1/*M///////////////////////////////////////////////////////////////////////////////////////
2//
3//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
4//
5//  By downloading, copying, installing or using the software you agree to this license.
6//  If you do not agree to this license, do not download, install,
7//  copy or use the software.
8//
9//
10//                           License Agreement
11//                For Open Source Computer Vision Library
12//
13// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
14// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
15// Third party copyrights are property of their respective owners.
16//
17// Redistribution and use in source and binary forms, with or without modification,
18// are permitted provided that the following conditions are met:
19//
20//   * Redistribution's of source code must retain the above copyright notice,
21//     this list of conditions and the following disclaimer.
22//
23//   * Redistribution's in binary form must reproduce the above copyright notice,
24//     this list of conditions and the following disclaimer in the documentation
25//     and/or other materials provided with the distribution.
26//
27//   * The name of the copyright holders may not be used to endorse or promote products
28//     derived from this software without specific prior written permission.
29//
30// This software is provided by the copyright holders and contributors "as is" and
31// any express or implied warranties, including, but not limited to, the implied
32// warranties of merchantability and fitness for a particular purpose are disclaimed.
33// In no event shall the Intel Corporation or contributors be liable for any direct,
34// indirect, incidental, special, exemplary, or consequential damages
35// (including, but not limited to, procurement of substitute goods or services;
36// loss of use, data, or profits; or business interruption) however caused
37// and on any theory of liability, whether in contract, strict liability,
38// or tort (including negligence or otherwise) arising in any way out of
39// the use of this software, even if advised of the possibility of such damage.
40//
41//M*/
42
43#include "opencv2/opencv_modules.hpp"
44
45#ifndef HAVE_OPENCV_CUDEV
46
47#error "opencv_cudev is required"
48
49#else
50
51#include "opencv2/cudaarithm.hpp"
52#include "opencv2/cudev.hpp"
53#include "opencv2/core/private.cuda.hpp"
54
55using namespace cv;
56using namespace cv::cuda;
57using namespace cv::cudev;
58
59////////////////////////////////////////////////////////////////////////
60/// merge
61
62namespace
63{
64    template <int cn, typename T> struct MergeFunc;
65
66    template <typename T> struct MergeFunc<2, T>
67    {
68        static void call(const GpuMat* src, GpuMat& dst, Stream& stream)
69        {
70            gridMerge(zipPtr(globPtr<T>(src[0]), globPtr<T>(src[1])),
71                    globPtr<typename MakeVec<T, 2>::type>(dst),
72                    stream);
73        }
74    };
75
76    template <typename T> struct MergeFunc<3, T>
77    {
78        static void call(const GpuMat* src, GpuMat& dst, Stream& stream)
79        {
80            gridMerge(zipPtr(globPtr<T>(src[0]), globPtr<T>(src[1]), globPtr<T>(src[2])),
81                    globPtr<typename MakeVec<T, 3>::type>(dst),
82                    stream);
83        }
84    };
85
86    template <typename T> struct MergeFunc<4, T>
87    {
88        static void call(const GpuMat* src, GpuMat& dst, Stream& stream)
89        {
90            gridMerge(zipPtr(globPtr<T>(src[0]), globPtr<T>(src[1]), globPtr<T>(src[2]), globPtr<T>(src[3])),
91                    globPtr<typename MakeVec<T, 4>::type>(dst),
92                    stream);
93        }
94    };
95
96    void mergeImpl(const GpuMat* src, size_t n, cv::OutputArray _dst, Stream& stream)
97    {
98        CV_Assert( src != 0 );
99        CV_Assert( n > 0 && n <= 4 );
100
101        const int depth = src[0].depth();
102        const cv::Size size = src[0].size();
103
104        for (size_t i = 0; i < n; ++i)
105        {
106            CV_Assert( src[i].size() == size );
107            CV_Assert( src[i].depth() == depth );
108            CV_Assert( src[i].channels() == 1 );
109        }
110
111        if (n == 1)
112        {
113            src[0].copyTo(_dst, stream);
114        }
115        else
116        {
117            typedef void (*func_t)(const GpuMat* src, GpuMat& dst, Stream& stream);
