added vecmath to gpu module.

modules/gpu/src/cuda/saturate_cast.hpp

@@ -163,92 +163,6 @@ namespace cv
             return saturate_cast<uint>((float)v);
         #endif
         }
-
-        template<typename _Tp> static __device__ _Tp saturate_cast(uchar4 v) { return _Tp(v); }
-        template<typename _Tp> static __device__ _Tp saturate_cast(char4 v) { return _Tp(v); }
-        template<typename _Tp> static __device__ _Tp saturate_cast(ushort4 v) { return _Tp(v); }
-        template<typename _Tp> static __device__ _Tp saturate_cast(short4 v) { return _Tp(v); }
-        template<typename _Tp> static __device__ _Tp saturate_cast(uint4 v) { return _Tp(v); }
-        template<typename _Tp> static __device__ _Tp saturate_cast(int4 v) { return _Tp(v); }
-        template<typename _Tp> static __device__ _Tp saturate_cast(float4 v) { return _Tp(v); }
-
-        template<> static __device__ uchar4 saturate_cast<uchar4>(char4 v)
-        { return make_uchar4(saturate_cast<uchar>(v.x), saturate_cast<uchar>(v.y), saturate_cast<uchar>(v.z), saturate_cast<uchar>(v.w)); }
-        template<> static __device__ uchar4 saturate_cast<uchar4>(ushort4 v)
-        { return make_uchar4(saturate_cast<uchar>(v.x), saturate_cast<uchar>(v.y), saturate_cast<uchar>(v.z), saturate_cast<uchar>(v.w)); }
-        template<> static __device__ uchar4 saturate_cast<uchar4>(short4 v)
-        { return make_uchar4(saturate_cast<uchar>(v.x), saturate_cast<uchar>(v.y), saturate_cast<uchar>(v.z), saturate_cast<uchar>(v.w)); }
-        template<> static __device__ uchar4 saturate_cast<uchar4>(uint4 v)
-        { return make_uchar4(saturate_cast<uchar>(v.x), saturate_cast<uchar>(v.y), saturate_cast<uchar>(v.z), saturate_cast<uchar>(v.w)); }
-        template<> static __device__ uchar4 saturate_cast<uchar4>(int4 v)
-        { return make_uchar4(saturate_cast<uchar>(v.x), saturate_cast<uchar>(v.y), saturate_cast<uchar>(v.z), saturate_cast<uchar>(v.w)); }
-        template<> static __device__ uchar4 saturate_cast<uchar4>(float4 v)
-        { return make_uchar4(saturate_cast<uchar>(v.x), saturate_cast<uchar>(v.y), saturate_cast<uchar>(v.z), saturate_cast<uchar>(v.w)); }
-
-        template<> static __device__ char4 saturate_cast<char4>(uchar4 v)
-        { return make_char4(saturate_cast<char>(v.x), saturate_cast<char>(v.y), saturate_cast<char>(v.z), saturate_cast<char>(v.w)); }
-        template<> static __device__ char4 saturate_cast<char4>(ushort4 v)
-        { return make_char4(saturate_cast<char>(v.x), saturate_cast<char>(v.y), saturate_cast<char>(v.z), saturate_cast<char>(v.w)); }
-        template<> static __device__ char4 saturate_cast<char4>(short4 v)
-        { return make_char4(saturate_cast<char>(v.x), saturate_cast<char>(v.y), saturate_cast<char>(v.z), saturate_cast<char>(v.w)); }
-        template<> static __device__ char4 saturate_cast<char4>(uint4 v)
-        { return make_char4(saturate_cast<char>(v.x), saturate_cast<char>(v.y), saturate_cast<char>(v.z), saturate_cast<char>(v.w)); }
-        template<> static __device__ char4 saturate_cast<char4>(int4 v)
-        { return make_char4(saturate_cast<char>(v.x), saturate_cast<char>(v.y), saturate_cast<char>(v.z), saturate_cast<char>(v.w)); }
-        template<> static __device__ char4 saturate_cast<char4>(float4 v)
-        { return make_char4(saturate_cast<char>(v.x), saturate_cast<char>(v.y), saturate_cast<char>(v.z), saturate_cast<char>(v.w)); }
-
-        template<> static __device__ ushort4 saturate_cast<ushort4>(uchar4 v)
-        { return make_ushort4(v.x, v.y, v.z, v.w); }
-        template<> static __device__ ushort4 saturate_cast<ushort4>(char4 v)
-        { return make_ushort4(saturate_cast<ushort>(v.x), saturate_cast<ushort>(v.y), saturate_cast<ushort>(v.z), saturate_cast<ushort>(v.w)); }
-        template<> static __device__ ushort4 saturate_cast<ushort4>(short4 v)
