>(const DevMem2Db, PtrStepb);\r
+    template int countNonZeroCaller<float>(const DevMem2Db, PtrStepb);\r
+    template int countNonZeroCaller<double>(const DevMem2Db, PtrStepb);\r
 \r
 \r
     template <int nthreads, typename T>\r
@@ -929,7 +929,7 @@ namespace cv { namespace gpu { namespace mathfunc
 \r
 \r
     template <typename T>\r
-    int countNonZeroMultipassCaller(const DevMem2D src, PtrStep buf)\r
+    int countNonZeroMultipassCaller(const DevMem2Db src, PtrStepb buf)\r
     {\r
         dim3 threads, grid;\r
         estimateThreadCfg(src.cols, src.rows, threads, grid);\r
@@ -950,12 +950,12 @@ namespace cv { namespace gpu { namespace mathfunc
         return count;\r
     }  \r
 \r
-    template int countNonZeroMultipassCaller<uchar>(const DevMem2D, PtrStep);\r
-    template int countNonZeroMultipassCaller<char>(const DevMem2D, PtrStep);\r
-    template int countNonZeroMultipassCaller<ushort>(const DevMem2D, PtrStep);\r
-    template int countNonZeroMultipassCaller<short>(const DevMem2D, PtrStep);\r
-    template int countNonZeroMultipassCaller<int>(const DevMem2D, PtrStep);\r
-    template int countNonZeroMultipassCaller<float>(const DevMem2D, PtrStep);\r
+    template int countNonZeroMultipassCaller<uchar>(const DevMem2Db, PtrStepb);\r
+    template int countNonZeroMultipassCaller<char>(const DevMem2Db, PtrStepb);\r
+    template int countNonZeroMultipassCaller<ushort>(const DevMem2Db, PtrStepb);\r
+    template int countNonZeroMultipassCaller<short>(const DevMem2Db, PtrStepb);\r
+    template int countNonZeroMultipassCaller<int>(const DevMem2Db, PtrStepb);\r
+    template int countNonZeroMultipassCaller<float>(const DevMem2Db, PtrStepb);\r
 \r
     } // namespace countnonzero\r
 \r
@@ -1022,7 +1022,7 @@ namespace cv { namespace gpu { namespace mathfunc
     }\r
 \r
     template <typename T, typename R, typename Op, int nthreads>\r
-    __global__ void sumKernel(const DevMem2D src, R* result)\r
+    __global__ void sumKernel(const DevMem2Db src, R* result)\r
     {\r
         __shared__ R smem[nthreads];\r
 \r
@@ -1094,7 +1094,7 @@ namespace cv { namespace gpu { namespace mathfunc
 \r
 \r
     template <typename T, typename R, typename Op, int nthreads>\r
-    __global__ void sumKernel_C2(const DevMem2D src, typename TypeVec<R, 2>::vec_type* result)\r
+    __global__ void sumKernel_C2(const DevMem2Db src, typename TypeVec<R, 2>::vec_type* result)\r
     {\r
         typedef typename TypeVec<T, 2>::vec_type SrcType;\r
         typedef typename TypeVec<R, 2>::vec_type DstType;\r
@@ -1199,7 +1199,7 @@ namespace cv { namespace gpu { namespace mathfunc
 \r
 \r
     template <typename T, typename R, typename Op, int nthreads>\r
-    __global__ void sumKernel_C3(const DevMem2D src, typename TypeVec<R, 3>::vec_type* result)\r
+    __global__ void sumKernel_C3(const DevMem2Db src, typename TypeVec<R, 3>::vec_type* result)\r
     {\r
         typedef typename TypeVec<T, 3>::vec_type SrcType;\r
         typedef typename TypeVec<R, 3>::vec_type DstType;\r
@@ -1313,7 +1313,7 @@ namespace cv { namespace gpu { namespace mathfunc
     }\r
 \r
     template <typename T, typename R, typename Op, int nthreads>\r
-    __global__ void sumKernel_C4(const DevMem2D src, typename TypeVec<R, 4>::vec_type* result)\r
+    __global__ void sumKernel_C4(const DevMem2Db src, typename TypeVec<R, 4>::vec_type* result)\r
     {\r
         typedef typename TypeVec<T, 4>::vec_type SrcType;\r
         typedef typename TypeVec<R, 4>::vec_type DstType;\r
@@ -1441,7 +1441,7 @@ namespace cv { namespace gpu { namespace mathfunc
 \r
 \r
     template <typename T>\r
-    void sumMultipassCaller(const DevMem2D src, PtrStep buf, double* sum, int cn)\r
+    void sumMultipassCaller(const DevMem2Db src, PtrStepb buf, double* sum, int cn)\r
     {\r
         using namespace sums;\r
         typedef typename SumType<T>::R R;\r
@@ -1504,16 +1504,16 @@ namespace cv { namespace gpu { namespace mathfunc
         sum[3] = result[3];\r
     }  \r
 \r
-    template void sumMultipassCaller<uchar>(const DevMem2D, PtrStep, double*, int);\r
-    template void sumMultipassCaller<char>(const DevMem2D, PtrStep, double*, int);\r
-    template void sumMultipassCaller<ushort>(const DevMem2D, PtrStep, double*, int);\r
-    template void sumMultipassCaller<short>(const DevMem2D, PtrStep, double*, int);\r
-    template void sumMultipassCaller<int>(const DevMem2D, PtrStep, double*, int);\r
-    template void sumMultipassCaller<float>(const DevMem2D, PtrStep, double*, int);\r
+    template void sumMultipassCaller<uchar>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sumMultipassCaller<char>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sumMultipassCaller<ushort>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sumMultipassCaller<short>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sumMultipassCaller<int>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sumMultipassCaller<float>(const DevMem2Db, PtrStepb, double*, int);\r
 \r
 \r
     template <typename T>\r
-    void sumCaller(const DevMem2D src, PtrStep buf, double* sum, int cn)\r
+    void sumCaller(const DevMem2Db src, PtrStepb buf, double* sum, int cn)\r
     {\r
         using namespace sums;\r
         typedef typename SumType<T>::R R;\r
@@ -1554,16 +1554,16 @@ namespace cv { namespace gpu { namespace mathfunc
         sum[3] = result[3];\r
     }  \r
 \r
-    template void sumCaller<uchar>(const DevMem2D, PtrStep, double*, int);\r
-    template void sumCaller<char>(const DevMem2D, PtrStep, double*, int);\r
-    template void sumCaller<ushort>(const DevMem2D, PtrStep, double*, int);\r
-    template void sumCaller<short>(const DevMem2D, PtrStep, double*, int);\r
-    template void sumCaller<int>(const DevMem2D, PtrStep, double*, int);\r
-    template void sumCaller<float>(const DevMem2D, PtrStep, double*, int);\r
+    template void sumCaller<uchar>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sumCaller<char>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sumCaller<ushort>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sumCaller<short>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sumCaller<int>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sumCaller<float>(const DevMem2Db, PtrStepb, double*, int);\r
 \r
 \r
     template <typename T>\r
-    void absSumMultipassCaller(const DevMem2D src, PtrStep buf, double* sum, int cn)\r
+    void absSumMultipassCaller(const DevMem2Db src, PtrStepb buf, double* sum, int cn)\r
     {\r
         using namespace sums;\r
         typedef typename SumType<T>::R R;\r
@@ -1626,16 +1626,16 @@ namespace cv { namespace gpu { namespace mathfunc
         sum[3] = result[3];\r
     }  \r
 \r
-    template void absSumMultipassCaller<uchar>(const DevMem2D, PtrStep, double*, int);\r
-    template void absSumMultipassCaller<char>(const DevMem2D, PtrStep, double*, int);\r
-    template void absSumMultipassCaller<ushort>(const DevMem2D, PtrStep, double*, int);\r
-    template void absSumMultipassCaller<short>(const DevMem2D, PtrStep, double*, int);\r
-    template void absSumMultipassCaller<int>(const DevMem2D, PtrStep, double*, int);\r
-    template void absSumMultipassCaller<float>(const DevMem2D, PtrStep, double*, int);\r
+    template void absSumMultipassCaller<uchar>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void absSumMultipassCaller<char>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void absSumMultipassCaller<ushort>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void absSumMultipassCaller<short>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void absSumMultipassCaller<int>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void absSumMultipassCaller<float>(const DevMem2Db, PtrStepb, double*, int);\r
 \r
 \r
     template <typename T>\r
-    void absSumCaller(const DevMem2D src, PtrStep buf, double* sum, int cn)\r
+    void absSumCaller(const DevMem2Db src, PtrStepb buf, double* sum, int cn)\r
     {\r
         using namespace sums;\r
         typedef typename SumType<T>::R R;\r
@@ -1676,16 +1676,16 @@ namespace cv { namespace gpu { namespace mathfunc
         sum[3] = result[3];\r
     }\r
 \r
-    template void absSumCaller<uchar>(const DevMem2D, PtrStep, double*, int);\r
-    template void absSumCaller<char>(const DevMem2D, PtrStep, double*, int);\r
-    template void absSumCaller<ushort>(const DevMem2D, PtrStep, double*, int);\r
-    template void absSumCaller<short>(const DevMem2D, PtrStep, double*, int);\r
-    template void absSumCaller<int>(const DevMem2D, PtrStep, double*, int);\r
-    template void absSumCaller<float>(const DevMem2D, PtrStep, double*, int);\r
+    template void absSumCaller<uchar>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void absSumCaller<char>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void absSumCaller<ushort>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void absSumCaller<short>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void absSumCaller<int>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void absSumCaller<float>(const DevMem2Db, PtrStepb, double*, int);\r
 \r
 \r
     template <typename T>\r
-    void sqrSumMultipassCaller(const DevMem2D src, PtrStep buf, double* sum, int cn)\r
+    void sqrSumMultipassCaller(const DevMem2Db src, PtrStepb buf, double* sum, int cn)\r
     {\r
         using namespace sums;\r
         typedef typename SumType<T>::R R;\r
@@ -1748,16 +1748,16 @@ namespace cv { namespace gpu { namespace mathfunc
         sum[3] = result[3];\r
     }  \r
 \r
-    template void sqrSumMultipassCaller<uchar>(const DevMem2D, PtrStep, double*, int);\r
-    template void sqrSumMultipassCaller<char>(const DevMem2D, PtrStep, double*, int);\r
-    template void sqrSumMultipassCaller<ushort>(const DevMem2D, PtrStep, double*, int);\r
-    template void sqrSumMultipassCaller<short>(const DevMem2D, PtrStep, double*, int);\r
-    template void sqrSumMultipassCaller<int>(const DevMem2D, PtrStep, double*, int);\r
-    template void sqrSumMultipassCaller<float>(const DevMem2D, PtrStep, double*, int);\r
+    template void sqrSumMultipassCaller<uchar>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sqrSumMultipassCaller<char>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sqrSumMultipassCaller<ushort>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sqrSumMultipassCaller<short>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sqrSumMultipassCaller<int>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sqrSumMultipassCaller<float>(const DevMem2Db, PtrStepb, double*, int);\r
 \r
 \r
     template <typename T>\r
-    void sqrSumCaller(const DevMem2D src, PtrStep buf, double* sum, int cn)\r
+    void sqrSumCaller(const DevMem2Db src, PtrStepb buf, double* sum, int cn)\r
     {\r
         using namespace sums;\r
         typedef typename SumType<T>::R R;\r
@@ -1798,12 +1798,12 @@ namespace cv { namespace gpu { namespace mathfunc
         sum[3] = result[3];\r
     }\r
 \r
-    template void sqrSumCaller<uchar>(const DevMem2D, PtrStep, double*, int);\r
-    template void sqrSumCaller<char>(const DevMem2D, PtrStep, double*, int);\r
-    template void sqrSumCaller<ushort>(const DevMem2D, PtrStep, double*, int);\r
-    template void sqrSumCaller<short>(const DevMem2D, PtrStep, double*, int);\r
-    template void sqrSumCaller<int>(const DevMem2D, PtrStep, double*, int);\r
-    template void sqrSumCaller<float>(const DevMem2D, PtrStep, double*, int);\r
+    template void sqrSumCaller<uchar>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sqrSumCaller<char>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sqrSumCaller<ushort>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sqrSumCaller<short>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sqrSumCaller<int>(const DevMem2Db, PtrStepb, double*, int);\r
+    template void sqrSumCaller<float>(const DevMem2Db, PtrStepb, double*, int);\r
 \r
     //////////////////////////////////////////////////////////////////////////////\r
     // reduce\r
@@ -1933,7 +1933,7 @@ namespace cv { namespace gpu { namespace mathfunc
 \r
     }\r
 \r
-    template <typename T, typename S, typename D> void reduceRows_gpu(const DevMem2D& src, const DevMem2D& dst, int reduceOp, cudaStream_t stream)\r
+    template <typename T, typename S, typename D> void reduceRows_gpu(const DevMem2Db& src, const DevMem2Db& dst, int reduceOp, cudaStream_t stream)\r
     {\r
         typedef void (*caller_t)(const DevMem2D_<T>& src, DevMem2D_<D> dst, cudaStream_t stream);\r
 \r
@@ -1948,22 +1948,22 @@ namespace cv { namespace gpu { namespace mathfunc
         callers[reduceOp](static_cast< DevMem2D_<T> >(src), static_cast< DevMem2D_<D> >(dst), stream);\r
     }\r
 \r
-    template void reduceRows_gpu<uchar, int, uchar>(const DevMem2D& src, const DevMem2D& dst, int reduceOp, cudaStream_t stream);\r
-    template void reduceRows_gpu<uchar, int, int>(const DevMem2D& src, const DevMem2D& dst, int reduceOp, cudaStream_t stream);\r
-    template void reduceRows_gpu<uchar, int, float>(const DevMem2D& src, const DevMem2D& dst, int reduceOp, cudaStream_t stream);  \r
+    template void reduceRows_gpu<uchar, int, uchar>(const DevMem2Db& src, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);\r
+    template void reduceRows_gpu<uchar, int, int>(const DevMem2Db& src, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);\r
+    template void reduceRows_gpu<uchar, int, float>(const DevMem2Db& src, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);  \r
 \r
-    template void reduceRows_gpu<ushort, int, ushort>(const DevMem2D& src, const DevMem2D& dst, int reduceOp, cudaStream_t stream);\r
-    template void reduceRows_gpu<ushort, int, int>(const DevMem2D& src, const DevMem2D& dst, int reduceOp, cudaStream_t stream);\r
-    template void reduceRows_gpu<ushort, int, float>(const DevMem2D& src, const DevMem2D& dst, int reduceOp, cudaStream_t stream); \r
+    template void reduceRows_gpu<ushort, int, ushort>(const DevMem2Db& src, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);\r
+    template void reduceRows_gpu<ushort, int, int>(const DevMem2Db& src, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);\r
+    template void reduceRows_gpu<ushort, int, float>(const DevMem2Db& src, const DevMem2Db& dst, int reduceOp, cudaStream_t stream); \r
 \r
-    template void reduceRows_gpu<short, int, short>(const DevMem2D& src, const DevMem2D& dst, int reduceOp, cudaStream_t stream);\r
-    template void reduceRows_gpu<short, int, int>(const DevMem2D& src, const DevMem2D& dst, int reduceOp, cudaStream_t stream);\r
-    template void reduceRows_gpu<short, int, float>(const DevMem2D& src, const DevMem2D& dst, int reduceOp, cudaStream_t stream); \r
+    template void reduceRows_gpu<short, int, short>(const DevMem2Db& src, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);\r
+    template void reduceRows_gpu<short, int, int>(const DevMem2Db& src, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);\r
+    template void reduceRows_gpu<short, int, float>(const DevMem2Db& src, const DevMem2Db& dst, int reduceOp, cudaStream_t stream); \r
 \r
-    template void reduceRows_gpu<int, int, int>(const DevMem2D& src, const DevMem2D& dst, int reduceOp, cudaStream_t stream);\r
-    template void reduceRows_gpu<int, int, float>(const DevMem2D& src, const DevMem2D& dst, int reduceOp, cudaStream_t stream);\r
+    template void reduceRows_gpu<int, int, int>(const DevMem2Db& src, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);\r
+    template void reduceRows_gpu<int, int, float>(const DevMem2Db& src, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);\r
 \r
-    template void reduceRows_gpu<float, float, float>(const DevMem2D& src, const DevMem2D& dst, int reduceOp, cudaStream_t stream);\r
+    template void reduceRows_gpu<float, float, float>(const DevMem2Db& src, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);\r
 \r
 \r
 \r
@@ -2070,7 +2070,7 @@ namespace cv { namespace gpu { namespace mathfunc
 \r
     }\r
 \r
-    template <typename T, typename S, typename D> void reduceCols_gpu(const DevMem2D& src, int cn, const DevMem2D& dst, int reduceOp, cudaStream_t stream)\r
+    template <typename T, typename S, typename D> void reduceCols_gpu(const DevMem2Db& src, int cn, const DevMem2Db& dst, int reduceOp, cudaStream_t stream)\r
     {\r
         typedef void (*caller_t)(const DevMem2D_<T>& src, DevMem2D_<D> dst, cudaStream_t stream);\r
 \r
@@ -2085,20 +2085,20 @@ namespace cv { namespace gpu { namespace mathfunc
         callers[cn - 1][reduceOp](static_cast< DevMem2D_<T> >(src), static_cast< DevMem2D_<D> >(dst), stream);\r
     }\r
 \r
-    template void reduceCols_gpu<uchar, int, uchar>(const DevMem2D& src, int cn, const DevMem2D& dst, int reduceOp, cudaStream_t stream);\r
-    template void reduceCols_gpu<uchar, int, int>(const DevMem2D& src, int cn, const DevMem2D& dst, int reduceOp, cudaStream_t stream);\r
-    template void reduceCols_gpu<uchar, int, float>(const DevMem2D& src, int cn, const DevMem2D& dst, int reduceOp, cudaStream_t stream);\r
+    template void reduceCols_gpu<uchar, int, uchar>(const DevMem2Db& src, int cn, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);\r
+    template void reduceCols_gpu<uchar, int, int>(const DevMem2Db& src, int cn, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);\r
+    template void reduceCols_gpu<uchar, int, float>(const DevMem2Db& src, int cn, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);\r
 \r
-    template void reduceCols_gpu<ushort, int, ushort>(const DevMem2D& src, int cn, const DevMem2D& dst, int reduceOp, cudaStream_t stream); \r
-    template void reduceCols_gpu<ushort, int, int>(const DevMem2D& src, int cn, const DevMem2D& dst, int reduceOp, cudaStream_t stream);                  \r
-    template void reduceCols_gpu<ushort, int, float>(const DevMem2D& src, int cn, const DevMem2D& dst, int reduceOp, cudaStream_t stream);\r
+    template void reduceCols_gpu<ushort, int, ushort>(const DevMem2Db& src, int cn, const DevMem2Db& dst, int reduceOp, cudaStream_t stream); \r
+    template void reduceCols_gpu<ushort, int, int>(const DevMem2Db& src, int cn, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);                  \r
+    template void reduceCols_gpu<ushort, int, float>(const DevMem2Db& src, int cn, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);\r
 \r
-    template void reduceCols_gpu<short, int, short>(const DevMem2D& src, int cn, const DevMem2D& dst, int reduceOp, cudaStream_t stream);  \r
-    template void reduceCols_gpu<short, int, int>(const DevMem2D& src, int cn, const DevMem2D& dst, int reduceOp, cudaStream_t stream);                  \r
-    template void reduceCols_gpu<short, int, float>(const DevMem2D& src, int cn, const DevMem2D& dst, int reduceOp, cudaStream_t stream);  \r
+    template void reduceCols_gpu<short, int, short>(const DevMem2Db& src, int cn, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);  \r
+    template void reduceCols_gpu<short, int, int>(const DevMem2Db& src, int cn, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);                  \r
+    template void reduceCols_gpu<short, int, float>(const DevMem2Db& src, int cn, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);  \r
 \r
-    template void reduceCols_gpu<int, int, int>(const DevMem2D& src, int cn, const DevMem2D& dst, int reduceOp, cudaStream_t stream);                  \r
-    template void reduceCols_gpu<int, int, float>(const DevMem2D& src, int cn, const DevMem2D& dst, int reduceOp, cudaStream_t stream);\r
+    template void reduceCols_gpu<int, int, int>(const DevMem2Db& src, int cn, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);                  \r
+    template void reduceCols_gpu<int, int, float>(const DevMem2Db& src, int cn, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);\r
 \r
-    template void reduceCols_gpu<float, float, float>(const DevMem2D& src, int cn, const DevMem2D& dst, int reduceOp, cudaStream_t stream);\r
+    template void reduceCols_gpu<float, float, float>(const DevMem2Db& src, int cn, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);\r
  }}}\r

