1 |
/***************************************************************************** |
/***************************************************************************** |
2 |
* |
* |
3 |
* XVID MPEG-4 VIDEO CODEC |
* XVID MPEG-4 VIDEO CODEC |
4 |
* - SAD calculation module (C part) - |
* - Sum Of Absolute Difference related code - |
5 |
* |
* |
6 |
* Copyright(C) 2002 Michael Militzer <isibaar@xvid.org> |
* Copyright(C) 2001-2010 Peter Ross <pross@xvid.org> |
7 |
* 2002 Peter Ross <pross@xvid.org> |
* 2010 Michael Militzer <michael@xvid.org> |
8 |
* |
* |
9 |
* This file is part of XviD, a free MPEG-4 video encoder/decoder |
* This program is free software ; you can redistribute it and/or modify |
10 |
* |
* it under the terms of the GNU General Public License as published by |
|
* XviD is free software; you can redistribute it and/or modify it |
|
|
* under the terms of the GNU General Public License as published by |
|
11 |
* the Free Software Foundation; either version 2 of the License, or |
* the Free Software Foundation; either version 2 of the License, or |
12 |
* (at your option) any later version. |
* (at your option) any later version. |
13 |
* |
* |
20 |
* along with this program; if not, write to the Free Software |
* along with this program; if not, write to the Free Software |
21 |
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
22 |
* |
* |
|
* Under section 8 of the GNU General Public License, the copyright |
|
|
* holders of XVID explicitly forbid distribution in the following |
|
|
* countries: |
|
|
* |
|
|
* - Japan |
|
|
* - United States of America |
|
|
* |
|
|
* Linking XviD statically or dynamically with other modules is making a |
|
|
* combined work based on XviD. Thus, the terms and conditions of the |
|
|
* GNU General Public License cover the whole combination. |
|
|
* |
|
|
* As a special exception, the copyright holders of XviD give you |
|
|
* permission to link XviD with independent modules that communicate with |
|
|
* XviD solely through the VFW1.1 and DShow interfaces, regardless of the |
|
|
* license terms of these independent modules, and to copy and distribute |
|
|
* the resulting combined work under terms of your choice, provided that |
|
|
* every copy of the combined work is accompanied by a complete copy of |
|
|
* the source code of XviD (the version of XviD used to produce the |
|
|
* combined work), being distributed under the terms of the GNU General |
|
|
* Public License plus this exception. An independent module is a module |
|
|
* which is not derived from or based on XviD. |
|
|
* |
|
|
* Note that people who make modified versions of XviD are not obligated |
|
|
* to grant this special exception for their modified versions; it is |
|
|
* their choice whether to do so. The GNU General Public License gives |
|
|
* permission to release a modified version without this exception; this |
|
|
* exception also makes it possible to release a modified version which |
|
|
* carries forward this exception. |
|
|
* |
|
23 |
* $Id$ |
* $Id$ |
24 |
* |
* |
25 |
****************************************************************************/ |
****************************************************************************/ |
26 |
|
|
27 |
#include "../portab.h" |
#include "../portab.h" |
28 |
|
#include "../global.h" |
29 |
#include "sad.h" |
#include "sad.h" |
30 |
|
|
31 |
|
#include <stdlib.h> |
32 |
|
|
33 |
sad16FuncPtr sad16; |
sad16FuncPtr sad16; |
34 |
sad8FuncPtr sad8; |
sad8FuncPtr sad8; |
35 |
sad16biFuncPtr sad16bi; |
sad16biFuncPtr sad16bi; |
36 |
sad8biFuncPtr sad8bi; /* not really sad16, but no difference in prototype */ |