-        { return make_ushort4(saturate_cast<ushort>(v.x), saturate_cast<ushort>(v.y), saturate_cast<ushort>(v.z), saturate_cast<ushort>(v.w)); }
-        template<> static __device__ ushort4 saturate_cast<ushort4>(uint4 v)
-        { return make_ushort4(saturate_cast<ushort>(v.x), saturate_cast<ushort>(v.y), saturate_cast<ushort>(v.z), saturate_cast<ushort>(v.w)); }
-        template<> static __device__ ushort4 saturate_cast<ushort4>(int4 v)
-        { return make_ushort4(saturate_cast<ushort>(v.x), saturate_cast<ushort>(v.y), saturate_cast<ushort>(v.z), saturate_cast<ushort>(v.w)); }
-        template<> static __device__ ushort4 saturate_cast<ushort4>(float4 v)
-        { return make_ushort4(saturate_cast<ushort>(v.x), saturate_cast<ushort>(v.y), saturate_cast<ushort>(v.z), saturate_cast<ushort>(v.w)); }
-
-        template<> static __device__ short4 saturate_cast<short4>(uchar4 v)
-        { return make_short4(v.x, v.y, v.z, v.w); }
-        template<> static __device__ short4 saturate_cast<short4>(char4 v)
-        { return make_short4(v.x, v.y, v.z, v.w); }
-        template<> static __device__ short4 saturate_cast<short4>(ushort4 v)
-        { return make_short4(saturate_cast<short>(v.x), saturate_cast<short>(v.y), saturate_cast<short>(v.z), saturate_cast<short>(v.w)); }
-        template<> static __device__ short4 saturate_cast<short4>(uint4 v)
-        { return make_short4(saturate_cast<short>(v.x), saturate_cast<short>(v.y), saturate_cast<short>(v.z), saturate_cast<short>(v.w)); }
-        template<> static __device__ short4 saturate_cast<short4>(int4 v)
-        { return make_short4(saturate_cast<short>(v.x), saturate_cast<short>(v.y), saturate_cast<short>(v.z), saturate_cast<short>(v.w)); }
-        template<> static __device__ short4 saturate_cast<short4>(float4 v)
-        { return make_short4(saturate_cast<short>(v.x), saturate_cast<short>(v.y), saturate_cast<short>(v.z), saturate_cast<short>(v.w)); }
-        
-        template<> static __device__ uint4 saturate_cast<uint4>(uchar4 v)
-        { return make_uint4(v.x, v.y, v.z, v.w); }
-        template<> static __device__ uint4 saturate_cast<uint4>(char4 v)
-        { return make_uint4(saturate_cast<uint>(v.x), saturate_cast<uint>(v.y), saturate_cast<uint>(v.z), saturate_cast<uint>(v.w)); }
-        template<> static __device__ uint4 saturate_cast<uint4>(ushort4 v)
-        { return make_uint4(v.x, v.y, v.z, v.w); }
-        template<> static __device__ uint4 saturate_cast<uint4>(short4 v)
-        { return make_uint4(saturate_cast<uint>(v.x), saturate_cast<uint>(v.y), saturate_cast<uint>(v.z), saturate_cast<uint>(v.w)); }
-        template<> static __device__ uint4 saturate_cast<uint4>(int4 v)
-        { return make_uint4(saturate_cast<uint>(v.x), saturate_cast<uint>(v.y), saturate_cast<uint>(v.z), saturate_cast<uint>(v.w)); }
-        template<> static __device__ uint4 saturate_cast<uint4>(float4 v)
-        { return make_uint4(saturate_cast<uint>(v.x), saturate_cast<uint>(v.y), saturate_cast<uint>(v.z), saturate_cast<uint>(v.w)); }
-        
-        template<> static __device__ int4 saturate_cast<int4>(uchar4 v)
-        { return make_int4(v.x, v.y, v.z, v.w); }
-        template<> static __device__ int4 saturate_cast<int4>(char4 v)
-        { return make_int4(v.x, v.y, v.z, v.w); }
-        template<> static __device__ int4 saturate_cast<int4>(ushort4 v)
-        { return make_int4(v.x, v.y, v.z, v.w); }
-        template<> static __device__ int4 saturate_cast<int4>(short4 v)
-        { return make_int4(v.x, v.y, v.z, v.w); }
-        template<> static __device__ int4 saturate_cast<int4>(uint4 v)
-        { return make_int4(saturate_cast<int>(v.x), saturate_cast<int>(v.y), saturate_cast<int>(v.z), saturate_cast<int>(v.w)); }
-        template<> static __device__ int4 saturate_cast<int4>(float4 v)
-        { return make_int4(saturate_cast<int>(v.x), saturate_cast<int>(v.y), saturate_cast<int>(v.z), saturate_cast<int>(v.w)); }
     }
 }
 