diff --git a/modules/gpu/src/cuda/pyr_down.cu b/modules/gpu/src/cuda/pyr_down.cu
index d701bec..385df7a 100644 (file)
--- a/modules/gpu/src/cuda/pyr_down.cu
+++ b/modules/gpu/src/cuda/pyr_down.cu
@@ -51,7 +51,7 @@ using namespace cv::gpu::device;
 \r
 namespace cv { namespace gpu { namespace imgproc\r
 {\r
-    template <typename T, typename B> __global__ void pyrDown(const PtrStep_<T> src, PtrStep_<T> dst, const B b, int dst_cols)\r
+    template <typename T, typename B> __global__ void pyrDown(const PtrStep<T> src, PtrStep<T> dst, const B b, int dst_cols)\r
     {\r
         typedef typename TypeVec<float, VecTraits<T>::cn>::vec_type value_type;\r
 \r
@@ -139,7 +139,7 @@ namespace cv { namespace gpu { namespace imgproc
             cudaSafeCall( cudaDeviceSynchronize() );\r
     }\r
 \r
-    template <typename T, int cn> void pyrDown_gpu(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream)\r
+    template <typename T, int cn> void pyrDown_gpu(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream)\r
     {\r
         typedef typename TypeVec<T, cn>::vec_type type;\r
 \r
@@ -153,33 +153,33 @@ namespace cv { namespace gpu { namespace imgproc
         callers[borderType](static_cast< DevMem2D_<type> >(src), static_cast< DevMem2D_<type> >(dst), stream);\r
     }\r
 \r
-    template void pyrDown_gpu<uchar, 1>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrDown_gpu<uchar, 2>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrDown_gpu<uchar, 3>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrDown_gpu<uchar, 4>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-\r
-    template void pyrDown_gpu<schar, 1>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrDown_gpu<schar, 2>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrDown_gpu<schar, 3>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrDown_gpu<schar, 4>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-\r
-    template void pyrDown_gpu<ushort, 1>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrDown_gpu<ushort, 2>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrDown_gpu<ushort, 3>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrDown_gpu<ushort, 4>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-\r
-    template void pyrDown_gpu<short, 1>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrDown_gpu<short, 2>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrDown_gpu<short, 3>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrDown_gpu<short, 4>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-\r
-    template void pyrDown_gpu<int, 1>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrDown_gpu<int, 2>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrDown_gpu<int, 3>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrDown_gpu<int, 4>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-\r
-    template void pyrDown_gpu<float, 1>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrDown_gpu<float, 2>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrDown_gpu<float, 3>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrDown_gpu<float, 4>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
+    template void pyrDown_gpu<uchar, 1>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrDown_gpu<uchar, 2>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrDown_gpu<uchar, 3>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrDown_gpu<uchar, 4>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+\r
+    template void pyrDown_gpu<schar, 1>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrDown_gpu<schar, 2>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrDown_gpu<schar, 3>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrDown_gpu<schar, 4>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+\r
+    template void pyrDown_gpu<ushort, 1>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrDown_gpu<ushort, 2>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrDown_gpu<ushort, 3>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrDown_gpu<ushort, 4>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+\r
+    template void pyrDown_gpu<short, 1>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrDown_gpu<short, 2>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrDown_gpu<short, 3>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrDown_gpu<short, 4>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+\r
+    template void pyrDown_gpu<int, 1>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrDown_gpu<int, 2>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrDown_gpu<int, 3>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrDown_gpu<int, 4>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+\r
+    template void pyrDown_gpu<float, 1>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrDown_gpu<float, 2>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrDown_gpu<float, 3>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrDown_gpu<float, 4>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
 }}}\r

diff --git a/modules/gpu/src/cuda/pyr_up.cu b/modules/gpu/src/cuda/pyr_up.cu
index 42d30c6..3555b89 100644 (file)
--- a/modules/gpu/src/cuda/pyr_up.cu
+++ b/modules/gpu/src/cuda/pyr_up.cu
@@ -51,7 +51,7 @@ using namespace cv::gpu::device;
 \r
 namespace cv { namespace gpu { namespace imgproc\r
 {\r
-    template <typename T, typename B> __global__ void pyrUp(const PtrStep_<T> src, DevMem2D_<T> dst, const B b)\r
+    template <typename T, typename B> __global__ void pyrUp(const PtrStep<T> src, DevMem2D_<T> dst, const B b)\r
     {\r
         typedef typename TypeVec<float, VecTraits<T>::cn>::vec_type value_type;\r
 \r
@@ -134,7 +134,7 @@ namespace cv { namespace gpu { namespace imgproc
             cudaSafeCall( cudaDeviceSynchronize() );\r
     }\r
 \r
-    template <typename T, int cn> void pyrUp_gpu(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream)\r
+    template <typename T, int cn> void pyrUp_gpu(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream)\r
     {\r
         typedef typename TypeVec<T, cn>::vec_type type;\r
 \r
@@ -148,33 +148,33 @@ namespace cv { namespace gpu { namespace imgproc
         callers[borderType](static_cast< DevMem2D_<type> >(src), static_cast< DevMem2D_<type> >(dst), stream);\r
     }\r
 \r
-    template void pyrUp_gpu<uchar, 1>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrUp_gpu<uchar, 2>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrUp_gpu<uchar, 3>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrUp_gpu<uchar, 4>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-\r
-    template void pyrUp_gpu<schar, 1>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrUp_gpu<schar, 2>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrUp_gpu<schar, 3>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrUp_gpu<schar, 4>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-\r
-    template void pyrUp_gpu<ushort, 1>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrUp_gpu<ushort, 2>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrUp_gpu<ushort, 3>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrUp_gpu<ushort, 4>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-\r
-    template void pyrUp_gpu<short, 1>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrUp_gpu<short, 2>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrUp_gpu<short, 3>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrUp_gpu<short, 4>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-\r
-    template void pyrUp_gpu<int, 1>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrUp_gpu<int, 2>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrUp_gpu<int, 3>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrUp_gpu<int, 4>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-\r
-    template void pyrUp_gpu<float, 1>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrUp_gpu<float, 2>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrUp_gpu<float, 3>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
-    template void pyrUp_gpu<float, 4>(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
+    template void pyrUp_gpu<uchar, 1>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrUp_gpu<uchar, 2>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrUp_gpu<uchar, 3>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrUp_gpu<uchar, 4>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+\r
+    template void pyrUp_gpu<schar, 1>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrUp_gpu<schar, 2>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrUp_gpu<schar, 3>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrUp_gpu<schar, 4>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+\r
+    template void pyrUp_gpu<ushort, 1>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrUp_gpu<ushort, 2>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrUp_gpu<ushort, 3>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrUp_gpu<ushort, 4>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+\r
+    template void pyrUp_gpu<short, 1>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrUp_gpu<short, 2>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrUp_gpu<short, 3>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrUp_gpu<short, 4>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+\r
+    template void pyrUp_gpu<int, 1>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrUp_gpu<int, 2>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrUp_gpu<int, 3>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrUp_gpu<int, 4>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+\r
+    template void pyrUp_gpu<float, 1>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrUp_gpu<float, 2>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrUp_gpu<float, 3>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
+    template void pyrUp_gpu<float, 4>(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
 }}}\r

diff --git a/modules/gpu/src/cuda/remap.cu b/modules/gpu/src/cuda/remap.cu
index fb342a8..1ab8447 100644 (file)
--- a/modules/gpu/src/cuda/remap.cu
+++ b/modules/gpu/src/cuda/remap.cu
@@ -77,8 +77,8 @@ namespace cv { namespace gpu { namespace imgproc
             dim3 grid(divUp(dst.cols, block.x), divUp(dst.rows, block.y));\r
 \r
             B<work_type> brd(src.rows, src.cols, VecTraits<work_type>::make(borderValue));\r
-            BorderReader< PtrStep_<T>, B<work_type> > brdSrc(src, brd);\r
-            Filter< BorderReader< PtrStep_<T>, B<work_type> > > filter_src(brdSrc);\r
+            BorderReader< PtrStep<T>, B<work_type> > brdSrc(src, brd);\r
+            Filter< BorderReader< PtrStep<T>, B<work_type> > > filter_src(brdSrc);\r
 \r
             remap<<<grid, block, 0, stream>>>(filter_src, mapx, mapy, dst);\r
             cudaSafeCall( cudaGetLastError() );\r
@@ -95,8 +95,8 @@ namespace cv { namespace gpu { namespace imgproc
             dim3 grid(divUp(dst.cols, block.x), divUp(dst.rows, block.y));\r
 \r
             B<work_type> brd(src.rows, src.cols, VecTraits<work_type>::make(borderValue));\r
-            BorderReader< PtrStep_<T>, B<work_type> > brdSrc(src, brd);\r
-            Filter< BorderReader< PtrStep_<T>, B<work_type> > > filter_src(brdSrc);\r
+            BorderReader< PtrStep<T>, B<work_type> > brdSrc(src, brd);\r
+            Filter< BorderReader< PtrStep<T>, B<work_type> > > filter_src(brdSrc);\r
 \r
             remap<<<grid, block>>>(filter_src, mapx, mapy, dst);\r
             cudaSafeCall( cudaGetLastError() );\r
@@ -185,7 +185,7 @@ namespace cv { namespace gpu { namespace imgproc
         }\r
     };\r
 \r
-    template <typename T> void remap_gpu(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream)\r
+    template <typename T> void remap_gpu(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream)\r
     {\r
         typedef void (*caller_t)(const DevMem2D_<T>& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D_<T>& dst, const float* borderValue, cudaStream_t stream);\r
 \r
@@ -217,33 +217,33 @@ namespace cv { namespace gpu { namespace imgproc
         callers[interpolation][borderMode](static_cast< DevMem2D_<T> >(src), xmap, ymap, static_cast< DevMem2D_<T> >(dst), borderValue, stream);\r
     }\r
 \r
-    template void remap_gpu<uchar >(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    //template void remap_gpu<uchar2>(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    template void remap_gpu<uchar3>(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    template void remap_gpu<uchar4>(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    template void remap_gpu<uchar >(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    //template void remap_gpu<uchar2>(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    template void remap_gpu<uchar3>(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    template void remap_gpu<uchar4>(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
     \r
-    //template void remap_gpu<schar>(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    //template void remap_gpu<char2>(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    //template void remap_gpu<char3>(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    //template void remap_gpu<char4>(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    //template void remap_gpu<schar>(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    //template void remap_gpu<char2>(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    //template void remap_gpu<char3>(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    //template void remap_gpu<char4>(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
     \r
-    template void remap_gpu<ushort >(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    //template void remap_gpu<ushort2>(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    template void remap_gpu<ushort3>(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    template void remap_gpu<ushort4>(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    template void remap_gpu<ushort >(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    //template void remap_gpu<ushort2>(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    template void remap_gpu<ushort3>(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    template void remap_gpu<ushort4>(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
     \r
-    template void remap_gpu<short >(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    //template void remap_gpu<short2>(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    template void remap_gpu<short3>(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    template void remap_gpu<short4>(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    template void remap_gpu<short >(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    //template void remap_gpu<short2>(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    template void remap_gpu<short3>(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    template void remap_gpu<short4>(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
     \r
-    //template void remap_gpu<int >(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    //template void remap_gpu<int2>(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    //template void remap_gpu<int3>(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    //template void remap_gpu<int4>(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    //template void remap_gpu<int >(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    //template void remap_gpu<int2>(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    //template void remap_gpu<int3>(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    //template void remap_gpu<int4>(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
     \r
-    template void remap_gpu<float >(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    //template void remap_gpu<float2>(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    template void remap_gpu<float3>(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
-    template void remap_gpu<float4>(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    template void remap_gpu<float >(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    //template void remap_gpu<float2>(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    template void remap_gpu<float3>(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    template void remap_gpu<float4>(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
 }}}\r

