1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
/*M///////////////////////////////////////////////////////////////////////////////////////
//
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
//
// By downloading, copying, installing or using the software you agree to this license.
// If you do not agree to this license, do not download, install,
// copy or use the software.
//
//
// License Agreement
// For Open Source Computer Vision Library
//
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
// Third party copyrights are property of their respective owners.
//
// Redistribution and use in source and binary forms, with or without modification,
// are permitted provided that the following conditions are met:
//
// * Redistribution's of source code must retain the above copyright notice,
// this list of conditions and the following disclaimer.
//
// * Redistribution's in binary form must reproduce the above copyright notice,
// this list of conditions and the following disclaimer in the documentation
// and/or other materials provided with the distribution.
//
// * The name of the copyright holders may not be used to endorse or promote products
// derived from this software without specific prior written permission.
//
// This software is provided by the copyright holders and contributors "as is" and
// any express or implied warranties, including, but not limited to, the implied
// warranties of merchantability and fitness for a particular purpose are disclaimed.
// In no event shall the Intel Corporation or contributors be liable for any direct,
// indirect, incidental, special, exemplary, or consequential damages
// (including, but not limited to, procurement of substitute goods or services;
// loss of use, data, or profits; or business interruption) however caused
// and on any theory of liability, whether in contract, strict liability,
// or tort (including negligence or otherwise) arising in any way out of
// the use of this software, even if advised of the possibility of such damage.
//
//M*/
#if !defined CUDA_DISABLER
#include <opencv2/gpu/device/common.hpp>
#include <opencv2/gpu/device/vec_traits.hpp>
#include <opencv2/gpu/device/vec_math.hpp>
#include <opencv2/gpu/device/limits.hpp>
namespace cv { namespace gpu {
namespace device
{
template <typename D>
__global__ void Bayer2BGR_8u(const PtrStepb src, PtrStepSz<D> dst, const bool blue_last, const bool start_with_green)
{
const int s_x = blockIdx.x * blockDim.x + threadIdx.x;
int s_y = blockIdx.y * blockDim.y + threadIdx.y;
if (s_y >= dst.rows || (s_x << 2) >= dst.cols)
return;
s_y = ::min(::max(s_y, 1), dst.rows - 2);
uchar4 patch[3][3];
patch[0][1] = ((const uchar4*) src.ptr(s_y - 1))[s_x];
patch[0][0] = ((const uchar4*) src.ptr(s_y - 1))[::max(s_x - 1, 0)];
patch[0][2] = ((const uchar4*) src.ptr(s_y - 1))[::min(s_x + 1, ((dst.cols + 3) >> 2) - 1)];
patch[1][1] = ((const uchar4*) src.ptr(s_y))[s_x];
patch[1][0] = ((const uchar4*) src.ptr(s_y))[::max(s_x - 1, 0)];
patch[1][2] = ((const uchar4*) src.ptr(s_y))[::min(s_x + 1, ((dst.cols + 3) >> 2) - 1)];
patch[2][1] = ((const uchar4*) src.ptr(s_y + 1))[s_x];
patch[2][0] = ((const uchar4*) src.ptr(s_y + 1))[::max(s_x - 1, 0)];
patch[2][2] = ((const uchar4*) src.ptr(s_y + 1))[::min(s_x + 1, ((dst.cols + 3) >> 2) - 1)];
D res0 = VecTraits<D>::all(numeric_limits<uchar>::max());