modules/gpu/src/cudastream.cpp

@@ -163,7 +163,7 @@ void cv::gpu::Stream::waitForCompletion() { cudaSafeCall( cudaStreamSynchronize(
 void cv::gpu::Stream::enqueueDownload(const GpuMat& src, Mat& dst)
 {
     // if not -> allocation will be done, but after that dst will not point to page locked memory
-    CV_Assert(src.cols == dst.cols && src.rows == dst.rows && src.type() == dst.type() )
+    CV_Assert(src.cols == dst.cols && src.rows == dst.rows && src.type() == dst.type() );
     devcopy(src, dst, impl->stream, cudaMemcpyDeviceToHost);
 }
 void cv::gpu::Stream::enqueueDownload(const GpuMat& src, CudaMem& dst) { devcopy(src, dst, impl->stream, cudaMemcpyDeviceToHost); }

modules/gpu/src/imgproc_gpu.cpp

@@ -652,7 +652,7 @@ void cv::gpu::cvtColor(const GpuMat& src, GpuMat& dst, int code, int dcn, const
 
 double cv::gpu::threshold(const GpuMat& src, GpuMat& dst, double thresh) 
 { 
-    CV_Assert(src.type() == CV_32FC1)
+    CV_Assert(src.type() == CV_32FC1);
 
     dst.create( src.size(), src.type() );
 

tests/gpu/src/filters.cpp

@@ -166,6 +166,8 @@ struct CV_GpuNppImageSobelTest : public CV_GpuNppFilterTest
 
     int test(const Mat& img)
     {
+        if (img.type() != CV_8UC1)
+            return CvTS::OK;
         int ksizes[] = {3, 5, 7};
         int ksizes_num = sizeof(ksizes) / sizeof(int);
 
@@ -181,8 +183,10 @@ struct CV_GpuNppImageSobelTest : public CV_GpuNppFilterTest
             cv::Sobel(img, cpudst, -1, dx, dy, ksizes[i]);
 
             GpuMat gpu1(img);
+            gpu1.convertTo(gpu1, CV_32S);
             GpuMat gpudst;
             cv::gpu::Sobel(gpu1, gpudst, -1, dx, dy, ksizes[i]);
+            gpudst.convertTo(gpudst, CV_8U);
 
             if (CheckNorm(cpudst, gpudst, Size(ksizes[i], ksizes[i])) != CvTS::OK)
                 test_res = CvTS::FAIL_GENERIC;
@@ -200,15 +204,20 @@ struct CV_GpuNppImageScharrTest : public CV_GpuNppFilterTest
 
     int test(const Mat& img)
     {
+        if (img.type() != CV_8UC1)
+            return CvTS::OK;
+
         int dx = 1, dy = 0;
 
         Mat cpudst;
         cv::Scharr(img, cpudst, -1, dx, dy);
 
         GpuMat gpu1(img);
+        gpu1.convertTo(gpu1, CV_32S);
         GpuMat gpudst;
         cv::gpu::Scharr(gpu1, gpudst, -1, dx, dy);
-
+        gpudst.convertTo(gpudst, CV_8U);
+        
         return CheckNorm(cpudst, gpudst, Size(3, 3));
     }
 };