generic-library/opencv
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Removed PtrElemStep, Marked DevMem1D as deprecated, now should use PtrStepSz now

Browse Source
This commit is contained in:
Anatoly Baksheev
2012-08-23 17:45:50 +04:00
parent 9be63c50dd
commit 70204a8e68
95 changed files with 2597 additions and 2563 deletions
Download Patch File Download Diff File Expand all files Collapse all files

76
modules/gpu/src/cuda/resize.cu
View File

@@ -53,7 +53,7 @@ namespace cv { namespace gpu { namespace device
{
namespace imgproc
{
template <typename Ptr2D, typename T> __global__ void resize(const Ptr2D src, float fx, float fy, DevMem2D_<T> dst)
template <typename Ptr2D, typename T> __global__ void resize(const Ptr2D src, float fx, float fy, PtrStepSz<T> dst)
{
const int x = blockDim.x * blockIdx.x + threadIdx.x;
const int y = blockDim.y * blockIdx.y + threadIdx.y;
@@ -67,7 +67,7 @@ namespace cv { namespace gpu { namespace device
}
}
template <typename Ptr2D, typename T> __global__ void resize_area(const Ptr2D src, float fx, float fy, DevMem2D_<T> dst)
template <typename Ptr2D, typename T> __global__ void resize_area(const Ptr2D src, float fx, float fy, PtrStepSz<T> dst)
{
const int x = blockDim.x * blockIdx.x + threadIdx.x;
const int y = blockDim.y * blockIdx.y + threadIdx.y;
@@ -80,7 +80,7 @@ namespace cv { namespace gpu { namespace device
template <template <typename> class Filter, typename T> struct ResizeDispatcherStream
{
static void call(DevMem2D_<T> src, float fx, float fy, DevMem2D_<T> dst, cudaStream_t stream)
static void call(PtrStepSz<T> src, float fx, float fy, PtrStepSz<T> dst, cudaStream_t stream)
{
dim3 block(32, 8);
dim3 grid(divUp(dst.cols, block.x), divUp(dst.rows, block.y));
@@ -96,7 +96,7 @@ namespace cv { namespace gpu { namespace device
template <typename T> struct ResizeDispatcherStream<AreaFilter, T>
{
static void call(DevMem2D_<T> src, float fx, float fy, DevMem2D_<T> dst, cudaStream_t stream)
static void call(PtrStepSz<T> src, float fx, float fy, PtrStepSz<T> dst, cudaStream_t stream)
{
dim3 block(32, 8);
dim3 grid(divUp(dst.cols, block.x), divUp(dst.rows, block.y));
@@ -113,7 +113,7 @@ namespace cv { namespace gpu { namespace device
template <typename T> struct ResizeDispatcherStream<IntegerAreaFilter, T>
{
static void call(DevMem2D_<T> src, float fx, float fy, DevMem2D_<T> dst, cudaStream_t stream)
static void call(PtrStepSz<T> src, float fx, float fy, PtrStepSz<T> dst, cudaStream_t stream)
{
dim3 block(32, 8);
dim3 grid(divUp(dst.cols, block.x), divUp(dst.rows, block.y));
@@ -129,7 +129,7 @@ namespace cv { namespace gpu { namespace device
template <template <typename> class Filter, typename T> struct ResizeDispatcherNonStream
{
static void call(DevMem2D_<T> src, DevMem2D_<T> srcWhole, int xoff, int yoff, float fx, float fy, DevMem2D_<T> dst)
static void call(PtrStepSz<T> src, PtrStepSz<T> srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSz<T> dst)
{
(void)srcWhole;
(void)xoff;
@@ -165,7 +165,7 @@ namespace cv { namespace gpu { namespace device
}; \
template <template <typename> class Filter> struct ResizeDispatcherNonStream<Filter, type > \
{ \
static void call(DevMem2D_< type > src, DevMem2D_< type > srcWhole, int xoff, int yoff, float fx, float fy, DevMem2D_< type > dst) \
static void call(PtrStepSz< type > src, PtrStepSz< type > srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSz< type > dst) \
{ \
dim3 block(32, 8); \
dim3 grid(divUp(dst.cols, block.x), divUp(dst.rows, block.y)); \
@@ -210,7 +210,7 @@ namespace cv { namespace gpu { namespace device
template <template <typename> class Filter, typename T> struct ResizeDispatcher
{
static void call(DevMem2D_<T> src, DevMem2D_<T> srcWhole, int xoff, int yoff, float fx, float fy, DevMem2D_<T> dst, cudaStream_t stream)
static void call(PtrStepSz<T> src, PtrStepSz<T> srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSz<T> dst, cudaStream_t stream)
{
if (stream == 0)
ResizeDispatcherNonStream<Filter, T>::call(src, srcWhole, xoff, yoff, fx, fy, dst);
@@ -221,7 +221,7 @@ namespace cv { namespace gpu { namespace device
template <typename T> struct ResizeDispatcher<AreaFilter, T>
{
static void call(DevMem2D_<T> src, DevMem2D_<T> srcWhole, int xoff, int yoff, float fx, float fy, DevMem2D_<T> dst, cudaStream_t stream)
static void call(PtrStepSz<T> src, PtrStepSz<T> srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSz<T> dst, cudaStream_t stream)
{
(void)srcWhole;
(void)xoff;
@@ -236,10 +236,10 @@ namespace cv { namespace gpu { namespace device
}
};
template <typename T> void resize_gpu(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy,
DevMem2Db dst, int interpolation, cudaStream_t stream)
