Merge pull request #3800 from vrabaud:cuda

Get CUDA code to compile with clang CUDA and without CUDA #3800

Changelist:
- there are some syntactic changes: `<< <` -> `<<<`. For some reason, I do not need to change all those in the code.
-  `::min` -> `std::min` in `__host__` code
- `modules/cudaimgproc/src/moments.cpp` needs to have the CUDA code in the `#ifdef`
- The signature of `cv::cuda::swapChannels` is not exactly the same as the C++ one in `modules/cudaimgproc/src/color.cpp` 
- `cv::cuda::FarnebackOpticalFlow::create` needs to be explicit about which FarnebackOpticalFlow it returns

### Pull Request Readiness Checklist

See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request

- [x] I agree to contribute to the project under Apache 2 License.
- [x] To the best of my knowledge, the proposed patch is not based on a code under GPL or another license that is incompatible with OpenCV
- [x] The PR is proposed to the proper branch

modules/cudaarithm/src/arithm.cpp

@@ -54,6 +54,8 @@ void cv::cuda::mulAndScaleSpectrums(InputArray, InputArray, OutputArray, int, fl
 
 void cv::cuda::dft(InputArray, OutputArray, Size, int, Stream&) { throw_no_cuda(); }
 
+Ptr<DFT> cv::cuda::createDFT(Size, int) { throw_no_cuda(); return Ptr<DFT>(); }
+
 Ptr<Convolution> cv::cuda::createConvolution(Size) { throw_no_cuda(); return Ptr<Convolution>(); }
 
 #else /* !defined (HAVE_CUDA) */

modules/cudaarithm/src/element_operations.cpp

@@ -84,8 +84,13 @@ void cv::cuda::magnitude(InputArray, InputArray, OutputArray, Stream&) { throw_n
 void cv::cuda::magnitudeSqr(InputArray, OutputArray, Stream&) { throw_no_cuda(); }
 void cv::cuda::magnitudeSqr(InputArray, InputArray, OutputArray, Stream&) { throw_no_cuda(); }
 void cv::cuda::phase(InputArray, InputArray, OutputArray, bool, Stream&) { throw_no_cuda(); }
+void cv::cuda::phase(InputArray, OutputArray, bool, Stream&) { throw_no_cuda(); }
 void cv::cuda::cartToPolar(InputArray, InputArray, OutputArray, OutputArray, bool, Stream&) { throw_no_cuda(); }
+void cv::cuda::cartToPolar(InputArray, OutputArray, OutputArray, bool, Stream&) { throw_no_cuda(); }
+void cv::cuda::cartToPolar(InputArray, OutputArray, bool, Stream&) { throw_no_cuda(); }
 void cv::cuda::polarToCart(InputArray, InputArray, OutputArray, OutputArray, bool, Stream&) { throw_no_cuda(); }
+void cv::cuda::polarToCart(InputArray, InputArray, OutputArray, bool, Stream&) { throw_no_cuda(); }
+void cv::cuda::polarToCart(InputArray, OutputArray, bool, Stream&) { throw_no_cuda(); }
 
 #else
 

modules/cudaarithm/src/reductions.cpp

@@ -69,8 +69,10 @@ void cv::cuda::countNonZero(InputArray, OutputArray, Stream&) { throw_no_cuda();
 
 void cv::cuda::reduce(InputArray, OutputArray, int, int, int, Stream&) { throw_no_cuda(); }
 
-void cv::cuda::meanStdDev(InputArray, Scalar&, Scalar&) { throw_no_cuda(); }
+void cv::cuda::meanStdDev(InputArray, OutputArray, InputArray, Stream&) { throw_no_cuda(); }
 void cv::cuda::meanStdDev(InputArray, OutputArray, Stream&) { throw_no_cuda(); }
+void cv::cuda::meanStdDev(InputArray, Scalar&, Scalar&, InputArray) { throw_no_cuda(); }
+void cv::cuda::meanStdDev(InputArray, Scalar&, Scalar&) { throw_no_cuda(); }
 
 void cv::cuda::rectStdDev(InputArray, InputArray, OutputArray, Rect, Stream&) { throw_no_cuda(); }
 

modules/cudaimgproc/src/color.cpp

@@ -51,7 +51,7 @@ void cv::cuda::cvtColor(InputArray, OutputArray, int, int, Stream&) { throw_no_c
 
 void cv::cuda::demosaicing(InputArray, OutputArray, int, int, Stream&) { throw_no_cuda(); }
 