D res1 = VecTraits<D>::all(numeric_limits<uchar>::max());
D res2 = VecTraits<D>::all(numeric_limits<uchar>::max());
D res3 = VecTraits<D>::all(numeric_limits<uchar>::max());
if ((s_y & 1) ^ start_with_green)
{
const int t0 = (patch[0][1].x + patch[2][1].x + 1) >> 1;
const int t1 = (patch[1][0].w + patch[1][1].y + 1) >> 1;
const int t2 = (patch[0][1].x + patch[0][1].z + patch[2][1].x + patch[2][1].z + 2) >> 2;
const int t3 = (patch[0][1].y + patch[1][1].x + patch[1][1].z + patch[2][1].y + 2) >> 2;
const int t4 = (patch[0][1].z + patch[2][1].z + 1) >> 1;
const int t5 = (patch[1][1].y + patch[1][1].w + 1) >> 1;
const int t6 = (patch[0][1].z + patch[0][2].x + patch[2][1].z + patch[2][2].x + 2) >> 2;
const int t7 = (patch[0][1].w + patch[1][1].z + patch[1][2].x + patch[2][1].w + 2) >> 2;
if ((s_y & 1) ^ blue_last)
{
res0.x = t1;
res0.y = patch[1][1].x;
res0.z = t0;
res1.x = patch[1][1].y;
res1.y = t3;
res1.z = t2;
res2.x = t5;
res2.y = patch[1][1].z;
res2.z = t4;
res3.x = patch[1][1].w;
res3.y = t7;
res3.z = t6;
}
else
{
res0.x = t0;
res0.y = patch[1][1].x;
res0.z = t1;
res1.x = t2;
res1.y = t3;
res1.z = patch[1][1].y;
res2.x = t4;
res2.y = patch[1][1].z;
res2.z = t5;
res3.x = t6;
res3.y = t7;
res3.z = patch[1][1].w;
}
}
else
{
const int t0 = (patch[0][0].w + patch[0][1].y + patch[2][0].w + patch[2][1].y + 2) >> 2;
const int t1 = (patch[0][1].x + patch[1][0].w + patch[1][1].y + patch[2][1].x + 2) >> 2;
const int t2 = (patch[0][1].y + patch[2][1].y + 1) >> 1;
const int t3 = (patch[1][1].x + patch[1][1].z + 1) >> 1;
const int t4 = (patch[0][1].y + patch[0][1].w + patch[2][1].y + patch[2][1].w + 2) >> 2;
const int t5 = (patch[0][1].z + patch[1][1].y + patch[1][1].w + patch[2][1].z + 2) >> 2;
const int t6 = (patch[0][1].w + patch[2][1].w + 1) >> 1;
const int t7 = (patch[1][1].z + patch[1][2].x + 1) >> 1;
if ((s_y & 1) ^ blue_last)
{
res0.x = patch[1][1].x;
res0.y = t1;
res0.z = t0;
res1.x = t3;
res1.y = patch[1][1].y;
res1.z = t2;
res2.x = patch[1][1].z;
res2.y = t5;
res2.z = t4;
res3.x = t7;
res3.y = patch[1][1].w;
res3.z = t6;
}
else
{
res0.x = t0;
res0.y = t1;
res0.z = patch[1][1].x;
res1.x = t2;
res1.y = patch[1][1].y;
res1.z = t3;
res2.x = t4;
res2.y = t5;
res2.z = patch[1][1].z;
res3.x = t6;
res3.y = patch[1][1].w;
res3.z = t7;
}
}
const int d_x = (blockIdx.x * blockDim.x + threadIdx.x) << 2;
const int d_y = blockIdx.y * blockDim.y + threadIdx.y;
dst(d_y, d_x) = res0;
if (d_x + 1 < dst.cols)
dst(d_y, d_x + 1) = res1;
if (d_x + 2 < dst.cols)
dst(d_y, d_x + 2) = res2;
if (d_x + 3 < dst.cols)
dst(d_y, d_x + 3) = res3;
}
template <typename D>
__global__ void Bayer2BGR_16u(const PtrStepb src, PtrStepSz<D> dst, const bool blue_last, const bool start_with_green)
{
const int s_x = blockIdx.x * blockDim.x + threadIdx.x;
int s_y = blockIdx.y * blockDim.y + threadIdx.y;
if (s_y >= dst.rows || (s_x << 1) >= dst.cols)
return;
s_y = ::min(::max(s_y, 1), dst.rows - 2);
ushort2 patch[3][3];
patch[0][1] = ((const ushort2*) src.ptr(s_y - 1))[s_x];
patch[0][0] = ((const ushort2*) src.ptr(s_y - 1))[::max(s_x - 1, 0)];
patch[0][2] = ((const ushort2*) src.ptr(s_y - 1))[::min(s_x + 1, ((dst.cols + 1) >> 1) - 1)];
patch[1][1] = ((const ushort2*) src.ptr(s_y))[s_x];
patch[1][0] = ((const ushort2*) src.ptr(s_y))[::max(s_x - 1, 0)];
patch[1][2] = ((const ushort2*) src.ptr(s_y))[::min(s_x + 1, ((dst.cols + 1) >> 1) - 1)];