diff --git a/modules/gpu/src/cuda/resize.cu b/modules/gpu/src/cuda/resize.cu
index 60f52c9..78871f7 100644 (file)
--- a/modules/gpu/src/cuda/resize.cu
+++ b/modules/gpu/src/cuda/resize.cu
@@ -88,8 +88,8 @@ namespace cv { namespace gpu { namespace imgproc
             dim3 grid(divUp(dst.cols, block.x), divUp(dst.rows, block.y));\r
 \r
             BrdReplicate<T> brd(src.rows, src.cols);\r
-            BorderReader< PtrStep_<T>, BrdReplicate<T> > brdSrc(src, brd);\r
-            Filter< BorderReader< PtrStep_<T>, BrdReplicate<T> > > filter_src(brdSrc);\r
+            BorderReader< PtrStep<T>, BrdReplicate<T> > brdSrc(src, brd);\r
+            Filter< BorderReader< PtrStep<T>, BrdReplicate<T> > > filter_src(brdSrc);\r
 \r
             resize<<<grid, block, 0, stream>>>(filter_src, fx, fy, dst);\r
             cudaSafeCall( cudaGetLastError() );\r
@@ -103,7 +103,7 @@ namespace cv { namespace gpu { namespace imgproc
             dim3 grid(divUp(dst.cols, block.x), divUp(dst.rows, block.y));\r
 \r
             BrdReplicate<T> brd(src.rows, src.cols);\r
-            BorderReader< PtrStep_<T>, BrdReplicate<T> > brdSrc(src, brd);\r
+            BorderReader< PtrStep<T>, BrdReplicate<T> > brdSrc(src, brd);\r
 \r
             resizeNN<<<grid, block, 0, stream>>>(brdSrc, fx, fy, dst);\r
             cudaSafeCall( cudaGetLastError() );\r
@@ -118,8 +118,8 @@ namespace cv { namespace gpu { namespace imgproc
             dim3 grid(divUp(dst.cols, block.x), divUp(dst.rows, block.y));\r
 \r
             BrdReplicate<T> brd(src.rows, src.cols);\r
-            BorderReader< PtrStep_<T>, BrdReplicate<T> > brdSrc(src, brd);\r
-            Filter< BorderReader< PtrStep_<T>, BrdReplicate<T> > > filter_src(brdSrc);\r
+            BorderReader< PtrStep<T>, BrdReplicate<T> > brdSrc(src, brd);\r
+            Filter< BorderReader< PtrStep<T>, BrdReplicate<T> > > filter_src(brdSrc);\r
 \r
             resize<<<grid, block>>>(filter_src, fx, fy, dst);\r
             cudaSafeCall( cudaGetLastError() );\r
@@ -135,7 +135,7 @@ namespace cv { namespace gpu { namespace imgproc
             dim3 grid(divUp(dst.cols, block.x), divUp(dst.rows, block.y));\r
 \r
             BrdReplicate<T> brd(src.rows, src.cols);\r
-            BorderReader< PtrStep_<T>, BrdReplicate<T> > brdSrc(src, brd);\r
+            BorderReader< PtrStep<T>, BrdReplicate<T> > brdSrc(src, brd);\r
 \r
             resizeNN<<<grid, block>>>(brdSrc, fx, fy, dst);\r
             cudaSafeCall( cudaGetLastError() );\r
@@ -220,7 +220,7 @@ namespace cv { namespace gpu { namespace imgproc
         }\r
     };\r
 \r
-    template <typename T> void resize_gpu(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream)\r
+    template <typename T> void resize_gpu(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream)\r
     {\r
         typedef void (*caller_t)(const DevMem2D_<T>& src, float fx, float fy, const DevMem2D_<T>& dst, cudaStream_t stream);\r
 \r
@@ -232,33 +232,33 @@ namespace cv { namespace gpu { namespace imgproc
         callers[interpolation](static_cast< DevMem2D_<T> >(src), fx, fy, static_cast< DevMem2D_<T> >(dst), stream);\r
     }\r
 \r
-    template void resize_gpu<uchar >(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
-    //template void resize_gpu<uchar2>(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
-    template void resize_gpu<uchar3>(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
-    template void resize_gpu<uchar4>(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
+    template void resize_gpu<uchar >(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
+    //template void resize_gpu<uchar2>(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
+    template void resize_gpu<uchar3>(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
+    template void resize_gpu<uchar4>(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
     \r
-    //template void resize_gpu<schar>(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
-    //template void resize_gpu<char2>(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
-    //template void resize_gpu<char3>(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
-    //template void resize_gpu<char4>(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
+    //template void resize_gpu<schar>(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
+    //template void resize_gpu<char2>(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
+    //template void resize_gpu<char3>(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
+    //template void resize_gpu<char4>(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
     \r
-    template void resize_gpu<ushort >(const DevMem2D& src,float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
-    //template void resize_gpu<ushort2>(const DevMem2D& src,float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
-    template void resize_gpu<ushort3>(const DevMem2D& src,float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
-    template void resize_gpu<ushort4>(const DevMem2D& src,float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
+    template void resize_gpu<ushort >(const DevMem2Db& src,float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
+    //template void resize_gpu<ushort2>(const DevMem2Db& src,float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
+    template void resize_gpu<ushort3>(const DevMem2Db& src,float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
+    template void resize_gpu<ushort4>(const DevMem2Db& src,float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
     \r
-    template void resize_gpu<short >(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
-    //template void resize_gpu<short2>(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
-    template void resize_gpu<short3>(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
-    template void resize_gpu<short4>(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
+    template void resize_gpu<short >(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
+    //template void resize_gpu<short2>(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
+    template void resize_gpu<short3>(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
+    template void resize_gpu<short4>(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
     \r
-    //template void resize_gpu<int >(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
-    //template void resize_gpu<int2>(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
-    //template void resize_gpu<int3>(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
-    //template void resize_gpu<int4>(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
+    //template void resize_gpu<int >(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
+    //template void resize_gpu<int2>(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
+    //template void resize_gpu<int3>(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
+    //template void resize_gpu<int4>(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
     \r
-    template void resize_gpu<float >(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
-    //template void resize_gpu<float2>(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
-    template void resize_gpu<float3>(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
-    template void resize_gpu<float4>(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
+    template void resize_gpu<float >(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
+    //template void resize_gpu<float2>(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
+    template void resize_gpu<float3>(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
+    template void resize_gpu<float4>(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
 }}}\r

diff --git a/modules/gpu/src/cuda/row_filter.cu b/modules/gpu/src/cuda/row_filter.cu
index 77c1227..cdcee7a 100644 (file)
--- a/modules/gpu/src/cuda/row_filter.cu
+++ b/modules/gpu/src/cuda/row_filter.cu
@@ -84,7 +84,7 @@ namespace filter_row
     };\r
 \r
     template <int KERNEL_SIZE, typename T, typename D, typename B>\r
-    __global__ void linearRowFilter(const DevMem2D_<T> src, PtrStep_<D> dst, int anchor, const B b)\r
+    __global__ void linearRowFilter(const DevMem2D_<T> src, PtrStep<D> dst, int anchor, const B b)\r
     {\r
         typedef typename SmemType<T>::smem_t smem_t;\r
         typedef typename TypeVec<float, VecTraits<T>::cn>::vec_type sum_t;\r
@@ -156,7 +156,7 @@ namespace cv { namespace gpu { namespace filters
     }\r
 \r
     template <typename T, typename D>\r
-    void linearRowFilter_gpu(const DevMem2D& src, const DevMem2D& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream)\r
+    void linearRowFilter_gpu(const DevMem2Db& src, const DevMem2Db& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream)\r
     {\r
         typedef void (*caller_t)(const DevMem2D_<T>& src, const DevMem2D_<D>& dst, int anchor, cudaStream_t stream);\r
         static const caller_t callers[5][17] = \r
@@ -263,11 +263,11 @@ namespace cv { namespace gpu { namespace filters
         callers[brd_type][ksize]((DevMem2D_<T>)src, (DevMem2D_<D>)dst, anchor, stream);\r
     }\r
 \r
-    template void linearRowFilter_gpu<uchar , float >(const DevMem2D& src, const DevMem2D& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
-    template void linearRowFilter_gpu<uchar4, float4>(const DevMem2D& src, const DevMem2D& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
-    //template void linearRowFilter_gpu<short , float >(const DevMem2D& src, const DevMem2D& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
-    //template void linearRowFilter_gpu<short2, float2>(const DevMem2D& src, const DevMem2D& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
-    template void linearRowFilter_gpu<short3, float3>(const DevMem2D& src, const DevMem2D& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
-    template void linearRowFilter_gpu<int   , float >(const DevMem2D& src, const DevMem2D& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
-    template void linearRowFilter_gpu<float , float >(const DevMem2D& src, const DevMem2D& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
+    template void linearRowFilter_gpu<uchar , float >(const DevMem2Db& src, const DevMem2Db& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
+    template void linearRowFilter_gpu<uchar4, float4>(const DevMem2Db& src, const DevMem2Db& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
+    //template void linearRowFilter_gpu<short , float >(const DevMem2Db& src, const DevMem2Db& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
+    //template void linearRowFilter_gpu<short2, float2>(const DevMem2Db& src, const DevMem2Db& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
+    template void linearRowFilter_gpu<short3, float3>(const DevMem2Db& src, const DevMem2Db& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
+    template void linearRowFilter_gpu<int   , float >(const DevMem2Db& src, const DevMem2Db& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
+    template void linearRowFilter_gpu<float , float >(const DevMem2Db& src, const DevMem2Db& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
 }}}\r

diff --git a/modules/gpu/src/cuda/split_merge.cu b/modules/gpu/src/cuda/split_merge.cu
index 1f5ad73..25bc294 100644 (file)
--- a/modules/gpu/src/cuda/split_merge.cu
+++ b/modules/gpu/src/cuda/split_merge.cu
@@ -89,8 +89,8 @@ namespace cv { namespace gpu { namespace split_merge {
         //typedef double4 type3;\r
     };\r
 \r
-    typedef void (*MergeFunction)(const DevMem2D* src, DevMem2D& dst, const cudaStream_t& stream);\r
-    typedef void (*SplitFunction)(const DevMem2D& src, DevMem2D* dst, const cudaStream_t& stream);\r
+    typedef void (*MergeFunction)(const DevMem2Db* src, DevMem2Db& dst, const cudaStream_t& stream);\r
+    typedef void (*SplitFunction)(const DevMem2Db& src, DevMem2Db* dst, const cudaStream_t& stream);\r
 \r
     //------------------------------------------------------------\r
     // Merge    \r
@@ -224,7 +224,7 @@ namespace cv { namespace gpu { namespace split_merge {
 \r
 \r
     template <typename T>\r
-    static void mergeC2_(const DevMem2D* src, DevMem2D& dst, const cudaStream_t& stream)\r
+    static void mergeC2_(const DevMem2Db* src, DevMem2Db& dst, const cudaStream_t& stream)\r
     {\r
         dim3 blockDim(32, 8);\r
         dim3 gridDim(divUp(dst.cols, blockDim.x), divUp(dst.rows, blockDim.y));\r
@@ -240,7 +240,7 @@ namespace cv { namespace gpu { namespace split_merge {
 \r
 \r
     template <typename T>\r
-    static void mergeC3_(const DevMem2D* src, DevMem2D& dst, const cudaStream_t& stream)\r
+    static void mergeC3_(const DevMem2Db* src, DevMem2Db& dst, const cudaStream_t& stream)\r
     {\r
         dim3 blockDim(32, 8);\r
         dim3 gridDim(divUp(dst.cols, blockDim.x), divUp(dst.rows, blockDim.y));\r
@@ -257,7 +257,7 @@ namespace cv { namespace gpu { namespace split_merge {
 \r
 \r
     template <typename T>\r
-    static void mergeC4_(const DevMem2D* src, DevMem2D& dst, const cudaStream_t& stream)\r
+    static void mergeC4_(const DevMem2Db* src, DevMem2Db& dst, const cudaStream_t& stream)\r
     {\r
         dim3 blockDim(32, 8);\r
         dim3 gridDim(divUp(dst.cols, blockDim.x), divUp(dst.rows, blockDim.y));\r
@@ -274,7 +274,7 @@ namespace cv { namespace gpu { namespace split_merge {
     }\r
 \r
 \r
-    extern "C" void merge_caller(const DevMem2D* src, DevMem2D& dst,\r
+    extern "C" void merge_caller(const DevMem2Db* src, DevMem2Db& dst,\r
                                  int total_channels, size_t elem_size,\r
                                  const cudaStream_t& stream)\r
     {\r
@@ -433,7 +433,7 @@ namespace cv { namespace gpu { namespace split_merge {
     }\r
 \r
     template <typename T>\r
-    static void splitC2_(const DevMem2D& src, DevMem2D* dst, const cudaStream_t& stream)\r
+    static void splitC2_(const DevMem2Db& src, DevMem2Db* dst, const cudaStream_t& stream)\r
     {\r
         dim3 blockDim(32, 8);\r
         dim3 gridDim(divUp(src.cols, blockDim.x), divUp(src.rows, blockDim.y));\r
@@ -449,7 +449,7 @@ namespace cv { namespace gpu { namespace split_merge {
 \r
 \r
     template <typename T>\r
-    static void splitC3_(const DevMem2D& src, DevMem2D* dst, const cudaStream_t& stream)\r
+    static void splitC3_(const DevMem2Db& src, DevMem2Db* dst, const cudaStream_t& stream)\r
     {\r
         dim3 blockDim(32, 8);\r
         dim3 gridDim(divUp(src.cols, blockDim.x), divUp(src.rows, blockDim.y));\r
@@ -466,7 +466,7 @@ namespace cv { namespace gpu { namespace split_merge {
 \r
 \r
     template <typename T>\r
-    static void splitC4_(const DevMem2D& src, DevMem2D* dst, const cudaStream_t& stream)\r
+    static void splitC4_(const DevMem2Db& src, DevMem2Db* dst, const cudaStream_t& stream)\r
     {\r
         dim3 blockDim(32, 8);\r
         dim3 gridDim(divUp(src.cols, blockDim.x), divUp(src.rows, blockDim.y));\r
@@ -483,7 +483,7 @@ namespace cv { namespace gpu { namespace split_merge {
     }\r
 \r
 \r
-    extern "C" void split_caller(const DevMem2D& src, DevMem2D* dst,\r
+    extern "C" void split_caller(const DevMem2Db& src, DevMem2Db* dst,\r
                                  int num_channels, size_t elem_size1,\r
                                  const cudaStream_t& stream)\r
     {\r

diff --git a/modules/gpu/src/cuda/stereobm.cu b/modules/gpu/src/cuda/stereobm.cu
index 928f729..3e99ce4 100644 (file)
--- a/modules/gpu/src/cuda/stereobm.cu
+++ b/modules/gpu/src/cuda/stereobm.cu
@@ -232,7 +232,7 @@ __device__ void InitColSSD(int x_tex, int y_tex, int im_pitch, unsigned char* im
 }\r
 \r
 template<int RADIUS>\r
-__global__ void stereoKernel(unsigned char *left, unsigned char *right, size_t img_step, PtrStep disp, int maxdisp)\r
+__global__ void stereoKernel(unsigned char *left, unsigned char *right, size_t img_step, PtrStepb disp, int maxdisp)\r
 {\r
     extern __shared__ unsigned int col_ssd_cache[];\r
     volatile unsigned int *col_ssd = col_ssd_cache + BLOCK_W + threadIdx.x;\r
@@ -313,7 +313,7 @@ __global__ void stereoKernel(unsigned char *left, unsigned char *right, size_t i
 }\r
 \r
 \r
-template<int RADIUS> void kernel_caller(const DevMem2D& left, const DevMem2D& right, const DevMem2D& disp, int maxdisp, cudaStream_t & stream)\r
+template<int RADIUS> void kernel_caller(const DevMem2Db& left, const DevMem2Db& right, const DevMem2Db& disp, int maxdisp, cudaStream_t & stream)\r
 {\r
     dim3 grid(1,1,1);\r
     dim3 threads(BLOCK_W, 1, 1);\r
@@ -331,7 +331,7 @@ template<int RADIUS> void kernel_caller(const DevMem2D& left, const DevMem2D& ri
         cudaSafeCall( cudaDeviceSynchronize() );\r
 };\r
 \r
-typedef void (*kernel_caller_t)(const DevMem2D& left, const DevMem2D& right, const DevMem2D& disp, int maxdisp, cudaStream_t & stream);\r
+typedef void (*kernel_caller_t)(const DevMem2Db& left, const DevMem2Db& right, const DevMem2Db& disp, int maxdisp, cudaStream_t & stream);\r
 \r
 const static kernel_caller_t callers[] =\r
 {\r
@@ -346,7 +346,7 @@ const static kernel_caller_t callers[] =
 };\r
 const int calles_num = sizeof(callers)/sizeof(callers[0]);\r
 \r
-extern "C" void stereoBM_GPU(const DevMem2D& left, const DevMem2D& right, const DevMem2D& disp, int maxdisp, int winsz, const DevMem2D_<unsigned int>& minSSD_buf, cudaStream_t& stream)\r
+extern "C" void stereoBM_GPU(const DevMem2Db& left, const DevMem2Db& right, const DevMem2Db& disp, int maxdisp, int winsz, const DevMem2D_<unsigned int>& minSSD_buf, cudaStream_t& stream)\r
 {\r
     int winsz2 = winsz >> 1;\r
 \r
@@ -375,7 +375,7 @@ extern "C" void stereoBM_GPU(const DevMem2D& left, const DevMem2D& right, const
 \r
 texture<unsigned char, 2, cudaReadModeElementType> texForSobel;\r
 \r
-extern "C" __global__ void prefilter_kernel(DevMem2D output, int prefilterCap)\r
+extern "C" __global__ void prefilter_kernel(DevMem2Db output, int prefilterCap)\r
 {\r
     int x = blockDim.x * blockIdx.x + threadIdx.x;\r
     int y = blockDim.y * blockIdx.y + threadIdx.y;\r
@@ -392,7 +392,7 @@ extern "C" __global__ void prefilter_kernel(DevMem2D output, int prefilterCap)
     }\r
 }\r
 \r
-extern "C" void prefilter_xsobel(const DevMem2D& input, const DevMem2D& output, int prefilterCap, cudaStream_t & stream)\r
+extern "C" void prefilter_xsobel(const DevMem2Db& input, const DevMem2Db& output, int prefilterCap, cudaStream_t & stream)\r
 {\r
     cudaChannelFormatDesc desc = cudaCreateChannelDesc<unsigned char>();\r
     cudaSafeCall( cudaBindTexture2D( 0, texForSobel, input.data, desc, input.cols, input.rows, input.step ) );\r
@@ -451,7 +451,7 @@ __device__ float CalcSums(float *cols, float *cols_cache, int winsz)
 \r
 #define RpT (2 * ROWSperTHREAD)  // got experimentally\r
 \r
-extern "C" __global__ void textureness_kernel(DevMem2D disp, int winsz, float threshold)\r
+extern "C" __global__ void textureness_kernel(DevMem2Db disp, int winsz, float threshold)\r
 {\r
     int winsz2 = winsz/2;\r
     int n_dirty_pixels = (winsz2) * 2;\r
@@ -510,7 +510,7 @@ extern "C" __global__ void textureness_kernel(DevMem2D disp, int winsz, float th
     }\r
 }\r
 \r
-extern "C" void postfilter_textureness(const DevMem2D& input, int winsz, float avgTexturenessThreshold, const DevMem2D& disp, cudaStream_t & stream)\r
+extern "C" void postfilter_textureness(const DevMem2Db& input, int winsz, float avgTexturenessThreshold, const DevMem2Db& disp, cudaStream_t & stream)\r
 {\r
     avgTexturenessThreshold *= winsz * winsz;\r
 \r