-void cv::cuda::swapChannels(InputOutputArray, const int[], Stream&) { throw_no_cuda(); }
+void cv::cuda::swapChannels(InputOutputArray, const int[4], Stream&) { throw_no_cuda(); }
 
 void cv::cuda::gammaCorrection(InputArray, OutputArray, bool, Stream&) { throw_no_cuda(); }
 

modules/cudaimgproc/src/connectedcomponents.cpp

@@ -9,8 +9,8 @@ using namespace cv::cuda;
 
 #if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
 
-void cv::cuda::connectedComponents(InputArray img_, OutputArray labels_, int connectivity,
-    int ltype, ConnectedComponentsAlgorithmsTypes ccltype) { throw_no_cuda(); }
+void cv::cuda::connectedComponents(InputArray, OutputArray, int, int, ConnectedComponentsAlgorithmsTypes) { throw_no_cuda(); }
+void cv::cuda::connectedComponents(InputArray, OutputArray, int, int) { throw_no_cuda(); }
 
 #else /* !defined (HAVE_CUDA) */
 

modules/cudaimgproc/src/cuda/canny.cu

@@ -428,7 +428,7 @@ namespace canny
             cudaSafeCall( cudaMemsetAsync(d_counter, 0, sizeof(int), stream) );
 
             const dim3 block(128);
-            const dim3 grid(::min(count, 65535u), divUp(count, 65535), 1);
+            const dim3 grid(std::min(count, 65535), divUp(count, 65535), 1);
 
             edgesHysteresisGlobalKernel<<<grid, block, 0, stream>>>(map, st1, st2, d_counter, count);
             cudaSafeCall( cudaGetLastError() );
@@ -439,7 +439,7 @@ namespace canny
             cudaSafeCall( cudaMemcpyAsync(&count, d_counter, sizeof(int), cudaMemcpyDeviceToHost, stream) );
             cudaSafeCall( cudaStreamSynchronize(stream) );
 
-            count = min(count, map.cols * map.rows);
+            count = std::min(count, map.cols * map.rows);
 
             //std::swap(st1, st2);
             short2* tmp = st1;

modules/cudaimgproc/src/cuda/generalized_hough.cu

@@ -302,7 +302,7 @@ namespace cv { namespace cuda { namespace device
             int totalCount;
             cudaSafeCall( cudaMemcpy(&totalCount, counterPtr, sizeof(int), cudaMemcpyDeviceToHost) );
 
-            totalCount = ::min(totalCount, maxSize);
+            totalCount = std::min(totalCount, maxSize);
 
             return totalCount;
         }
@@ -812,7 +812,7 @@ namespace cv { namespace cuda { namespace device
             int totalCount;
             cudaSafeCall( cudaMemcpy(&totalCount, counterPtr, sizeof(int), cudaMemcpyDeviceToHost) );
 
-            totalCount = ::min(totalCount, maxSize);
+            totalCount = std::min(totalCount, maxSize);
 
             return totalCount;
         }

modules/cudaimgproc/src/cuda/hough_circles.cu

@@ -238,7 +238,7 @@ namespace cv { namespace cuda { namespace device
             cudaSafeCall( cudaMemcpyAsync(&totalCount, counterPtr, sizeof(int), cudaMemcpyDeviceToHost, stream) );
             cudaSafeCall( cudaStreamSynchronize(stream) );
 
-            totalCount = ::min(totalCount, maxCircles);
+            totalCount = std::min(totalCount, maxCircles);
 
             return totalCount;
         }

modules/cudaimgproc/src/cuda/hough_lines.cu

@@ -189,7 +189,7 @@ namespace cv { namespace cuda { namespace device
 
             cudaSafeCall( cudaStreamSynchronize(stream) );
 
-            totalCount = ::min(totalCount, maxSize);
+            totalCount = std::min(totalCount, maxSize);
 
             if (doSort && totalCount > 0)
             {

modules/cudaimgproc/src/cuda/hough_segments.cu

@@ -241,7 +241,7 @@ namespace cv { namespace cuda { namespace device
 
             cudaSafeCall( cudaStreamSynchronize(stream) );
 
-            totalCount = ::min(totalCount, maxSize);
+            totalCount = std::min(totalCount, maxSize);
             return totalCount;
         }
     }

modules/cudaimgproc/src/histogram.cpp

@@ -48,6 +48,7 @@ using namespace cv::cuda;
 #if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
 
 void cv::cuda::calcHist(InputArray, OutputArray, Stream&) { throw_no_cuda(); }
+void cv::cuda::calcHist(InputArray, InputArray, OutputArray, Stream&) { throw_no_cuda(); }
 
 void cv::cuda::equalizeHist(InputArray, OutputArray, Stream&) { throw_no_cuda(); }
 

modules/cudaimgproc/src/moments.cpp

@@ -3,15 +3,10 @@
 // of this distribution and at http://opencv.org/license.html.
 