118            static const func_t funcs[3][5] =
119            {
120                {MergeFunc<2, uchar>::call, MergeFunc<2, ushort>::call, MergeFunc<2, int>::call, 0, MergeFunc<2, double>::call},
121                {MergeFunc<3, uchar>::call, MergeFunc<3, ushort>::call, MergeFunc<3, int>::call, 0, MergeFunc<3, double>::call},
122                {MergeFunc<4, uchar>::call, MergeFunc<4, ushort>::call, MergeFunc<4, int>::call, 0, MergeFunc<4, double>::call}
123            };
124
125            const int channels = static_cast<int>(n);
126
127            GpuMat dst = getOutputMat(_dst, size, CV_MAKE_TYPE(depth, channels), stream);
128
129            const func_t func = funcs[channels - 2][CV_ELEM_SIZE(depth) / 2];
130
131            if (func == 0)
132                CV_Error(cv::Error::StsUnsupportedFormat, "Unsupported channel count or data type");
133
134            func(src, dst, stream);
135
136            syncOutput(dst, _dst, stream);
137        }
138    }
139}
140
141void cv::cuda::merge(const GpuMat* src, size_t n, OutputArray dst, Stream& stream)
142{
143    mergeImpl(src, n, dst, stream);
144}
145
146
147void cv::cuda::merge(const std::vector<GpuMat>& src, OutputArray dst, Stream& stream)
148{
149    mergeImpl(&src[0], src.size(), dst, stream);
150}
151
152////////////////////////////////////////////////////////////////////////
153/// split
154
155namespace
156{
157    template <int cn, typename T> struct SplitFunc;
158
159    template <typename T> struct SplitFunc<2, T>
160    {
161        static void call(const GpuMat& src, GpuMat* dst, Stream& stream)
162        {
163            GlobPtrSz<T> dstarr[2] =
164            {
165                globPtr<T>(dst[0]), globPtr<T>(dst[1])
166            };
167
168            gridSplit(globPtr<typename MakeVec<T, 2>::type>(src), dstarr, stream);
169        }
170    };
171
172    template <typename T> struct SplitFunc<3, T>
173    {
174        static void call(const GpuMat& src, GpuMat* dst, Stream& stream)
175        {
176            GlobPtrSz<T> dstarr[3] =
177            {
178                globPtr<T>(dst[0]), globPtr<T>(dst[1]), globPtr<T>(dst[2])
179            };
180
181            gridSplit(globPtr<typename MakeVec<T, 3>::type>(src), dstarr, stream);
182        }
183    };
184
185    template <typename T> struct SplitFunc<4, T>
186    {
187        static void call(const GpuMat& src, GpuMat* dst, Stream& stream)
188        {
189            GlobPtrSz<T> dstarr[4] =
190            {
191                globPtr<T>(dst[0]), globPtr<T>(dst[1]), globPtr<T>(dst[2]), globPtr<T>(dst[3])
192            };
193
194            gridSplit(globPtr<typename MakeVec<T, 4>::type>(src), dstarr, stream);
195        }
196    };
197
198    void splitImpl(const GpuMat& src, GpuMat* dst, Stream& stream)
199    {
200        typedef void (*func_t)(const GpuMat& src, GpuMat* dst, Stream& stream);
201        static const func_t funcs[3][5] =
202        {
203            {SplitFunc<2, uchar>::call, SplitFunc<2, ushort>::call, SplitFunc<2, int>::call, 0, SplitFunc<2, double>::call},
204            {SplitFunc<3, uchar>::call, SplitFunc<3, ushort>::call, SplitFunc<3, int>::call, 0, SplitFunc<3, double>::call},
205            {SplitFunc<4, uchar>::call, SplitFunc<4, ushort>::call, SplitFunc<4, int>::call, 0, SplitFunc<4, double>::call}
206        };
207
208        CV_Assert( dst != 0 );
209
210        const int depth = src.depth();
211        const int channels = src.channels();
212
213        CV_Assert( channels <= 4 );
214
215        if (channels == 0)
216            return;
217
218        if (channels == 1)
219        {
220            src.copyTo(dst[0], stream);
221            return;
222        }
223
224        for (int i = 0; i < channels; ++i)
225            dst[i].create(src.size(), depth);
226
227        const func_t func = funcs[channels - 2][CV_ELEM_SIZE(depth) / 2];
228
229        if (func == 0)
230            CV_Error(cv::Error::StsUnsupportedFormat, "Unsupported channel count or data type");
231
232        func(src, dst, stream);
233    }
234}
235
236void cv::cuda::split(InputArray _src, GpuMat* dst, Stream& stream)
237{
238    GpuMat src = getInputMat(_src, stream);
239    splitImpl(src, dst, stream);
240}
241
242void cv::cuda::split(InputArray _src, std::vector<GpuMat>& dst, Stream& stream)
243{
244    GpuMat src = getInputMat(_src, stream);
245    dst.resize(src.channels());
246    if (src.channels() > 0)
247        splitImpl(src, &dst[0], stream);
248}
249
250#endif
251