diff --git a/modules/gpu/src/cuda/stereobp.cu b/modules/gpu/src/cuda/stereobp.cu
index e0f5fd4..9a980cf 100644 (file)
--- a/modules/gpu/src/cuda/stereobp.cu
+++ b/modules/gpu/src/cuda/stereobp.cu
@@ -129,7 +129,7 @@ namespace cv { namespace gpu { namespace bp
     };\r
 \r
     template <int cn, typename D>\r
-    __global__ void comp_data(const DevMem2D left, const PtrStep right, PtrElemStep_<D> data)\r
+    __global__ void comp_data(const DevMem2Db left, const PtrStepb right, PtrElemStep_<D> data)\r
     {\r
         const int x = blockIdx.x * blockDim.x + threadIdx.x;\r
         const int y = blockIdx.y * blockDim.y + threadIdx.y;\r
@@ -160,9 +160,9 @@ namespace cv { namespace gpu { namespace bp
     }\r
 \r
     template<typename T, typename D>\r
-    void comp_data_gpu(const DevMem2D& left, const DevMem2D& right, const DevMem2D& data, cudaStream_t stream);\r
+    void comp_data_gpu(const DevMem2Db& left, const DevMem2Db& right, const DevMem2Db& data, cudaStream_t stream);\r
 \r
-    template <> void comp_data_gpu<uchar, short>(const DevMem2D& left, const DevMem2D& right, const DevMem2D& data, cudaStream_t stream)\r
+    template <> void comp_data_gpu<uchar, short>(const DevMem2Db& left, const DevMem2Db& right, const DevMem2Db& data, cudaStream_t stream)\r
     {\r
         dim3 threads(32, 8, 1);\r
         dim3 grid(1, 1, 1);\r
@@ -176,7 +176,7 @@ namespace cv { namespace gpu { namespace bp
         if (stream == 0)\r
             cudaSafeCall( cudaDeviceSynchronize() );\r
     }\r
-    template <> void comp_data_gpu<uchar, float>(const DevMem2D& left, const DevMem2D& right, const DevMem2D& data, cudaStream_t stream)\r
+    template <> void comp_data_gpu<uchar, float>(const DevMem2Db& left, const DevMem2Db& right, const DevMem2Db& data, cudaStream_t stream)\r
     {\r
         dim3 threads(32, 8, 1);\r
         dim3 grid(1, 1, 1);\r
@@ -191,7 +191,7 @@ namespace cv { namespace gpu { namespace bp
             cudaSafeCall( cudaDeviceSynchronize() );\r
     }\r
 \r
-    template <> void comp_data_gpu<uchar3, short>(const DevMem2D& left, const DevMem2D& right, const DevMem2D& data, cudaStream_t stream)\r
+    template <> void comp_data_gpu<uchar3, short>(const DevMem2Db& left, const DevMem2Db& right, const DevMem2Db& data, cudaStream_t stream)\r
     {\r
         dim3 threads(32, 8, 1);\r
         dim3 grid(1, 1, 1);\r
@@ -205,7 +205,7 @@ namespace cv { namespace gpu { namespace bp
         if (stream == 0)\r
             cudaSafeCall( cudaDeviceSynchronize() );\r
     }\r
-    template <> void comp_data_gpu<uchar3, float>(const DevMem2D& left, const DevMem2D& right, const DevMem2D& data, cudaStream_t stream)\r
+    template <> void comp_data_gpu<uchar3, float>(const DevMem2Db& left, const DevMem2Db& right, const DevMem2Db& data, cudaStream_t stream)\r
     {\r
         dim3 threads(32, 8, 1);\r
         dim3 grid(1, 1, 1);\r
@@ -220,7 +220,7 @@ namespace cv { namespace gpu { namespace bp
             cudaSafeCall( cudaDeviceSynchronize() );\r
     }\r
 \r
-    template <> void comp_data_gpu<uchar4, short>(const DevMem2D& left, const DevMem2D& right, const DevMem2D& data, cudaStream_t stream)\r
+    template <> void comp_data_gpu<uchar4, short>(const DevMem2Db& left, const DevMem2Db& right, const DevMem2Db& data, cudaStream_t stream)\r
     {\r
         dim3 threads(32, 8, 1);\r
         dim3 grid(1, 1, 1);\r
@@ -234,7 +234,7 @@ namespace cv { namespace gpu { namespace bp
         if (stream == 0)\r
             cudaSafeCall( cudaDeviceSynchronize() );\r
     }\r
-    template <> void comp_data_gpu<uchar4, float>(const DevMem2D& left, const DevMem2D& right, const DevMem2D& data, cudaStream_t stream)\r
+    template <> void comp_data_gpu<uchar4, float>(const DevMem2Db& left, const DevMem2Db& right, const DevMem2Db& data, cudaStream_t stream)\r
     {\r
         dim3 threads(32, 8, 1);\r
         dim3 grid(1, 1, 1);\r
@@ -254,7 +254,7 @@ namespace cv { namespace gpu { namespace bp
 ///////////////////////////////////////////////////////////////\r
 \r
     template <typename T>\r
-    __global__ void data_step_down(int dst_cols, int dst_rows, int src_rows, const PtrStep_<T> src, PtrStep_<T> dst)\r
+    __global__ void data_step_down(int dst_cols, int dst_rows, int src_rows, const PtrStep<T> src, PtrStep<T> dst)\r
     {\r
         const int x = blockIdx.x * blockDim.x + threadIdx.x;\r
         const int y = blockIdx.y * blockDim.y + threadIdx.y;\r
@@ -274,7 +274,7 @@ namespace cv { namespace gpu { namespace bp
     }\r
 \r
     template<typename T>\r
-    void data_step_down_gpu(int dst_cols, int dst_rows, int src_rows, const DevMem2D& src, const DevMem2D& dst, cudaStream_t stream)\r
+    void data_step_down_gpu(int dst_cols, int dst_rows, int src_rows, const DevMem2Db& src, const DevMem2Db& dst, cudaStream_t stream)\r
     {\r
         dim3 threads(32, 8, 1);\r
         dim3 grid(1, 1, 1);\r
@@ -289,8 +289,8 @@ namespace cv { namespace gpu { namespace bp
             cudaSafeCall( cudaDeviceSynchronize() );\r
     }\r
 \r
-    template void data_step_down_gpu<short>(int dst_cols, int dst_rows, int src_rows, const DevMem2D& src, const DevMem2D& dst, cudaStream_t stream);\r
-    template void data_step_down_gpu<float>(int dst_cols, int dst_rows, int src_rows, const DevMem2D& src, const DevMem2D& dst, cudaStream_t stream);\r
+    template void data_step_down_gpu<short>(int dst_cols, int dst_rows, int src_rows, const DevMem2Db& src, const DevMem2Db& dst, cudaStream_t stream);\r
+    template void data_step_down_gpu<float>(int dst_cols, int dst_rows, int src_rows, const DevMem2Db& src, const DevMem2Db& dst, cudaStream_t stream);\r
 \r
 ///////////////////////////////////////////////////////////////\r
 /////////////////// level up messages  ////////////////////////\r
@@ -316,7 +316,7 @@ namespace cv { namespace gpu { namespace bp
     }\r
 \r
     template <typename T>\r
-    void level_up_messages_gpu(int dst_idx, int dst_cols, int dst_rows, int src_rows, DevMem2D* mus, DevMem2D* mds, DevMem2D* mls, DevMem2D* mrs, cudaStream_t stream)\r
+    void level_up_messages_gpu(int dst_idx, int dst_cols, int dst_rows, int src_rows, DevMem2Db* mus, DevMem2Db* mds, DevMem2Db* mls, DevMem2Db* mrs, cudaStream_t stream)\r
     {\r
         dim3 threads(32, 8, 1);\r
         dim3 grid(1, 1, 1);\r
@@ -339,8 +339,8 @@ namespace cv { namespace gpu { namespace bp
             cudaSafeCall( cudaDeviceSynchronize() );\r
     }\r
 \r
-    template void level_up_messages_gpu<short>(int dst_idx, int dst_cols, int dst_rows, int src_rows, DevMem2D* mus, DevMem2D* mds, DevMem2D* mls, DevMem2D* mrs, cudaStream_t stream);\r
-    template void level_up_messages_gpu<float>(int dst_idx, int dst_cols, int dst_rows, int src_rows, DevMem2D* mus, DevMem2D* mds, DevMem2D* mls, DevMem2D* mrs, cudaStream_t stream);\r
+    template void level_up_messages_gpu<short>(int dst_idx, int dst_cols, int dst_rows, int src_rows, DevMem2Db* mus, DevMem2Db* mds, DevMem2Db* mls, DevMem2Db* mrs, cudaStream_t stream);\r
+    template void level_up_messages_gpu<float>(int dst_idx, int dst_cols, int dst_rows, int src_rows, DevMem2Db* mus, DevMem2Db* mds, DevMem2Db* mls, DevMem2Db* mrs, cudaStream_t stream);\r
 \r
 ///////////////////////////////////////////////////////////////\r
 ////////////////////  calc all iterations /////////////////////\r
@@ -441,8 +441,8 @@ namespace cv { namespace gpu { namespace bp
     }\r
 \r
     template <typename T>\r
-    void calc_all_iterations_gpu(int cols, int rows, int iters, const DevMem2D& u, const DevMem2D& d,\r
-        const DevMem2D& l, const DevMem2D& r, const DevMem2D& data, cudaStream_t stream)\r
+    void calc_all_iterations_gpu(int cols, int rows, int iters, const DevMem2Db& u, const DevMem2Db& d,\r
+        const DevMem2Db& l, const DevMem2Db& r, const DevMem2Db& data, cudaStream_t stream)\r
     {\r
         dim3 threads(32, 8, 1);\r
         dim3 grid(1, 1, 1);\r
@@ -460,8 +460,8 @@ namespace cv { namespace gpu { namespace bp
         }\r
     }\r
 \r
-    template void calc_all_iterations_gpu<short>(int cols, int rows, int iters, const DevMem2D& u, const DevMem2D& d, const DevMem2D& l, const DevMem2D& r, const DevMem2D& data, cudaStream_t stream);\r
-    template void calc_all_iterations_gpu<float>(int cols, int rows, int iters, const DevMem2D& u, const DevMem2D& d, const DevMem2D& l, const DevMem2D& r, const DevMem2D& data, cudaStream_t stream);\r
+    template void calc_all_iterations_gpu<short>(int cols, int rows, int iters, const DevMem2Db& u, const DevMem2Db& d, const DevMem2Db& l, const DevMem2Db& r, const DevMem2Db& data, cudaStream_t stream);\r
+    template void calc_all_iterations_gpu<float>(int cols, int rows, int iters, const DevMem2Db& u, const DevMem2Db& d, const DevMem2Db& l, const DevMem2Db& r, const DevMem2Db& data, cudaStream_t stream);\r
 \r
 ///////////////////////////////////////////////////////////////\r
 /////////////////////////// output ////////////////////////////\r
@@ -506,7 +506,7 @@ namespace cv { namespace gpu { namespace bp
     }\r
 \r
     template <typename T>\r
-    void output_gpu(const DevMem2D& u, const DevMem2D& d, const DevMem2D& l, const DevMem2D& r, const DevMem2D& data,\r
+    void output_gpu(const DevMem2Db& u, const DevMem2Db& d, const DevMem2Db& l, const DevMem2Db& r, const DevMem2Db& data,\r
         const DevMem2D_<short>& disp, cudaStream_t stream)\r
     {\r
         dim3 threads(32, 8, 1);\r
@@ -522,6 +522,6 @@ namespace cv { namespace gpu { namespace bp
             cudaSafeCall( cudaDeviceSynchronize() );\r
     }\r
 \r
-    template void output_gpu<short>(const DevMem2D& u, const DevMem2D& d, const DevMem2D& l, const DevMem2D& r, const DevMem2D& data, const DevMem2D_<short>& disp, cudaStream_t stream);\r
-    template void output_gpu<float>(const DevMem2D& u, const DevMem2D& d, const DevMem2D& l, const DevMem2D& r, const DevMem2D& data, const DevMem2D_<short>& disp, cudaStream_t stream);\r
+    template void output_gpu<short>(const DevMem2Db& u, const DevMem2Db& d, const DevMem2Db& l, const DevMem2Db& r, const DevMem2Db& data, const DevMem2D_<short>& disp, cudaStream_t stream);\r
+    template void output_gpu<float>(const DevMem2Db& u, const DevMem2Db& d, const DevMem2Db& l, const DevMem2Db& r, const DevMem2Db& data, const DevMem2D_<short>& disp, cudaStream_t stream);\r
 }}}\r

diff --git a/modules/gpu/src/cuda/stereocsbp.cu b/modules/gpu/src/cuda/stereocsbp.cu
index f3f0680..e43800b 100644 (file)
--- a/modules/gpu/src/cuda/stereocsbp.cu
+++ b/modules/gpu/src/cuda/stereocsbp.cu
@@ -76,7 +76,7 @@ namespace cv { namespace gpu { namespace csbp
 \r
 \r
     void load_constants(int ndisp, float max_data_term, float data_weight, float max_disc_term, float disc_single_jump, int min_disp_th,\r
-                        const DevMem2D& left, const DevMem2D& right, const DevMem2D& temp)\r
+                        const DevMem2Db& left, const DevMem2Db& right, const DevMem2Db& temp)\r
     {\r
         cudaSafeCall( cudaMemcpyToSymbol(cndisp, &ndisp, sizeof(int)) );\r
 \r

diff --git a/modules/gpu/src/cudastream.cpp b/modules/gpu/src/cudastream.cpp
index b00391e..880794b 100644 (file)
--- a/modules/gpu/src/cudastream.cpp
+++ b/modules/gpu/src/cudastream.cpp
@@ -72,14 +72,14 @@ cv::gpu::Stream::operator bool() const { throw_nogpu(); return false; }
 #include "opencv2/gpu/stream_accessor.hpp"\r
 \r
 namespace cv { namespace gpu { namespace device {            \r
-    void copy_to_with_mask(const DevMem2D& src, DevMem2D dst, int depth, const DevMem2D& mask, int channels, const cudaStream_t & stream = 0);\r
+    void copy_to_with_mask(const DevMem2Db& src, DevMem2Db dst, int depth, const DevMem2Db& mask, int channels, const cudaStream_t & stream = 0);\r
 \r
     template <typename T>\r
-    void set_to_gpu(const DevMem2D& mat, const T* scalar, int channels, cudaStream_t stream);\r
+    void set_to_gpu(const DevMem2Db& mat, const T* scalar, int channels, cudaStream_t stream);\r
     template <typename T>\r
-    void set_to_gpu(const DevMem2D& mat, const T* scalar, const DevMem2D& mask, int channels, cudaStream_t stream);\r
+    void set_to_gpu(const DevMem2Db& mat, const T* scalar, const DevMem2Db& mask, int channels, cudaStream_t stream);\r
 \r
-    void convert_gpu(const DevMem2D& src, int sdepth, const DevMem2D& dst, int ddepth, double alpha, double beta, cudaStream_t stream = 0);\r
+    void convert_gpu(const DevMem2Db& src, int sdepth, const DevMem2Db& dst, int ddepth, double alpha, double beta, cudaStream_t stream = 0);\r
 }}}\r
 \r
 struct Stream::Impl\r