patch[2][1] = ((const ushort2*) src.ptr(s_y + 1))[s_x];
patch[2][0] = ((const ushort2*) src.ptr(s_y + 1))[::max(s_x - 1, 0)];
patch[2][2] = ((const ushort2*) src.ptr(s_y + 1))[::min(s_x + 1, ((dst.cols + 1) >> 1) - 1)];
D res0 = VecTraits<D>::all(numeric_limits<ushort>::max());
D res1 = VecTraits<D>::all(numeric_limits<ushort>::max());
if ((s_y & 1) ^ start_with_green)
{
const int t0 = (patch[0][1].x + patch[2][1].x + 1) >> 1;
const int t1 = (patch[1][0].y + patch[1][1].y + 1) >> 1;
const int t2 = (patch[0][1].x + patch[0][2].x + patch[2][1].x + patch[2][2].x + 2) >> 2;
const int t3 = (patch[0][1].y + patch[1][1].x + patch[1][2].x + patch[2][1].y + 2) >> 2;
if ((s_y & 1) ^ blue_last)
{
res0.x = t1;
res0.y = patch[1][1].x;
res0.z = t0;
res1.x = patch[1][1].y;
res1.y = t3;
res1.z = t2;
}
else
{
res0.x = t0;
res0.y = patch[1][1].x;
res0.z = t1;
res1.x = t2;
res1.y = t3;
res1.z = patch[1][1].y;
}
}
else
{
const int t0 = (patch[0][0].y + patch[0][1].y + patch[2][0].y + patch[2][1].y + 2) >> 2;
const int t1 = (patch[0][1].x + patch[1][0].y + patch[1][1].y + patch[2][1].x + 2) >> 2;
const int t2 = (patch[0][1].y + patch[2][1].y + 1) >> 1;
const int t3 = (patch[1][1].x + patch[1][2].x + 1) >> 1;
if ((s_y & 1) ^ blue_last)
{
res0.x = patch[1][1].x;
res0.y = t1;
res0.z = t0;
res1.x = t3;
res1.y = patch[1][1].y;
res1.z = t2;
}
else
{
res0.x = t0;
res0.y = t1;
res0.z = patch[1][1].x;
res1.x = t2;
res1.y = patch[1][1].y;
res1.z = t3;
}
}
const int d_x = (blockIdx.x * blockDim.x + threadIdx.x) << 1;
const int d_y = blockIdx.y * blockDim.y + threadIdx.y;
dst(d_y, d_x) = res0;
if (d_x + 1 < dst.cols)
dst(d_y, d_x + 1) = res1;
}
template <int cn>
void Bayer2BGR_8u_gpu(PtrStepSzb src, PtrStepSzb dst, bool blue_last, bool start_with_green, cudaStream_t stream)
{
typedef typename TypeVec<uchar, cn>::vec_type dst_t;
const dim3 block(32, 8);
const dim3 grid(divUp(dst.cols, 4 * block.x), divUp(dst.rows, block.y));
cudaSafeCall( cudaFuncSetCacheConfig(Bayer2BGR_8u<dst_t>, cudaFuncCachePreferL1) );
Bayer2BGR_8u<dst_t><<<grid, block, 0, stream>>>(src, (PtrStepSz<dst_t>)dst, blue_last, start_with_green);
cudaSafeCall( cudaGetLastError() );
if (stream == 0)
cudaSafeCall( cudaDeviceSynchronize() );
}
template <int cn>
void Bayer2BGR_16u_gpu(PtrStepSzb src, PtrStepSzb dst, bool blue_last, bool start_with_green, cudaStream_t stream)
{
typedef typename TypeVec<ushort, cn>::vec_type dst_t;
const dim3 block(32, 8);
const dim3 grid(divUp(dst.cols, 2 * block.x), divUp(dst.rows, block.y));
cudaSafeCall( cudaFuncSetCacheConfig(Bayer2BGR_16u<dst_t>, cudaFuncCachePreferL1) );
Bayer2BGR_16u<dst_t><<<grid, block, 0, stream>>>(src, (PtrStepSz<dst_t>)dst, blue_last, start_with_green);
cudaSafeCall( cudaGetLastError() );
if (stream == 0)
cudaSafeCall( cudaDeviceSynchronize() );
}
template void Bayer2BGR_8u_gpu<3>(PtrStepSzb src, PtrStepSzb dst, bool blue_last, bool start_with_green, cudaStream_t stream);
template void Bayer2BGR_8u_gpu<4>(PtrStepSzb src, PtrStepSzb dst, bool blue_last, bool start_with_green, cudaStream_t stream);
template void Bayer2BGR_16u_gpu<3>(PtrStepSzb src, PtrStepSzb dst, bool blue_last, bool start_with_green, cudaStream_t stream);
template void Bayer2BGR_16u_gpu<4>(PtrStepSzb src, PtrStepSzb dst, bool blue_last, bool start_with_green, cudaStream_t stream);
}
}}
#endif /* CUDA_DISABLER */