template <typename T> void resize_gpu(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy,
PtrStepSzb dst, int interpolation, cudaStream_t stream)
{
typedef void (*caller_t)(DevMem2D_<T> src, DevMem2D_<T> srcWhole, int xoff, int yoff, float fx, float fy, DevMem2D_<T> dst, cudaStream_t stream);
typedef void (*caller_t)(PtrStepSz<T> src, PtrStepSz<T> srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSz<T> dst, cudaStream_t stream);
static const caller_t callers[4] =
{
@@ -252,39 +252,39 @@ namespace cv { namespace gpu { namespace device
if (interpolation == 3 && (fx <= 1.f || fy <= 1.f))
interpolation = 1;
callers[interpolation](static_cast< DevMem2D_<T> >(src), static_cast< DevMem2D_<T> >(srcWhole), xoff, yoff, fx, fy,
static_cast< DevMem2D_<T> >(dst), stream);
callers[interpolation](static_cast< PtrStepSz<T> >(src), static_cast< PtrStepSz<T> >(srcWhole), xoff, yoff, fx, fy,
static_cast< PtrStepSz<T> >(dst), stream);
}
template void resize_gpu<uchar >(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<uchar2>(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
template void resize_gpu<uchar3>(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
template void resize_gpu<uchar4>(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
template void resize_gpu<uchar >(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<uchar2>(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
template void resize_gpu<uchar3>(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
template void resize_gpu<uchar4>(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<schar>(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<char2>(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<char3>(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<char4>(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<schar>(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<char2>(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<char3>(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<char4>(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
template void resize_gpu<ushort >(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<ushort2>(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
template void resize_gpu<ushort3>(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
template void resize_gpu<ushort4>(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
template void resize_gpu<ushort >(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<ushort2>(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
template void resize_gpu<ushort3>(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
template void resize_gpu<ushort4>(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
template void resize_gpu<short >(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<short2>(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
template void resize_gpu<short3>(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
template void resize_gpu<short4>(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
template void resize_gpu<short >(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<short2>(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
template void resize_gpu<short3>(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
template void resize_gpu<short4>(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<int >(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<int2>(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<int3>(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<int4>(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<int >(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<int2>(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<int3>(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<int4>(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
template void resize_gpu<float >(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<float2>(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
template void resize_gpu<float3>(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
template void resize_gpu<float4>(DevMem2Db src, DevMem2Db srcWhole, int xoff, int yoff, float fx, float fy, DevMem2Db dst, int interpolation, cudaStream_t stream);
template void resize_gpu<float >(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
//template void resize_gpu<float2>(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
template void resize_gpu<float3>(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
template void resize_gpu<float4>(PtrStepSzb src, PtrStepSzb srcWhole, int xoff, int yoff, float fx, float fy, PtrStepSzb dst, int interpolation, cudaStream_t stream);
template<typename T> struct scan_traits{};