diff --git a/modules/gpu/src/element_operations.cpp b/modules/gpu/src/element_operations.cpp
index b816de2..ae2f41a 100644 (file)
--- a/modules/gpu/src/element_operations.cpp
+++ b/modules/gpu/src/element_operations.cpp
@@ -126,17 +126,17 @@ namespace
 namespace cv { namespace gpu { namespace device\r
 {\r
     template <typename T, typename D> \r
-    void add_gpu(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
+    void add_gpu(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
 \r
     template <typename T, typename D> \r
-    void add_gpu(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
+    void add_gpu(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
 }}}\r
 \r
 void cv::gpu::add(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, const GpuMat& mask, int dtype, Stream& s)\r
 {\r
     using namespace cv::gpu::device;\r
 \r
-    typedef void (*func_t)(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
+    typedef void (*func_t)(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
 \r
     static const func_t funcs[7][7] = \r
     {\r
@@ -175,7 +175,7 @@ void cv::gpu::add(const GpuMat& src, const Scalar& sc, GpuMat& dst, const GpuMat
 {\r
     using namespace cv::gpu::device;\r
 \r
-    typedef void (*func_t)(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
+    typedef void (*func_t)(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
 \r
     static const func_t funcs[7][7] = \r
     {\r
@@ -238,17 +238,17 @@ void cv::gpu::add(const GpuMat& src, const Scalar& sc, GpuMat& dst, const GpuMat
 namespace cv { namespace gpu { namespace device\r
 {\r
     template <typename T, typename D> \r
-    void subtract_gpu(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
+    void subtract_gpu(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
 \r
     template <typename T, typename D> \r
-    void subtract_gpu(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
+    void subtract_gpu(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
 }}}\r
 \r
 void cv::gpu::subtract(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, const GpuMat& mask, int dtype, Stream& s)\r
 {\r
     using namespace cv::gpu::device;\r
 \r
-    typedef void (*func_t)(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
+    typedef void (*func_t)(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
 \r
     static const func_t funcs[7][7] = \r
     {\r
@@ -287,7 +287,7 @@ void cv::gpu::subtract(const GpuMat& src, const Scalar& sc, GpuMat& dst, const G
 {\r
     using namespace cv::gpu::device;\r
 \r
-    typedef void (*func_t)(const DevMem2D& src1, double val, const DevMem2D& dst, const PtrStep& mask, cudaStream_t stream);\r
+    typedef void (*func_t)(const DevMem2Db& src1, double val, const DevMem2Db& dst, const PtrStepb& mask, cudaStream_t stream);\r
 \r
     static const func_t funcs[7][7] = \r
     {\r
@@ -353,17 +353,17 @@ namespace cv { namespace gpu { namespace device
     void multiply_gpu(const DevMem2D_<short4>& src1, const DevMem2Df& src2, const DevMem2D_<short4>& dst, cudaStream_t stream);\r
 \r
     template <typename T, typename D> \r
-    void multiply_gpu(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
+    void multiply_gpu(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
 \r
     template <typename T, typename D> \r
-    void multiply_gpu(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
+    void multiply_gpu(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
 }}}\r
 \r
 void cv::gpu::multiply(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, double scale, int dtype, Stream& s)\r
 {\r
     using namespace cv::gpu::device;\r
 \r
-    typedef void (*func_t)(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
+    typedef void (*func_t)(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
 \r
     static const func_t funcs[7][7] = \r
     {\r
@@ -421,7 +421,7 @@ void cv::gpu::multiply(const GpuMat& src, const Scalar& sc, GpuMat& dst, double
 {\r
     using namespace cv::gpu::device;\r
 \r
-    typedef void (*func_t)(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
+    typedef void (*func_t)(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
 \r
     static const func_t funcs[7][7] = \r
     {\r
@@ -475,20 +475,20 @@ namespace cv { namespace gpu { namespace device
     void divide_gpu(const DevMem2D_<short4>& src1, const DevMem2Df& src2, const DevMem2D_<short4>& dst, cudaStream_t stream);\r
     \r
     template <typename T, typename D> \r
-    void divide_gpu(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
+    void divide_gpu(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
     \r
     template <typename T, typename D> \r
-    void divide_gpu(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
+    void divide_gpu(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
     \r
     template <typename T, typename D> \r
-    void divide_gpu(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
+    void divide_gpu(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
 }}}\r
 \r
 void cv::gpu::divide(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, double scale, int dtype, Stream& s)\r
 {\r
     using namespace cv::gpu::device;\r
 \r
-    typedef void (*func_t)(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, double scale, cudaStream_t stream);\r
+    typedef void (*func_t)(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
 \r
     static const func_t funcs[7][7] = \r
     {\r
@@ -546,7 +546,7 @@ void cv::gpu::divide(const GpuMat& src, const Scalar& sc, GpuMat& dst, double sc
 {\r
     using namespace cv::gpu::device;\r
 \r
-    typedef void (*func_t)(const DevMem2D& src1, double val, const DevMem2D& dst, double scale, cudaStream_t stream);\r
+    typedef void (*func_t)(const DevMem2Db& src1, double val, const DevMem2Db& dst, double scale, cudaStream_t stream);\r
 \r
     static const func_t funcs[7][7] = \r
     {\r
@@ -595,7 +595,7 @@ void cv::gpu::divide(double scale, const GpuMat& src, GpuMat& dst, int dtype, St
 {\r
     using namespace cv::gpu::device;\r
 \r
-    typedef void (*func_t)(double scalar, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
+    typedef void (*func_t)(double scalar, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
 \r
     static const func_t funcs[7][7] = \r
     {\r
@@ -629,17 +629,17 @@ void cv::gpu::divide(double scale, const GpuMat& src, GpuMat& dst, int dtype, St
 namespace cv { namespace gpu { namespace device\r
 {\r
     template <typename T> \r
-    void absdiff_gpu(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
+    void absdiff_gpu(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
     \r
     template <typename T> \r
-    void absdiff_gpu(const DevMem2D& src1, double val, const DevMem2D& dst, cudaStream_t stream);\r
+    void absdiff_gpu(const DevMem2Db& src1, double val, const DevMem2Db& dst, cudaStream_t stream);\r
 }}}\r
 \r
 void cv::gpu::absdiff(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, Stream& s)\r
 {\r
     using namespace cv::gpu::device;\r
 \r
-    typedef void (*func_t)(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
+    typedef void (*func_t)(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
 \r
     static const func_t funcs[] = \r
     {\r
@@ -711,7 +711,7 @@ void cv::gpu::absdiff(const GpuMat& src1, const Scalar& src2, GpuMat& dst, Strea
 {\r
     using namespace cv::gpu::device;\r
 \r
-    typedef void (*func_t)(const DevMem2D& src1, double val, const DevMem2D& dst, cudaStream_t stream);\r
+    typedef void (*func_t)(const DevMem2Db& src1, double val, const DevMem2Db& dst, cudaStream_t stream);\r
 \r
     static const func_t funcs[] = \r
     {\r
@@ -755,17 +755,17 @@ void cv::gpu::absdiff(const GpuMat& src1, const Scalar& src2, GpuMat& dst, Strea
 \r
 namespace cv { namespace gpu { namespace device\r
 {\r
-    template <typename T> void compare_eq(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template <typename T> void compare_ne(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template <typename T> void compare_lt(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
-    template <typename T> void compare_le(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
+    template <typename T> void compare_eq(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template <typename T> void compare_ne(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template <typename T> void compare_lt(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
+    template <typename T> void compare_le(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
 }}}\r
 \r
 void cv::gpu::compare(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, int cmpop, Stream& stream)\r
 {\r
     using namespace cv::gpu::device;\r
 \r
-    typedef void (*func_t)(const DevMem2D& src1, const DevMem2D& src2, const DevMem2D& dst, cudaStream_t stream);\r
+    typedef void (*func_t)(const DevMem2Db& src1, const DevMem2Db& src2, const DevMem2Db& dst, cudaStream_t stream);\r
 \r
     static const func_t funcs[7][4] = \r
     {\r
@@ -831,10 +831,10 @@ void cv::gpu::compare(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, int c
 \r
 namespace cv { namespace gpu { namespace device\r
 {\r
-    void bitwiseNotCaller(int rows, int cols, size_t elem_size1, int cn, const PtrStep src, PtrStep dst, cudaStream_t stream);\r
+    void bitwiseNotCaller(int rows, int cols, size_t elem_size1, int cn, const PtrStepb src, PtrStepb dst, cudaStream_t stream);\r
 \r
     template <typename T>\r
-    void bitwiseMaskNotCaller(int rows, int cols, int cn, const PtrStep src, const PtrStep mask, PtrStep dst, cudaStream_t stream);\r
+    void bitwiseMaskNotCaller(int rows, int cols, int cn, const PtrStepb src, const PtrStepb mask, PtrStepb dst, cudaStream_t stream);\r
 }}}\r
 \r
 namespace\r
@@ -852,7 +852,7 @@ namespace
     {\r
         using namespace cv::gpu;\r
 \r
-        typedef void (*Caller)(int, int, int, const PtrStep, const PtrStep, PtrStep, cudaStream_t);\r
+        typedef void (*Caller)(int, int, int, const PtrStepb, const PtrStepb, PtrStepb, cudaStream_t);\r
         static Caller callers[] = {device::bitwiseMaskNotCaller<unsigned char>, device::bitwiseMaskNotCaller<unsigned char>, \r
                                    device::bitwiseMaskNotCaller<unsigned short>, device::bitwiseMaskNotCaller<unsigned short>,\r
                                    device::bitwiseMaskNotCaller<unsigned int>, device::bitwiseMaskNotCaller<unsigned int>,\r
@@ -885,20 +885,20 @@ void cv::gpu::bitwise_not(const GpuMat& src, GpuMat& dst, const GpuMat& mask, St
 \r
 namespace cv { namespace gpu { namespace device\r
 {\r
-    void bitwiseOrCaller(int rows, int cols, size_t elem_size1, int cn, const PtrStep src1, const PtrStep src2, PtrStep dst, cudaStream_t stream);\r
+    void bitwiseOrCaller(int rows, int cols, size_t elem_size1, int cn, const PtrStepb src1, const PtrStepb src2, PtrStepb dst, cudaStream_t stream);\r
 \r
     template <typename T>\r
-    void bitwiseMaskOrCaller(int rows, int cols, int cn, const PtrStep src1, const PtrStep src2, const PtrStep mask, PtrStep dst, cudaStream_t stream);\r
+    void bitwiseMaskOrCaller(int rows, int cols, int cn, const PtrStepb src1, const PtrStepb src2, const PtrStepb mask, PtrStepb dst, cudaStream_t stream);\r
 \r
-    void bitwiseAndCaller(int rows, int cols, size_t elem_size1, int cn, const PtrStep src1, const PtrStep src2, PtrStep dst, cudaStream_t stream);\r
+    void bitwiseAndCaller(int rows, int cols, size_t elem_size1, int cn, const PtrStepb src1, const PtrStepb src2, PtrStepb dst, cudaStream_t stream);\r
 \r
     template <typename T>\r
-    void bitwiseMaskAndCaller(int rows, int cols, int cn, const PtrStep src1, const PtrStep src2, const PtrStep mask, PtrStep dst, cudaStream_t stream);\r
+    void bitwiseMaskAndCaller(int rows, int cols, int cn, const PtrStepb src1, const PtrStepb src2, const PtrStepb mask, PtrStepb dst, cudaStream_t stream);\r
 \r
-    void bitwiseXorCaller(int rows, int cols, size_t elem_size1, int cn, const PtrStep src1, const PtrStep src2, PtrStep dst, cudaStream_t stream);\r
+    void bitwiseXorCaller(int rows, int cols, size_t elem_size1, int cn, const PtrStepb src1, const PtrStepb src2, PtrStepb dst, cudaStream_t stream);\r
 \r
     template <typename T>\r
-    void bitwiseMaskXorCaller(int rows, int cols, int cn, const PtrStep src1, const PtrStep src2, const PtrStep mask, PtrStep dst, cudaStream_t stream);\r
+    void bitwiseMaskXorCaller(int rows, int cols, int cn, const PtrStepb src1, const PtrStepb src2, const PtrStepb mask, PtrStepb dst, cudaStream_t stream);\r
 }}}\r
 \r
 \r
@@ -918,7 +918,7 @@ namespace
     {\r
         using namespace cv::gpu;\r
 \r
-        typedef void (*Caller)(int, int, int, const PtrStep, const PtrStep, const PtrStep, PtrStep, cudaStream_t);\r
+        typedef void (*Caller)(int, int, int, const PtrStepb, const PtrStepb, const PtrStepb, PtrStepb, cudaStream_t);\r
         static Caller callers[] = {device::bitwiseMaskOrCaller<unsigned char>, device::bitwiseMaskOrCaller<unsigned char>, \r
                                    device::bitwiseMaskOrCaller<unsigned short>, device::bitwiseMaskOrCaller<unsigned short>,\r
                                    device::bitwiseMaskOrCaller<unsigned int>, device::bitwiseMaskOrCaller<unsigned int>,\r
@@ -949,7 +949,7 @@ namespace
     {\r
         using namespace cv::gpu;\r
 \r
-        typedef void (*Caller)(int, int, int, const PtrStep, const PtrStep, const PtrStep, PtrStep, cudaStream_t);\r
+        typedef void (*Caller)(int, int, int, const PtrStepb, const PtrStepb, const PtrStepb, PtrStepb, cudaStream_t);\r
         static Caller callers[] = {device::bitwiseMaskAndCaller<unsigned char>, device::bitwiseMaskAndCaller<unsigned char>, \r
                                    device::bitwiseMaskAndCaller<unsigned short>, device::bitwiseMaskAndCaller<unsigned short>,\r
                                    device::bitwiseMaskAndCaller<unsigned int>, device::bitwiseMaskAndCaller<unsigned int>,\r
@@ -980,7 +980,7 @@ namespace
     {\r
         using namespace cv::gpu;\r
 \r
-        typedef void (*Caller)(int, int, int, const PtrStep, const PtrStep, const PtrStep, PtrStep, cudaStream_t);\r
+        typedef void (*Caller)(int, int, int, const PtrStepb, const PtrStepb, const PtrStepb, PtrStepb, cudaStream_t);\r
         static Caller callers[] = {device::bitwiseMaskXorCaller<unsigned char>, device::bitwiseMaskXorCaller<unsigned char>, \r
                                    device::bitwiseMaskXorCaller<unsigned short>, device::bitwiseMaskXorCaller<unsigned short>,\r
                                    device::bitwiseMaskXorCaller<unsigned int>, device::bitwiseMaskXorCaller<unsigned int>,\r
@@ -1139,7 +1139,7 @@ void cv::gpu::max(const GpuMat& src1, double src2, GpuMat& dst, Stream& stream)
 namespace cv { namespace gpu { namespace device\r
 {\r
     template <typename T>\r
-    void threshold_gpu(const DevMem2D& src, const DevMem2D& dst, T thresh, T maxVal, int type,\r
+    void threshold_gpu(const DevMem2Db& src, const DevMem2Db& dst, T thresh, T maxVal, int type,\r
         cudaStream_t stream);\r
 }}}\r
 \r
@@ -1207,7 +1207,7 @@ double cv::gpu::threshold(const GpuMat& src, GpuMat& dst, double thresh, double
 namespace cv { namespace gpu { namespace device\r
 {\r
     template<typename T>\r
-    void pow_caller(const DevMem2D& src, float power, DevMem2D dst, cudaStream_t stream);\r
+    void pow_caller(const DevMem2Db& src, float power, DevMem2Db dst, cudaStream_t stream);\r
 }}}\r
 \r
 void cv::gpu::pow(const GpuMat& src, double power, GpuMat& dst, Stream& stream)\r
@@ -1215,7 +1215,7 @@ void cv::gpu::pow(const GpuMat& src, double power, GpuMat& dst, Stream& stream)
     CV_Assert( src.depth() != CV_64F );\r
     dst.create(src.size(), src.type());\r
 \r
-    typedef void (*caller_t)(const DevMem2D& src, float power, DevMem2D dst, cudaStream_t stream);\r
+    typedef void (*caller_t)(const DevMem2Db& src, float power, DevMem2Db dst, cudaStream_t stream);\r
 \r
     static const caller_t callers[] = \r
     {\r
@@ -1233,7 +1233,7 @@ void cv::gpu::pow(const GpuMat& src, double power, GpuMat& dst, Stream& stream)
 namespace cv { namespace gpu { namespace device\r
 {\r
     template <typename T1, typename T2, typename D>\r
-    void addWeighted_gpu(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
+    void addWeighted_gpu(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
 }}}\r
 \r
 void cv::gpu::addWeighted(const GpuMat& src1, double alpha, const GpuMat& src2, double beta, double gamma, GpuMat& dst, int dtype, Stream& stream)\r
@@ -1254,7 +1254,7 @@ void cv::gpu::addWeighted(const GpuMat& src1, double alpha, const GpuMat& src2,
         std::swap(alpha, beta);\r
     }\r
 \r
-    typedef void (*caller_t)(const DevMem2D& src1, double alpha, const DevMem2D& src2, double beta, double gamma, const DevMem2D& dst, cudaStream_t stream);\r
+    typedef void (*caller_t)(const DevMem2Db& src1, double alpha, const DevMem2Db& src2, double beta, double gamma, const DevMem2Db& dst, cudaStream_t stream);\r
 \r
     using namespace cv::gpu::device;\r
 \r

diff --git a/modules/gpu/src/filtering.cpp b/modules/gpu/src/filtering.cpp
index 747451c..b5e7509 100644 (file)
--- a/modules/gpu/src/filtering.cpp
+++ b/modules/gpu/src/filtering.cpp
@@ -738,10 +738,10 @@ void cv::gpu::filter2D(const GpuMat& src, GpuMat& dst, int ddepth, const Mat& ke
 namespace cv { namespace gpu { namespace filters\r
 {\r
     template <typename T, typename D>\r
-    void linearRowFilter_gpu(const DevMem2D& src, const DevMem2D& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
+    void linearRowFilter_gpu(const DevMem2Db& src, const DevMem2Db& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
 \r
     template <typename T, typename D>\r
-    void linearColumnFilter_gpu(const DevMem2D& src, const DevMem2D& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
+    void linearColumnFilter_gpu(const DevMem2Db& src, const DevMem2Db& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
 }}}\r
 \r
 namespace\r
@@ -749,7 +749,7 @@ namespace
     typedef NppStatus (*nppFilter1D_t)(const Npp8u * pSrc, Npp32s nSrcStep, Npp8u * pDst, Npp32s nDstStep, NppiSize oROI, \r
         const Npp32s * pKernel, Npp32s nMaskSize, Npp32s nAnchor, Npp32s nDivisor);\r
 \r
-    typedef void (*gpuFilter1D_t)(const DevMem2D& src, const DevMem2D& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
+    typedef void (*gpuFilter1D_t)(const DevMem2Db& src, const DevMem2Db& dst, const float kernel[], int ksize, int anchor, int brd_type, cudaStream_t stream);\r
 \r
     struct NppLinearRowFilter : public BaseRowFilter_GPU\r
     {\r

diff --git a/modules/gpu/src/gpumat.cpp b/modules/gpu/src/gpumat.cpp
index 3ba510a..8d36057 100644 (file)
--- a/modules/gpu/src/gpumat.cpp
+++ b/modules/gpu/src/gpumat.cpp
@@ -395,14 +395,14 @@ void cv::gpu::ensureSizeIsEnough(int, int, int, GpuMat&) { throw_nogpu(); }
 \r
 namespace cv { namespace gpu { namespace device\r
 {\r
-    void copy_to_with_mask(const DevMem2D& src, DevMem2D dst, int depth, const DevMem2D& mask, int channels, const cudaStream_t & stream = 0);\r
+    void copy_to_with_mask(const DevMem2Db& src, DevMem2Db dst, int depth, const DevMem2Db& mask, int channels, const cudaStream_t & stream = 0);\r
 \r
     template <typename T>\r
-    void set_to_gpu(const DevMem2D& mat, const T* scalar, int channels, cudaStream_t stream);\r
+    void set_to_gpu(const DevMem2Db& mat, const T* scalar, int channels, cudaStream_t stream);\r
     template <typename T>\r
-    void set_to_gpu(const DevMem2D& mat, const T* scalar, const DevMem2D& mask, int channels, cudaStream_t stream);\r
+    void set_to_gpu(const DevMem2Db& mat, const T* scalar, const DevMem2Db& mask, int channels, cudaStream_t stream);\r
 \r
-    void convert_gpu(const DevMem2D& src, int sdepth, const DevMem2D& dst, int ddepth, double alpha, double beta, cudaStream_t stream = 0);\r
+    void convert_gpu(const DevMem2Db& src, int sdepth, const DevMem2Db& dst, int ddepth, double alpha, double beta, cudaStream_t stream = 0);\r
 }}}\r
 \r
 \r