 #include "precomp.hpp"
-#include "cuda/moments.cuh"
 
 using namespace cv;
 using namespace cv::cuda;
 
-int cv::cuda::numMoments(const MomentsOrder order) {
-    return order == MomentsOrder::FIRST_ORDER_MOMENTS ? device::imgproc::n1 : order == MomentsOrder::SECOND_ORDER_MOMENTS ? device::imgproc::n12 : device::imgproc::n123;
-}
-
 template<typename T>
 cv::Moments convertSpatialMomentsT(Mat spatialMoments, const MomentsOrder order) {
     switch (order) {
@@ -32,10 +27,17 @@ cv::Moments cv::cuda::convertSpatialMoments(Mat spatialMoments, const MomentsOrd
 }
 
 #if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
+    int cv::cuda::numMoments(MomentsOrder) { throw_no_cuda(); return 0; }
     Moments cv::cuda::moments(InputArray src, const bool binary, const MomentsOrder order, const int momentsType) { throw_no_cuda(); }
-    void spatialMoments(InputArray src, OutputArray moments, const bool binary, const MomentsOrder order, const int momentsType, Stream& stream) { throw_no_cuda(); }
+    void cv::cuda::spatialMoments(InputArray src, OutputArray moments, const bool binary, const MomentsOrder order, const int momentsType, Stream& stream) { throw_no_cuda(); }
 #else /* !defined (HAVE_CUDA) */
 
+#include "cuda/moments.cuh"
+
+int cv::cuda::numMoments(const MomentsOrder order) {
+    return order == MomentsOrder::FIRST_ORDER_MOMENTS ? device::imgproc::n1 : order == MomentsOrder::SECOND_ORDER_MOMENTS ? device::imgproc::n12 : device::imgproc::n123;
+}
+
 namespace cv { namespace cuda { namespace device { namespace imgproc {
         template <typename TSrc, typename TMoments>
         void moments(const PtrStepSzb src, PtrStepSzb moments, const bool binary, const int order, const int offsetX, const cudaStream_t stream);

modules/cudaoptflow/src/farneback.cpp

@@ -47,7 +47,7 @@ using namespace cv::cuda;
 
 #if !defined HAVE_CUDA || defined(CUDA_DISABLER)
 
-Ptr<FarnebackOpticalFlow> cv::cuda::FarnebackOpticalFlow::create(int, double, bool, int, int, int, double, int) { throw_no_cuda(); return Ptr<FarnebackOpticalFlow>(); }
+Ptr<cv::cuda::FarnebackOpticalFlow> cv::cuda::FarnebackOpticalFlow::create(int, double, bool, int, int, int, double, int) { throw_no_cuda(); return Ptr<FarnebackOpticalFlow>(); }
 
 #else
 

modules/cudaoptflow/src/precomp.hpp

@@ -52,7 +52,9 @@
 #include "opencv2/video.hpp"
 
 #include "opencv2/core/private.cuda.hpp"
+#if defined HAVE_CUDA
 #include "opencv2/core/cuda/vec_traits.hpp"
+#endif
 #include "opencv2/opencv_modules.hpp"
 
 #ifdef HAVE_OPENCV_CUDALEGACY

modules/cudev/include/opencv2/cudev/grid/detail/minmaxloc.hpp

@@ -148,8 +148,8 @@ namespace grid_minmaxloc_detail
         block = dim3(Policy::block_size_x, Policy::block_size_y);
         grid = dim3(divUp(cols, block.x * Policy::patch_size_x), divUp(rows, block.y * Policy::patch_size_y));
 
-        grid.x = ::min(grid.x, block.x);
-        grid.y = ::min(grid.y, block.y);
+        grid.x = std::min(grid.x, block.x);
+        grid.y = std::min(grid.y, block.y);
     }
 
     template <class Policy, class SrcPtr, typename ResType, class MaskPtr>