diff --git a/modules/gpu/src/hog.cpp b/modules/gpu/src/hog.cpp
index bd4f86d..da00258 100644 (file)
--- a/modules/gpu/src/hog.cpp
+++ b/modules/gpu/src/hog.cpp
@@ -67,7 +67,7 @@ void set_up_constants(int nbins, int block_stride_x, int block_stride_y,
 \r
 void compute_hists(int nbins, int block_stride_x, int blovck_stride_y,\r
                    int height, int width, const cv::gpu::DevMem2Df& grad, \r
-                   const cv::gpu::DevMem2D& qangle, float sigma, float* block_hists);\r
+                   const cv::gpu::DevMem2Db& qangle, float sigma, float* block_hists);\r
 \r
 void normalize_hists(int nbins, int block_stride_x, int block_stride_y, \r
                      int height, int width, float* block_hists, float threshold);\r
@@ -84,13 +84,13 @@ void extract_descrs_by_cols(int win_height, int win_width, int block_stride_y, i
                             int win_stride_y, int win_stride_x, int height, int width, float* block_hists, \r
                             cv::gpu::DevMem2Df descriptors);\r
 \r
-void compute_gradients_8UC1(int nbins, int height, int width, const cv::gpu::DevMem2D& img, \r
-                            float angle_scale, cv::gpu::DevMem2Df grad, cv::gpu::DevMem2D qangle, bool correct_gamma);\r
-void compute_gradients_8UC4(int nbins, int height, int width, const cv::gpu::DevMem2D& img, \r
-                            float angle_scale, cv::gpu::DevMem2Df grad, cv::gpu::DevMem2D qangle, bool correct_gamma);\r
+void compute_gradients_8UC1(int nbins, int height, int width, const cv::gpu::DevMem2Db& img, \r
+                            float angle_scale, cv::gpu::DevMem2Df grad, cv::gpu::DevMem2Db qangle, bool correct_gamma);\r
+void compute_gradients_8UC4(int nbins, int height, int width, const cv::gpu::DevMem2Db& img, \r
+                            float angle_scale, cv::gpu::DevMem2Df grad, cv::gpu::DevMem2Db qangle, bool correct_gamma);\r
 \r
-void resize_8UC1(const cv::gpu::DevMem2D& src, cv::gpu::DevMem2D dst);\r
-void resize_8UC4(const cv::gpu::DevMem2D& src, cv::gpu::DevMem2D dst);\r
+void resize_8UC1(const cv::gpu::DevMem2Db& src, cv::gpu::DevMem2Db dst);\r
+void resize_8UC4(const cv::gpu::DevMem2Db& src, cv::gpu::DevMem2Db dst);\r
 \r
 }}}\r
 \r

diff --git a/modules/gpu/src/imgproc.cpp b/modules/gpu/src/imgproc.cpp
index 67a36f0..f78d0a5 100644 (file)
--- a/modules/gpu/src/imgproc.cpp
+++ b/modules/gpu/src/imgproc.cpp
@@ -107,7 +107,7 @@ void cv::gpu::CannyBuf::release() { throw_nogpu(); }
 \r
 namespace cv { namespace gpu {  namespace imgproc\r
 {\r
-    template <typename T> void remap_gpu(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, \r
+    template <typename T> void remap_gpu(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, \r
                                          int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
 }}}\r
 \r
@@ -115,7 +115,7 @@ void cv::gpu::remap(const GpuMat& src, GpuMat& dst, const GpuMat& xmap, const Gp
 {\r
     using namespace cv::gpu::imgproc;\r
 \r
-    typedef void (*caller_t)(const DevMem2D& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2D& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
+    typedef void (*caller_t)(const DevMem2Db& src, const DevMem2Df& xmap, const DevMem2Df& ymap, const DevMem2Db& dst, int interpolation, int borderMode, const float* borderValue, cudaStream_t stream);\r
     static const caller_t callers[6][4] = \r
     {\r
         {remap_gpu<uchar>, 0/*remap_gpu<uchar2>*/, remap_gpu<uchar3>, remap_gpu<uchar4>},\r
@@ -151,7 +151,7 @@ void cv::gpu::remap(const GpuMat& src, GpuMat& dst, const GpuMat& xmap, const Gp
 \r
 namespace cv { namespace gpu {  namespace imgproc\r
 {\r
-    extern "C" void meanShiftFiltering_gpu(const DevMem2D& src, DevMem2D dst, int sp, int sr, int maxIter, float eps);\r
+    extern "C" void meanShiftFiltering_gpu(const DevMem2Db& src, DevMem2Db dst, int sp, int sr, int maxIter, float eps);\r
 }}}\r
 \r
 void cv::gpu::meanShiftFiltering(const GpuMat& src, GpuMat& dst, int sp, int sr, TermCriteria criteria)\r
@@ -182,7 +182,7 @@ void cv::gpu::meanShiftFiltering(const GpuMat& src, GpuMat& dst, int sp, int sr,
 \r
 namespace cv { namespace gpu {  namespace imgproc\r
 {\r
-    extern "C" void meanShiftProc_gpu(const DevMem2D& src, DevMem2D dstr, DevMem2D dstsp, int sp, int sr, int maxIter, float eps);\r
+    extern "C" void meanShiftProc_gpu(const DevMem2Db& src, DevMem2Db dstr, DevMem2Db dstsp, int sp, int sr, int maxIter, float eps);\r
 }}}\r
 \r
 void cv::gpu::meanShiftProc(const GpuMat& src, GpuMat& dstr, GpuMat& dstsp, int sp, int sr, TermCriteria criteria)\r
@@ -214,8 +214,8 @@ void cv::gpu::meanShiftProc(const GpuMat& src, GpuMat& dstr, GpuMat& dstsp, int
 \r
 namespace cv { namespace gpu {  namespace imgproc\r
 {\r
-    void drawColorDisp_gpu(const DevMem2D& src, const DevMem2D& dst, int ndisp, const cudaStream_t& stream);\r
-    void drawColorDisp_gpu(const DevMem2D_<short>& src, const DevMem2D& dst, int ndisp, const cudaStream_t& stream);\r
+    void drawColorDisp_gpu(const DevMem2Db& src, const DevMem2Db& dst, int ndisp, const cudaStream_t& stream);\r
+    void drawColorDisp_gpu(const DevMem2D_<short>& src, const DevMem2Db& dst, int ndisp, const cudaStream_t& stream);\r
 }}}\r
 \r
 namespace\r
@@ -245,7 +245,7 @@ void cv::gpu::drawColorDisp(const GpuMat& src, GpuMat& dst, int ndisp, Stream& s
 \r
 namespace cv { namespace gpu {  namespace imgproc\r
 {\r
-    void reprojectImageTo3D_gpu(const DevMem2D& disp, const DevMem2Df& xyzw, const float* q, const cudaStream_t& stream);\r
+    void reprojectImageTo3D_gpu(const DevMem2Db& disp, const DevMem2Df& xyzw, const float* q, const cudaStream_t& stream);\r
     void reprojectImageTo3D_gpu(const DevMem2D_<short>& disp, const DevMem2Df& xyzw, const float* q, const cudaStream_t& stream);\r
 }}}\r
 \r
@@ -275,7 +275,7 @@ void cv::gpu::reprojectImageTo3D(const GpuMat& disp, GpuMat& xyzw, const Mat& Q,
 \r
 namespace cv { namespace gpu {  namespace imgproc\r
 {\r
-    template <typename T> void resize_gpu(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
+    template <typename T> void resize_gpu(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
 }}}\r
 \r
 void cv::gpu::resize(const GpuMat& src, GpuMat& dst, Size dsize, double fx, double fy, int interpolation, Stream& s)\r
@@ -342,7 +342,7 @@ void cv::gpu::resize(const GpuMat& src, GpuMat& dst, Size dsize, double fx, doub
     {\r
         using namespace cv::gpu::imgproc;\r
 \r
-        typedef void (*caller_t)(const DevMem2D& src, float fx, float fy, const DevMem2D& dst, int interpolation, cudaStream_t stream);\r
+        typedef void (*caller_t)(const DevMem2Db& src, float fx, float fy, const DevMem2Db& dst, int interpolation, cudaStream_t stream);\r
         static const caller_t callers[6][4] = \r
         {\r
             {resize_gpu<uchar>, 0/*resize_gpu<uchar2>*/, resize_gpu<uchar3>, resize_gpu<uchar4>},\r
@@ -362,12 +362,12 @@ void cv::gpu::resize(const GpuMat& src, GpuMat& dst, Size dsize, double fx, doub
 \r
 namespace cv { namespace gpu {  namespace imgproc\r
 {\r
-    template <typename T, int cn> void copyMakeBorder_gpu(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderMode, const T* borderValue, cudaStream_t stream);\r
+    template <typename T, int cn> void copyMakeBorder_gpu(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderMode, const T* borderValue, cudaStream_t stream);\r
 }}}\r
 \r
 namespace\r
 {\r
-    template <typename T, int cn> void copyMakeBorder_caller(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderType, const Scalar& value, cudaStream_t stream)\r
+    template <typename T, int cn> void copyMakeBorder_caller(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderType, const Scalar& value, cudaStream_t stream)\r
     {\r
         Scalar_<T> val(saturate_cast<T>(value[0]), saturate_cast<T>(value[1]), saturate_cast<T>(value[2]), saturate_cast<T>(value[3]));\r
 \r
@@ -434,7 +434,7 @@ void cv::gpu::copyMakeBorder(const GpuMat& src, GpuMat& dst, int top, int bottom
     }\r
     else\r
     {\r
-        typedef void (*caller_t)(const DevMem2D& src, const DevMem2D& dst, int top, int left, int borderType, const Scalar& value, cudaStream_t stream);\r
+        typedef void (*caller_t)(const DevMem2Db& src, const DevMem2Db& dst, int top, int left, int borderType, const Scalar& value, cudaStream_t stream);\r
         static const caller_t callers[6][4] = \r
         {\r
             {   copyMakeBorder_caller<uchar, 1>  , 0/*copyMakeBorder_caller<uchar, 2>*/ ,    copyMakeBorder_caller<uchar, 3>  ,    copyMakeBorder_caller<uchar, 4>},\r
@@ -839,7 +839,7 @@ void cv::gpu::sqrIntegral(const GpuMat& src, GpuMat& sqsum, Stream& s)
 \r
 namespace cv { namespace gpu { namespace imgproc\r
 {\r
-    void columnSum_32F(const DevMem2D src, const DevMem2D dst);\r
+    void columnSum_32F(const DevMem2Db src, const DevMem2Db dst);\r
 }}}\r
 \r
 void cv::gpu::columnSum(const GpuMat& src, GpuMat& dst)\r
@@ -1179,7 +1179,7 @@ void cv::gpu::histRange(const GpuMat& src, GpuMat hist[4], const GpuMat levels[4
 \r
 namespace cv { namespace gpu { namespace histograms\r
 {\r
-    void histogram256_gpu(DevMem2D src, int* hist, unsigned int* buf, cudaStream_t stream);\r
+    void histogram256_gpu(DevMem2Db src, int* hist, unsigned int* buf, cudaStream_t stream);\r
 \r
     const int PARTIAL_HISTOGRAM256_COUNT = 240;\r
     const int HISTOGRAM256_BIN_COUNT     = 256;\r
@@ -1219,7 +1219,7 @@ void cv::gpu::equalizeHist(const GpuMat& src, GpuMat& dst, GpuMat& hist, Stream&
 \r
 namespace cv { namespace gpu { namespace histograms\r
 {\r
-    void equalizeHist_gpu(DevMem2D src, DevMem2D dst, const int* lut, cudaStream_t stream);\r
+    void equalizeHist_gpu(DevMem2Db src, DevMem2Db dst, const int* lut, cudaStream_t stream);\r
 }}}\r
 \r
 void cv::gpu::equalizeHist(const GpuMat& src, GpuMat& dst, GpuMat& hist, GpuMat& buf, Stream& s)\r
@@ -1261,8 +1261,8 @@ void cv::gpu::equalizeHist(const GpuMat& src, GpuMat& dst, GpuMat& hist, GpuMat&
 namespace cv { namespace gpu { namespace imgproc {\r
 \r
     void extractCovData_caller(const DevMem2Df Dx, const DevMem2Df Dy, PtrStepf dst);\r
-    void cornerHarris_caller(const int block_size, const float k, const DevMem2D Dx, const DevMem2D Dy, DevMem2D dst, int border_type);\r
-    void cornerMinEigenVal_caller(const int block_size, const DevMem2D Dx, const DevMem2D Dy, DevMem2D dst, int border_type);\r
+    void cornerHarris_caller(const int block_size, const float k, const DevMem2Db Dx, const DevMem2Db Dy, DevMem2Db dst, int border_type);\r
+    void cornerMinEigenVal_caller(const int block_size, const DevMem2Db Dx, const DevMem2Db Dy, DevMem2Db dst, int border_type);\r
 \r
 }}}\r
 \r
@@ -1379,10 +1379,10 @@ void cv::gpu::cornerMinEigenVal(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuM
 \r
 namespace cv { namespace gpu { namespace imgproc \r
 {\r
-    void mulSpectrums(const PtrStep_<cufftComplex> a, const PtrStep_<cufftComplex> b, \r
+    void mulSpectrums(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, \r
                       DevMem2D_<cufftComplex> c);\r
 \r
-    void mulSpectrums_CONJ(const PtrStep_<cufftComplex> a, const PtrStep_<cufftComplex> b, \r
+    void mulSpectrums_CONJ(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b, \r
                            DevMem2D_<cufftComplex> c);\r
 }}}\r
 \r
@@ -1390,7 +1390,7 @@ namespace cv { namespace gpu { namespace imgproc
 void cv::gpu::mulSpectrums(const GpuMat& a, const GpuMat& b, GpuMat& c, \r
                            int flags, bool conjB) \r
 {\r
-    typedef void (*Caller)(const PtrStep_<cufftComplex>, const PtrStep_<cufftComplex>, \r
+    typedef void (*Caller)(const PtrStep<cufftComplex>, const PtrStep<cufftComplex>, \r
                            DevMem2D_<cufftComplex>);\r
     static Caller callers[] = { imgproc::mulSpectrums, \r
                                 imgproc::mulSpectrums_CONJ };\r
@@ -1409,10 +1409,10 @@ void cv::gpu::mulSpectrums(const GpuMat& a, const GpuMat& b, GpuMat& c,
 \r
 namespace cv { namespace gpu { namespace imgproc \r
 {\r
-    void mulAndScaleSpectrums(const PtrStep_<cufftComplex> a, const PtrStep_<cufftComplex> b,\r
+    void mulAndScaleSpectrums(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b,\r
                              float scale, DevMem2D_<cufftComplex> c);\r
 \r
-    void mulAndScaleSpectrums_CONJ(const PtrStep_<cufftComplex> a, const PtrStep_<cufftComplex> b,\r
+    void mulAndScaleSpectrums_CONJ(const PtrStep<cufftComplex> a, const PtrStep<cufftComplex> b,\r
                                   float scale, DevMem2D_<cufftComplex> c);\r
 }}}\r
 \r
@@ -1420,7 +1420,7 @@ namespace cv { namespace gpu { namespace imgproc
 void cv::gpu::mulAndScaleSpectrums(const GpuMat& a, const GpuMat& b, GpuMat& c,\r
                                   int flags, float scale, bool conjB) \r
 {\r
-    typedef void (*Caller)(const PtrStep_<cufftComplex>, const PtrStep_<cufftComplex>,\r
+    typedef void (*Caller)(const PtrStep<cufftComplex>, const PtrStep<cufftComplex>,\r
                            float scale, DevMem2D_<cufftComplex>);\r
     static Caller callers[] = { imgproc::mulAndScaleSpectrums, \r
                                 imgproc::mulAndScaleSpectrums_CONJ };\r
@@ -1663,14 +1663,14 @@ void cv::gpu::convolve(const GpuMat& image, const GpuMat& templ, GpuMat& result,
 \r
 namespace cv { namespace gpu { namespace imgproc\r
 {\r
-    template <typename T, int cn> void pyrDown_gpu(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
+    template <typename T, int cn> void pyrDown_gpu(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
 }}}\r
 \r
 void cv::gpu::pyrDown(const GpuMat& src, GpuMat& dst, int borderType, Stream& stream)\r
 {\r
     using namespace cv::gpu::imgproc;\r
 \r
-    typedef void (*func_t)(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
+    typedef void (*func_t)(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
 \r
     static const func_t funcs[6][4] = \r
     {\r
@@ -1699,14 +1699,14 @@ void cv::gpu::pyrDown(const GpuMat& src, GpuMat& dst, int borderType, Stream& st
 \r
 namespace cv { namespace gpu { namespace imgproc\r
 {\r
-    template <typename T, int cn> void pyrUp_gpu(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
+    template <typename T, int cn> void pyrUp_gpu(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
 }}}\r
 \r
 void cv::gpu::pyrUp(const GpuMat& src, GpuMat& dst, int borderType, Stream& stream)\r
 {\r
     using namespace cv::gpu::imgproc;\r
 \r
-    typedef void (*func_t)(const DevMem2D& src, const DevMem2D& dst, int borderType, cudaStream_t stream);\r
+    typedef void (*func_t)(const DevMem2Db& src, const DevMem2Db& dst, int borderType, cudaStream_t stream);\r
 \r
     static const func_t funcs[6][4] = \r
     {\r
@@ -1777,7 +1777,7 @@ void cv::gpu::CannyBuf::release()
 \r
 namespace cv { namespace gpu { namespace canny\r
 {    \r
-    void calcSobelRowPass_gpu(PtrStep src, PtrStepi dx_buf, PtrStepi dy_buf, int rows, int cols);\r
+    void calcSobelRowPass_gpu(PtrStepb src, PtrStepi dx_buf, PtrStepi dy_buf, int rows, int cols);\r
 \r
     void calcMagnitude_gpu(PtrStepi dx_buf, PtrStepi dy_buf, PtrStepi dx, PtrStepi dy, PtrStepf mag, int rows, int cols, bool L2Grad);\r
     void calcMagnitude_gpu(PtrStepi dx, PtrStepi dy, PtrStepf mag, int rows, int cols, bool L2Grad);\r
@@ -1788,7 +1788,7 @@ namespace cv { namespace gpu { namespace canny
 \r
     void edgesHysteresisGlobal_gpu(PtrStepi map, ushort2* st1, ushort2* st2, int rows, int cols);\r
 \r
-    void getEdges_gpu(PtrStepi map, PtrStep dst, int rows, int cols);\r
+    void getEdges_gpu(PtrStepi map, PtrStepb dst, int rows, int cols);\r
 }}}\r
 \r
 namespace\r

diff --git a/modules/gpu/src/match_template.cpp b/modules/gpu/src/match_template.cpp
index d534503..58bc8bb 100644 (file)
--- a/modules/gpu/src/match_template.cpp
+++ b/modules/gpu/src/match_template.cpp
@@ -54,16 +54,16 @@ void cv::gpu::matchTemplate(const GpuMat&, const GpuMat&, GpuMat&, int) { throw_
 namespace cv { namespace gpu { namespace imgproc \r
 {  \r
     void matchTemplateNaive_CCORR_8U(\r
-            const DevMem2D image, const DevMem2D templ, DevMem2Df result, int cn);\r
+            const DevMem2Db image, const DevMem2Db templ, DevMem2Df result, int cn);\r
 \r
     void matchTemplateNaive_CCORR_32F(\r
-            const DevMem2D image, const DevMem2D templ, DevMem2Df result, int cn);\r
+            const DevMem2Db image, const DevMem2Db templ, DevMem2Df result, int cn);\r
 \r
     void matchTemplateNaive_SQDIFF_8U(\r
-            const DevMem2D image, const DevMem2D templ, DevMem2Df result, int cn);\r
+            const DevMem2Db image, const DevMem2Db templ, DevMem2Df result, int cn);\r
 \r
     void matchTemplateNaive_SQDIFF_32F(\r
-            const DevMem2D image, const DevMem2D templ, DevMem2Df result, int cn);\r
+            const DevMem2Db image, const DevMem2Db templ, DevMem2Df result, int cn);\r
 \r
     void matchTemplatePrepared_SQDIFF_8U(\r
             int w, int h, const DevMem2D_<unsigned long long> image_sqsum, \r
@@ -145,7 +145,7 @@ namespace cv { namespace gpu { namespace imgproc
     void normalize_8U(int w, int h, const DevMem2D_<unsigned long long> image_sqsum, \r
                       unsigned int templ_sqsum, DevMem2Df result, int cn);\r
 \r
-    void extractFirstChannel_32F(const DevMem2D image, DevMem2Df result, int cn);\r
+    void extractFirstChannel_32F(const DevMem2Db image, DevMem2Df result, int cn);\r
 }}}\r
 \r
 \r

diff --git a/modules/gpu/src/matrix_reductions.cpp b/modules/gpu/src/matrix_reductions.cpp
index 11cf4bc..60c3ed9 100644 (file)
--- a/modules/gpu/src/matrix_reductions.cpp
+++ b/modules/gpu/src/matrix_reductions.cpp
@@ -193,22 +193,22 @@ double cv::gpu::norm(const GpuMat& src1, const GpuMat& src2, int normType)
 namespace cv { namespace gpu { namespace mathfunc\r
 {\r
     template <typename T>\r
-    void sumCaller(const DevMem2D src, PtrStep buf, double* sum, int cn);\r
+    void sumCaller(const DevMem2Db src, PtrStepb buf, double* sum, int cn);\r
 \r
     template <typename T>\r
-    void sumMultipassCaller(const DevMem2D src, PtrStep buf, double* sum, int cn);\r
+    void sumMultipassCaller(const DevMem2Db src, PtrStepb buf, double* sum, int cn);\r
 \r
     template <typename T>\r
-    void absSumCaller(const DevMem2D src, PtrStep buf, double* sum, int cn);\r
+    void absSumCaller(const DevMem2Db src, PtrStepb buf, double* sum, int cn);\r
 \r
     template <typename T>\r
-    void absSumMultipassCaller(const DevMem2D src, PtrStep buf, double* sum, int cn);\r
+    void absSumMultipassCaller(const DevMem2Db src, PtrStepb buf, double* sum, int cn);\r
 \r
     template <typename T>\r
-    void sqrSumCaller(const DevMem2D src, PtrStep buf, double* sum, int cn);\r
+    void sqrSumCaller(const DevMem2Db src, PtrStepb buf, double* sum, int cn);\r
 \r
     template <typename T>\r
-    void sqrSumMultipassCaller(const DevMem2D src, PtrStep buf, double* sum, int cn);\r
+    void sqrSumMultipassCaller(const DevMem2Db src, PtrStepb buf, double* sum, int cn);\r
 \r
     namespace sums\r
     {\r
@@ -228,7 +228,7 @@ Scalar cv::gpu::sum(const GpuMat& src, GpuMat& buf)
 {\r
     using namespace mathfunc;\r
 \r
-    typedef void (*Caller)(const DevMem2D, PtrStep, double*, int);\r
+    typedef void (*Caller)(const DevMem2Db, PtrStepb, double*, int);\r
 \r
     static Caller multipass_callers[7] = { \r
             sumMultipassCaller<unsigned char>, sumMultipassCaller<char>, \r
@@ -269,7 +269,7 @@ Scalar cv::gpu::absSum(const GpuMat& src, GpuMat& buf)
 {\r
     using namespace mathfunc;\r
 \r
-    typedef void (*Caller)(const DevMem2D, PtrStep, double*, int);\r
+    typedef void (*Caller)(const DevMem2Db, PtrStepb, double*, int);\r
 \r
     static Caller multipass_callers[7] = { \r
             absSumMultipassCaller<unsigned char>, absSumMultipassCaller<char>, \r
@@ -310,7 +310,7 @@ Scalar cv::gpu::sqrSum(const GpuMat& src, GpuMat& buf)
 {\r
     using namespace mathfunc;\r
 \r
-    typedef void (*Caller)(const DevMem2D, PtrStep, double*, int);\r
+    typedef void (*Caller)(const DevMem2Db, PtrStepb, double*, int);\r
 \r
     static Caller multipass_callers[7] = { \r
             sqrSumMultipassCaller<unsigned char>, sqrSumMultipassCaller<char>, \r
@@ -350,16 +350,16 @@ namespace cv { namespace gpu { namespace mathfunc { namespace minmax {
     void getBufSizeRequired(int cols, int rows, int elem_size, int& bufcols, int& bufrows);\r
     \r
     template <typename T> \r
-    void minMaxCaller(const DevMem2D src, double* minval, double* maxval, PtrStep buf);\r
+    void minMaxCaller(const DevMem2Db src, double* minval, double* maxval, PtrStepb buf);\r
 \r
     template <typename T> \r
-    void minMaxMaskCaller(const DevMem2D src, const PtrStep mask, double* minval, double* maxval, PtrStep buf);\r
+    void minMaxMaskCaller(const DevMem2Db src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf);\r
 \r
     template <typename T> \r
-    void minMaxMultipassCaller(const DevMem2D src, double* minval, double* maxval, PtrStep buf);\r
+    void minMaxMultipassCaller(const DevMem2Db src, double* minval, double* maxval, PtrStepb buf);\r
 \r
     template <typename T> \r
-    void minMaxMaskMultipassCaller(const DevMem2D src, const PtrStep mask, double* minval, double* maxval, PtrStep buf);\r
+    void minMaxMaskMultipassCaller(const DevMem2Db src, const PtrStepb mask, double* minval, double* maxval, PtrStepb buf);\r
 \r
 }}}}\r
 \r
@@ -375,8 +375,8 @@ void cv::gpu::minMax(const GpuMat& src, double* minVal, double* maxVal, const Gp
 {\r
     using namespace mathfunc::minmax;\r
 \r
-    typedef void (*Caller)(const DevMem2D, double*, double*, PtrStep);\r
-    typedef void (*MaskedCaller)(const DevMem2D, const PtrStep, double*, double*, PtrStep);\r
+    typedef void (*Caller)(const DevMem2Db, double*, double*, PtrStepb);\r
+    typedef void (*MaskedCaller)(const DevMem2Db, const PtrStepb, double*, double*, PtrStepb);\r
 \r
     static Caller multipass_callers[7] = { \r
             minMaxMultipassCaller<unsigned char>, minMaxMultipassCaller<char>, \r
@@ -445,20 +445,20 @@ namespace cv { namespace gpu { namespace mathfunc { namespace minmaxloc {
                                int& b1rows, int& b2cols, int& b2rows);\r
 \r
     template <typename T> \r
-    void minMaxLocCaller(const DevMem2D src, double* minval, double* maxval, \r
-                            int minloc[2], int maxloc[2], PtrStep valBuf, PtrStep locBuf);\r
+    void minMaxLocCaller(const DevMem2Db src, double* minval, double* maxval, \r
+                            int minloc[2], int maxloc[2], PtrStepb valBuf, PtrStepb locBuf);\r
 \r
     template <typename T> \r
-    void minMaxLocMaskCaller(const DevMem2D src, const PtrStep mask, double* minval, double* maxval, \r
-                                 int minloc[2], int maxloc[2], PtrStep valBuf, PtrStep locBuf);\r
+    void minMaxLocMaskCaller(const DevMem2Db src, const PtrStepb mask, double* minval, double* maxval, \r
+                                 int minloc[2], int maxloc[2], PtrStepb valBuf, PtrStepb locBuf);\r
 \r
     template <typename T> \r
-    void minMaxLocMultipassCaller(const DevMem2D src, double* minval, double* maxval, \r
-                                     int minloc[2], int maxloc[2], PtrStep valBuf, PtrStep locBuf);\r
+    void minMaxLocMultipassCaller(const DevMem2Db src, double* minval, double* maxval, \r
+                                     int minloc[2], int maxloc[2], PtrStepb valBuf, PtrStepb locBuf);\r
 \r
     template <typename T> \r
-    void minMaxLocMaskMultipassCaller(const DevMem2D src, const PtrStep mask, double* minval, double* maxval, \r
-                                           int minloc[2], int maxloc[2], PtrStep valBuf, PtrStep locBuf);\r
+    void minMaxLocMaskMultipassCaller(const DevMem2Db src, const PtrStepb mask, double* minval, double* maxval, \r
+                                           int minloc[2], int maxloc[2], PtrStepb valBuf, PtrStepb locBuf);\r
 }}}}\r
 \r
 \r
@@ -474,8 +474,8 @@ void cv::gpu::minMaxLoc(const GpuMat& src, double* minVal, double* maxVal, Point
 {\r
     using namespace mathfunc::minmaxloc;\r
 \r
-    typedef void (*Caller)(const DevMem2D, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
-    typedef void (*MaskedCaller)(const DevMem2D, const PtrStep, double*, double*, int[2], int[2], PtrStep, PtrStep);\r
+    typedef void (*Caller)(const DevMem2Db, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
+    typedef void (*MaskedCaller)(const DevMem2Db, const PtrStepb, double*, double*, int[2], int[2], PtrStepb, PtrStepb);\r
 \r
     static Caller multipass_callers[7] = { \r
             minMaxLocMultipassCaller<unsigned char>, minMaxLocMultipassCaller<char>, \r
@@ -549,10 +549,10 @@ namespace cv { namespace gpu { namespace mathfunc { namespace countnonzero {
     void getBufSizeRequired(int cols, int rows, int& bufcols, int& bufrows);\r
 \r
     template <typename T> \r
-    int countNonZeroCaller(const DevMem2D src, PtrStep buf);\r
+    int countNonZeroCaller(const DevMem2Db src, PtrStepb buf);\r
 \r
     template <typename T> \r
-    int countNonZeroMultipassCaller(const DevMem2D src, PtrStep buf);\r
+    int countNonZeroMultipassCaller(const DevMem2Db src, PtrStepb buf);\r
 \r
 }}}}\r
 \r
@@ -568,7 +568,7 @@ int cv::gpu::countNonZero(const GpuMat& src, GpuMat& buf)
 {\r
     using namespace mathfunc::countnonzero;\r
 \r
-    typedef int (*Caller)(const DevMem2D src, PtrStep buf);\r
+    typedef int (*Caller)(const DevMem2Db src, PtrStepb buf);\r
 \r
     static Caller multipass_callers[7] = { \r
             countNonZeroMultipassCaller<unsigned char>, countNonZeroMultipassCaller<char>,\r
@@ -603,8 +603,8 @@ int cv::gpu::countNonZero(const GpuMat& src, GpuMat& buf)
 // reduce\r
 \r
 namespace cv { namespace gpu { namespace mathfunc {\r
-    template <typename T, typename S, typename D> void reduceRows_gpu(const DevMem2D& src, const DevMem2D& dst, int reduceOp, cudaStream_t stream);\r
-    template <typename T, typename S, typename D> void reduceCols_gpu(const DevMem2D& src, int cn, const DevMem2D& dst, int reduceOp, cudaStream_t stream);\r
+    template <typename T, typename S, typename D> void reduceRows_gpu(const DevMem2Db& src, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);\r
+    template <typename T, typename S, typename D> void reduceCols_gpu(const DevMem2Db& src, int cn, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);\r
 }}}\r
 \r
 void cv::gpu::reduce(const GpuMat& src, GpuMat& dst, int dim, int reduceOp, int dtype, Stream& stream)\r
@@ -621,7 +621,7 @@ void cv::gpu::reduce(const GpuMat& src, GpuMat& dst, int dim, int reduceOp, int
 \r
     if (dim == 0)\r
     {\r
-        typedef void (*caller_t)(const DevMem2D& src, const DevMem2D& dst, int reduceOp, cudaStream_t stream);\r
+        typedef void (*caller_t)(const DevMem2Db& src, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);\r
 \r
         static const caller_t callers[6][6] = \r
         {\r
@@ -683,7 +683,7 @@ void cv::gpu::reduce(const GpuMat& src, GpuMat& dst, int dim, int reduceOp, int
     }\r
     else\r
     {\r
-        typedef void (*caller_t)(const DevMem2D& src, int cn, const DevMem2D& dst, int reduceOp, cudaStream_t stream);\r
+        typedef void (*caller_t)(const DevMem2Db& src, int cn, const DevMem2Db& dst, int reduceOp, cudaStream_t stream);\r
 \r
         static const caller_t callers[6][6] = \r
         {\r

diff --git a/modules/gpu/src/opencv2/gpu/device/detail/transform_detail.hpp b/modules/gpu/src/opencv2/gpu/device/detail/transform_detail.hpp
index 96eca2e..212dcbc 100644 (file)
--- a/modules/gpu/src/opencv2/gpu/device/detail/transform_detail.hpp
+++ b/modules/gpu/src/opencv2/gpu/device/detail/transform_detail.hpp
@@ -55,11 +55,11 @@ namespace cv { namespace gpu { namespace device
 \r
         struct MaskReader\r
         {\r
-            explicit MaskReader(const PtrStep& mask_): mask(mask_) {}\r
+            explicit MaskReader(const PtrStepb& mask_): mask(mask_) {}\r
 \r
             __device__ __forceinline__ bool operator()(int y, int x) const { return mask.ptr(y)[x]; }\r
 \r
-            const PtrStep mask;\r
+            const PtrStepb mask;\r
         };\r
 \r
         struct NoMask \r
@@ -219,7 +219,7 @@ namespace cv { namespace gpu { namespace device
         };\r
 \r
         template <typename T, typename D, typename UnOp, typename Mask>\r
-        __global__ static void transformSmart(const DevMem2D_<T> src_, PtrStep_<D> dst_, const Mask mask, const UnOp op)\r
+        __global__ static void transformSmart(const DevMem2D_<T> src_, PtrStep<D> dst_, const Mask mask, const UnOp op)\r
         {\r
             typedef TransformFunctorTraits<UnOp> ft;\r
             typedef typename UnaryReadWriteTraits<T, D, ft::smart_shift>::read_type read_type;\r
@@ -255,7 +255,7 @@ namespace cv { namespace gpu { namespace device
         }\r
 \r
         template <typename T, typename D, typename UnOp, typename Mask>\r
-        static __global__ void transformSimple(const DevMem2D_<T> src, PtrStep_<D> dst, const Mask mask, const UnOp op)\r
+        static __global__ void transformSimple(const DevMem2D_<T> src, PtrStep<D> dst, const Mask mask, const UnOp op)\r
         {\r
                    const int x = blockDim.x * blockIdx.x + threadIdx.x;\r
                    const int y = blockDim.y * blockIdx.y + threadIdx.y;\r
@@ -267,7 +267,7 @@ namespace cv { namespace gpu { namespace device
         }\r
 \r
         template <typename T1, typename T2, typename D, typename BinOp, typename Mask>\r
-        __global__ static void transformSmart(const DevMem2D_<T1> src1_, const PtrStep_<T2> src2_, PtrStep_<D> dst_, \r
+        __global__ static void transformSmart(const DevMem2D_<T1> src1_, const PtrStep<T2> src2_, PtrStep<D> dst_, \r
             const Mask mask, const BinOp op)\r
         {\r
             typedef TransformFunctorTraits<BinOp> ft;\r
@@ -307,7 +307,7 @@ namespace cv { namespace gpu { namespace device
         }\r
 \r
         template <typename T1, typename T2, typename D, typename BinOp, typename Mask>\r
-        static __global__ void transformSimple(const DevMem2D_<T1> src1, const PtrStep_<T2> src2, PtrStep_<D> dst, \r
+        static __global__ void transformSimple(const DevMem2D_<T1> src1, const PtrStep<T2> src2, PtrStep<D> dst, \r
             const Mask mask, const BinOp op)\r
         {\r
                    const int x = blockDim.x * blockIdx.x + threadIdx.x;\r

diff --git a/modules/gpu/src/opencv2/gpu/device/transform.hpp b/modules/gpu/src/opencv2/gpu/device/transform.hpp
index 92d5065..4787292 100644 (file)
--- a/modules/gpu/src/opencv2/gpu/device/transform.hpp
+++ b/modules/gpu/src/opencv2/gpu/device/transform.hpp
@@ -54,7 +54,7 @@ namespace cv { namespace gpu { namespace device
         detail::transform_caller(src, dst, op, WithOutMask(), stream);\r
     }\r
     template <typename T, typename D, typename UnOp>\r
-    void transform(const DevMem2D_<T>& src, const DevMem2D_<D>& dst, const PtrStep& mask, const UnOp& op, cudaStream_t stream = 0)\r
+    void transform(const DevMem2D_<T>& src, const DevMem2D_<D>& dst, const PtrStepb& mask, const UnOp& op, cudaStream_t stream = 0)\r
     {\r
         detail::transform_caller(src, dst, op, SingleMask(mask), stream);\r
     }\r
@@ -65,7 +65,7 @@ namespace cv { namespace gpu { namespace device
         detail::transform_caller(src1, src2, dst, op, WithOutMask(), stream);\r
     }\r
     template <typename T1, typename T2, typename D, typename BinOp>\r
-    void transform(const DevMem2D_<T1>& src1, const DevMem2D_<T2>& src2, const DevMem2D_<D>& dst, const PtrStep& mask, const BinOp& op, cudaStream_t stream = 0)\r
+    void transform(const DevMem2D_<T1>& src1, const DevMem2D_<T2>& src2, const DevMem2D_<D>& dst, const PtrStepb& mask, const BinOp& op, cudaStream_t stream = 0)\r
     {\r
         detail::transform_caller(src1, src2, dst, op, SingleMask(mask), stream);\r
     }\r

diff --git a/modules/gpu/src/opencv2/gpu/device/utility.hpp b/modules/gpu/src/opencv2/gpu/device/utility.hpp
index 678470d..aaa8052 100644 (file)
--- a/modules/gpu/src/opencv2/gpu/device/utility.hpp
+++ b/modules/gpu/src/opencv2/gpu/device/utility.hpp
@@ -70,19 +70,19 @@ namespace cv {  namespace gpu { namespace device
 \r
     struct SingleMask\r
     {\r
-        explicit __host__ __device__ __forceinline__ SingleMask(const PtrStep& mask_) : mask(mask_) {}\r
+        explicit __host__ __device__ __forceinline__ SingleMask(const PtrStepb& mask_) : mask(mask_) {}\r
         \r
         __device__ __forceinline__ bool operator()(int y, int x) const\r
         {            \r
             return mask.ptr(y)[x] != 0;\r
         }\r
 \r
-        const PtrStep mask;\r
+        const PtrStepb mask;\r
     };\r
 \r
     struct MaskCollection\r
     {\r
-        explicit __host__ __device__ __forceinline__ MaskCollection(PtrStep* maskCollection_) : maskCollection(maskCollection_) {}\r
+        explicit __host__ __device__ __forceinline__ MaskCollection(PtrStepb* maskCollection_) : maskCollection(maskCollection_) {}\r
 \r
         __device__ __forceinline__ void next()\r
         {\r
@@ -99,8 +99,8 @@ namespace cv {  namespace gpu { namespace device
             return curMask.data == 0 || (ForceGlob<uchar>::Load(curMask.ptr(y), x, val), (val != 0));\r
         }\r
 \r
-        const PtrStep* maskCollection;\r
-        PtrStep curMask;\r
+        const PtrStepb* maskCollection;\r
+        PtrStepb curMask;\r
     };\r
 \r
     struct WithOutMask\r

diff --git a/modules/gpu/src/split_merge.cpp b/modules/gpu/src/split_merge.cpp
index a8785d4..a791862 100644 (file)
--- a/modules/gpu/src/split_merge.cpp
+++ b/modules/gpu/src/split_merge.cpp
@@ -55,11 +55,11 @@ void cv::gpu::split(const GpuMat& /*src*/, vector<GpuMat>& /*dst*/, Stream& /*st
 \r
 namespace cv { namespace gpu { namespace split_merge \r
 {    \r
-    extern "C" void merge_caller(const DevMem2D* src, DevMem2D& dst, \r
+    extern "C" void merge_caller(const DevMem2Db* src, DevMem2Db& dst, \r
                                  int total_channels, size_t elem_size, \r
                                  const cudaStream_t& stream);\r
 \r
-    extern "C" void split_caller(const DevMem2D& src, DevMem2D* dst, \r
+    extern "C" void split_caller(const DevMem2Db& src, DevMem2Db* dst, \r
                                  int num_channels, size_t elem_size1, \r
                                  const cudaStream_t& stream);\r
 \r
@@ -95,11 +95,11 @@ namespace cv { namespace gpu { namespace split_merge
         {\r
             dst.create(size, CV_MAKETYPE(depth, total_channels));\r
 \r
-            DevMem2D src_as_devmem[4];\r
+            DevMem2Db src_as_devmem[4];\r
             for(size_t i = 0; i < n; ++i)\r
                 src_as_devmem[i] = src[i];\r
 \r
-            DevMem2D dst_as_devmem(dst);\r
+            DevMem2Db dst_as_devmem(dst);\r
             split_merge::merge_caller(src_as_devmem, dst_as_devmem,\r
                                       total_channels, CV_ELEM_SIZE(depth),\r
                                       stream);\r
@@ -130,11 +130,11 @@ namespace cv { namespace gpu { namespace split_merge
 \r
         CV_Assert(num_channels <= 4);\r
 \r
-        DevMem2D dst_as_devmem[4];\r
+        DevMem2Db dst_as_devmem[4];\r
         for (int i = 0; i < num_channels; ++i)\r
             dst_as_devmem[i] = dst[i];\r
 \r
-        DevMem2D src_as_devmem(src);\r
+        DevMem2Db src_as_devmem(src);\r
         split_merge::split_caller(src_as_devmem, dst_as_devmem,\r
                                   num_channels, src.elemSize1(), \r
                                   stream);\r

diff --git a/modules/gpu/src/stereobm.cpp b/modules/gpu/src/stereobm.cpp
index f29758b..8ab2d29 100644 (file)
--- a/modules/gpu/src/stereobm.cpp
+++ b/modules/gpu/src/stereobm.cpp
@@ -59,10 +59,10 @@ namespace cv { namespace gpu
 {\r
     namespace bm\r
     {\r
-        //extern "C" void stereoBM_GPU(const DevMem2D& left, const DevMem2D& right, const DevMem2D& disp, int ndisp, int winsz, const DevMem2D_<uint>& minSSD_buf);\r
-        extern "C" void stereoBM_GPU(const DevMem2D& left, const DevMem2D& right, const DevMem2D& disp, int ndisp, int winsz, const DevMem2D_<uint>& minSSD_buf, cudaStream_t & stream);\r
-        extern "C" void prefilter_xsobel(const DevMem2D& input, const DevMem2D& output, int prefilterCap /*= 31*/, cudaStream_t & stream);\r
-        extern "C" void postfilter_textureness(const DevMem2D& input, int winsz, float avgTexturenessThreshold, const DevMem2D& disp, cudaStream_t & stream);\r
+        //extern "C" void stereoBM_GPU(const DevMem2Db& left, const DevMem2Db& right, const DevMem2Db& disp, int ndisp, int winsz, const DevMem2D_<uint>& minSSD_buf);\r
+        extern "C" void stereoBM_GPU(const DevMem2Db& left, const DevMem2Db& right, const DevMem2Db& disp, int ndisp, int winsz, const DevMem2D_<uint>& minSSD_buf, cudaStream_t & stream);\r
+        extern "C" void prefilter_xsobel(const DevMem2Db& input, const DevMem2Db& output, int prefilterCap /*= 31*/, cudaStream_t & stream);\r
+        extern "C" void postfilter_textureness(const DevMem2Db& input, int winsz, float avgTexturenessThreshold, const DevMem2Db& disp, cudaStream_t & stream);\r
     }\r
 }}\r
 \r

diff --git a/modules/gpu/src/stereobp.cpp b/modules/gpu/src/stereobp.cpp
index ea151fc..288e2f8 100644 (file)
--- a/modules/gpu/src/stereobp.cpp
+++ b/modules/gpu/src/stereobp.cpp
@@ -63,16 +63,16 @@ namespace cv { namespace gpu { namespace bp
 {\r
     void load_constants(int ndisp, float max_data_term, float data_weight, float max_disc_term, float disc_single_jump);\r
     template<typename T, typename D>\r
-    void comp_data_gpu(const DevMem2D& left, const DevMem2D& right, const DevMem2D& data, cudaStream_t stream);\r
+    void comp_data_gpu(const DevMem2Db& left, const DevMem2Db& right, const DevMem2Db& data, cudaStream_t stream);\r
     template<typename T>\r
-    void data_step_down_gpu(int dst_cols, int dst_rows, int src_rows, const DevMem2D& src, const DevMem2D& dst, cudaStream_t stream);\r
+    void data_step_down_gpu(int dst_cols, int dst_rows, int src_rows, const DevMem2Db& src, const DevMem2Db& dst, cudaStream_t stream);\r
     template <typename T>\r
-    void level_up_messages_gpu(int dst_idx, int dst_cols, int dst_rows, int src_rows, DevMem2D* mus, DevMem2D* mds, DevMem2D* mls, DevMem2D* mrs, cudaStream_t stream);\r
+    void level_up_messages_gpu(int dst_idx, int dst_cols, int dst_rows, int src_rows, DevMem2Db* mus, DevMem2Db* mds, DevMem2Db* mls, DevMem2Db* mrs, cudaStream_t stream);\r
     template <typename T>\r
-    void calc_all_iterations_gpu(int cols, int rows, int iters, const DevMem2D& u, const DevMem2D& d, \r
-        const DevMem2D& l, const DevMem2D& r, const DevMem2D& data, cudaStream_t stream);\r
+    void calc_all_iterations_gpu(int cols, int rows, int iters, const DevMem2Db& u, const DevMem2Db& d, \r
+        const DevMem2Db& l, const DevMem2Db& r, const DevMem2Db& data, cudaStream_t stream);\r
     template <typename T>\r
-    void output_gpu(const DevMem2D& u, const DevMem2D& d, const DevMem2D& l, const DevMem2D& r, const DevMem2D& data, \r
+    void output_gpu(const DevMem2Db& u, const DevMem2Db& d, const DevMem2Db& l, const DevMem2Db& r, const DevMem2Db& data, \r
         const DevMem2D_<short>& disp, cudaStream_t stream);\r
 }}}\r
 \r
@@ -133,7 +133,7 @@ namespace
 \r
         void operator()(const GpuMat& left, const GpuMat& right, GpuMat& disp, Stream& stream)\r
         {\r
-            typedef void (*comp_data_t)(const DevMem2D& left, const DevMem2D& right, const DevMem2D& data, cudaStream_t stream);\r
+            typedef void (*comp_data_t)(const DevMem2Db& left, const DevMem2Db& right, const DevMem2Db& data, cudaStream_t stream);\r
             static const comp_data_t comp_data_callers[2][5] = \r
             {\r
                 {0, bp::comp_data_gpu<unsigned char, short>, 0, bp::comp_data_gpu<uchar3, short>, bp::comp_data_gpu<uchar4, short>},\r
@@ -251,25 +251,25 @@ namespace
         {\r
             using namespace cv::gpu::bp;\r
 \r
-            typedef void (*data_step_down_t)(int dst_cols, int dst_rows, int src_rows, const DevMem2D& src, const DevMem2D& dst, cudaStream_t stream);\r
+            typedef void (*data_step_down_t)(int dst_cols, int dst_rows, int src_rows, const DevMem2Db& src, const DevMem2Db& dst, cudaStream_t stream);\r
             static const data_step_down_t data_step_down_callers[2] = \r
             {\r
                 data_step_down_gpu<short>, data_step_down_gpu<float>\r
             };\r
             \r
-            typedef void (*level_up_messages_t)(int dst_idx, int dst_cols, int dst_rows, int src_rows, DevMem2D* mus, DevMem2D* mds, DevMem2D* mls, DevMem2D* mrs, cudaStream_t stream);\r
+            typedef void (*level_up_messages_t)(int dst_idx, int dst_cols, int dst_rows, int src_rows, DevMem2Db* mus, DevMem2Db* mds, DevMem2Db* mls, DevMem2Db* mrs, cudaStream_t stream);\r
             static const level_up_messages_t level_up_messages_callers[2] = \r
             {\r
                 level_up_messages_gpu<short>, level_up_messages_gpu<float>\r
             };\r
 \r
-            typedef void (*calc_all_iterations_t)(int cols, int rows, int iters, const DevMem2D& u, const DevMem2D& d, const DevMem2D& l, const DevMem2D& r, const DevMem2D& data, cudaStream_t stream);\r
+            typedef void (*calc_all_iterations_t)(int cols, int rows, int iters, const DevMem2Db& u, const DevMem2Db& d, const DevMem2Db& l, const DevMem2Db& r, const DevMem2Db& data, cudaStream_t stream);\r
             static const calc_all_iterations_t calc_all_iterations_callers[2] = \r
             {\r
                 calc_all_iterations_gpu<short>, calc_all_iterations_gpu<float>\r
             };\r
 \r
-            typedef void (*output_t)(const DevMem2D& u, const DevMem2D& d, const DevMem2D& l, const DevMem2D& r, const DevMem2D& data, const DevMem2D_<short>& disp, cudaStream_t stream);\r
+            typedef void (*output_t)(const DevMem2Db& u, const DevMem2Db& d, const DevMem2Db& l, const DevMem2Db& r, const DevMem2Db& data, const DevMem2D_<short>& disp, cudaStream_t stream);\r
             static const output_t output_callers[2] = \r
             {\r
                 output_gpu<short>, output_gpu<float>\r
@@ -289,10 +289,10 @@ namespace
                 data_step_down_callers[funcIdx](cols_all[i], rows_all[i], rows_all[i-1], datas[i-1], datas[i], cudaStream);\r
             }\r
 \r
-            DevMem2D mus[] = {u, u2};\r
-            DevMem2D mds[] = {d, d2};\r
-            DevMem2D mrs[] = {r, r2};\r
-            DevMem2D mls[] = {l, l2};\r
+            DevMem2Db mus[] = {u, u2};\r
+            DevMem2Db mds[] = {d, d2};\r
+            DevMem2Db mrs[] = {r, r2};\r
+            DevMem2Db mls[] = {l, l2};\r
 \r
             int mem_idx = (rthis.levels & 1) ? 0 : 1;\r
 \r

diff --git a/modules/gpu/src/stereocsbp.cpp b/modules/gpu/src/stereocsbp.cpp
index d026ebb..85faaf8 100644 (file)
--- a/modules/gpu/src/stereocsbp.cpp
+++ b/modules/gpu/src/stereocsbp.cpp
@@ -60,7 +60,7 @@ void cv::gpu::StereoConstantSpaceBP::operator()(const GpuMat&, const GpuMat&, Gp
 namespace cv { namespace gpu { namespace csbp\r
 {\r
     void load_constants(int ndisp, float max_data_term, float data_weight, float max_disc_term, float disc_single_jump, int min_disp_th,\r
-        const DevMem2D& left, const DevMem2D& right, const DevMem2D& temp);\r
+        const DevMem2Db& left, const DevMem2Db& right, const DevMem2Db& temp);\r
 \r
     template<class T>\r
     void init_data_cost(int rows, int cols, T* disp_selected_pyr, T* data_cost_selected, size_t msg_step,\r

diff --git a/modules/gpu/src/surf.cpp b/modules/gpu/src/surf.cpp
index 1d50fb2..dba9696 100644 (file)
--- a/modules/gpu/src/surf.cpp
+++ b/modules/gpu/src/surf.cpp
@@ -143,7 +143,7 @@ namespace
             uploadConstant("cv::gpu::surf::c_nOctaveLayers",     nOctaveLayers);\r
             uploadConstant("cv::gpu::surf::c_hessianThreshold",  static_cast<float>(hessianThreshold));\r
 \r
-            imgTex.bind("cv::gpu::surf::imgTex", (DevMem2D)img);\r
+            imgTex.bind("cv::gpu::surf::imgTex", (DevMem2Db)img);\r
 \r
             integralBuffered(img, sum, intBuffer);\r
             sumTex.bind("cv::gpu::surf::sumTex", (DevMem2D_<unsigned int>)sum);\r

diff --git a/modules/objdetect/include/opencv2/objdetect/objdetect.hpp b/modules/objdetect/include/opencv2/objdetect/objdetect.hpp
index 49d4237..a8dedb1 100644 (file)
--- a/modules/objdetect/include/opencv2/objdetect/objdetect.hpp
+++ b/modules/objdetect/include/opencv2/objdetect/objdetect.hpp
@@ -474,7 +474,7 @@ public:
     {
         public:
             virtual cv::Mat generateMask(const cv::Mat& src)=0;
-            virtual void initializeMask(const cv::Mat& src) {};
+            virtual void initializeMask(const cv::Mat& /*src*/) {};
     };
     void setMaskGenerator(Ptr<MaskGenerator> maskGenerator);
     Ptr<MaskGenerator> getMaskGenerator();