sad8biFuncPtr sad8bi; /* not really sad16, but no difference in prototype */ |
37 |
dev16FuncPtr dev16; |
dev16FuncPtr dev16; |
38 |
|
sad16vFuncPtr sad16v; |
39 |
|
sse8Func_16bitPtr sse8_16bit; |
40 |
|
sse8Func_8bitPtr sse8_8bit; |
41 |
|
|
42 |
|
sseh8Func_16bitPtr sseh8_16bit; |
43 |
|
coeff8_energyFunc_Ptr coeff8_energy; |
44 |
|
blocksum8Func_Ptr blocksum8; |
45 |
|
|
46 |
sadInitFuncPtr sadInit; |
sadInitFuncPtr sadInit; |
47 |
|
|
|
#define ABS(X) (((X)>0)?(X):-(X)) |
|
|
|
|
|
#define MRSAD16_CORRFACTOR 8 |
|
|
uint32_t |
|
|
mrsad16_c(const uint8_t * const cur, |
|
|
const uint8_t * const ref, |
|
|
const uint32_t stride, |
|
|
const uint32_t best_sad) |
|
|
{ |
|
|
|
|
|
uint32_t sad = 0; |
|
|
int32_t mean = 0; |
|
|
uint32_t i, j; |
|
|
uint8_t const *ptr_cur = cur; |
|
|
uint8_t const *ptr_ref = ref; |
|
|
|
|
|
for (j = 0; j < 16; j++) { |
|
|
for (i = 0; i < 16; i++) { |
|
|
mean += ((int) *(ptr_cur + i) - (int) *(ptr_ref + i)); |
|
|
} |
|
|
ptr_cur += stride; |
|
|
ptr_ref += stride; |
|
|
|
|
|
} |
|
|
mean /= 256; |
|
|
|
|
|
for (j = 0; j < 16; j++) { |
|
|
|
|
|
ptr_cur -= stride; |
|
|
ptr_ref -= stride; |
|
|
|
|
|
for (i = 0; i < 16; i++) { |
|
|
|
|
|
sad += ABS(*(ptr_cur + i) - *(ptr_ref + i) - mean); |
|
|
if (sad >= best_sad) { |
|
|
return MRSAD16_CORRFACTOR * sad; |
|
|
} |
|
|
} |
|
|
} |
|
|
|
|
|
return MRSAD16_CORRFACTOR * sad; |
|
|
|
|
|
} |
|
|
|
|
48 |
|
|
49 |
uint32_t |
uint32_t |
50 |
sad16_c(const uint8_t * const cur, |
sad16_c(const uint8_t * const cur, |
54 |
{ |
{ |
55 |
|
|
56 |
uint32_t sad = 0; |
uint32_t sad = 0; |
57 |
uint32_t i, j; |
uint32_t j; |
58 |
uint8_t const *ptr_cur = cur; |
uint8_t const *ptr_cur = cur; |
59 |
uint8_t const *ptr_ref = ref; |
uint8_t const *ptr_ref = ref; |
60 |
|
|
61 |
for (j = 0; j < 16; j++) { |
for (j = 0; j < 16; j++) { |
62 |
|
sad += abs(ptr_cur[0] - ptr_ref[0]); |
63 |
|
sad += abs(ptr_cur[1] - ptr_ref[1]); |
64 |
|
sad += abs(ptr_cur[2] - ptr_ref[2]); |
65 |
|
sad += abs(ptr_cur[3] - ptr_ref[3]); |
66 |
|
sad += abs(ptr_cur[4] - ptr_ref[4]); |
67 |
|
sad += abs(ptr_cur[5] - ptr_ref[5]); |
68 |
|
sad += abs(ptr_cur[6] - ptr_ref[6]); |
69 |
|
sad += abs(ptr_cur[7] - ptr_ref[7]); |
70 |
|
sad += abs(ptr_cur[8] - ptr_ref[8]); |
71 |
|
sad += abs(ptr_cur[9] - ptr_ref[9]); |
72 |
|
sad += abs(ptr_cur[10] - ptr_ref[10]); |
73 |
|
sad += abs(ptr_cur[11] - ptr_ref[11]); |
74 |
|
sad += abs(ptr_cur[12] - ptr_ref[12]); |
75 |
|
sad += abs(ptr_cur[13] - ptr_ref[13]); |
76 |
|
sad += abs(ptr_cur[14] - ptr_ref[14]); |
77 |
|
sad += abs(ptr_cur[15] - ptr_ref[15]); |
78 |
|
|
79 |
for (i = 0; i < 16; i++) { |
if (sad >= best_sad) |
|
|
|
|
sad += ABS(*(ptr_cur + i) - *(ptr_ref + i)); |
|
|
|
|
|
if (sad >= best_sad) { |
|
80 |
return sad; |
return sad; |
|
} |
|
|
|
|
|
|
|
|
} |
|
81 |
|
|
82 |
ptr_cur += stride; |
ptr_cur += stride; |
83 |
ptr_ref += stride; |
ptr_ref += stride; |
88 |
|
|
89 |
} |
} |
90 |
|
|
|
|
|
|
|
|
91 |
uint32_t |
uint32_t |
92 |
sad16bi_c(const uint8_t * const cur, |
sad16bi_c(const uint8_t * const cur, |
93 |
const uint8_t * const ref1, |
const uint8_t * const ref1, |
105 |
|
|
106 |
for (i = 0; i < 16; i++) { |
for (i = 0; i < 16; i++) { |
107 |
int pixel = (ptr_ref1[i] + ptr_ref2[i] + 1) / 2; |
int pixel = (ptr_ref1[i] + ptr_ref2[i] + 1) / 2; |
108 |
|
sad += abs(ptr_cur[i] - pixel); |
|
if (pixel < 0) { |
|
|
pixel = 0; |
|
|
} else if (pixel > 255) { |
|
|
pixel = 255; |
|
|
} |
|
|
|
|
|
sad += ABS(ptr_cur[i] - pixel); |
|
109 |
} |
} |
110 |
|
|
111 |
ptr_cur += stride; |
ptr_cur += stride; |
135 |
|
|
136 |
for (i = 0; i < 8; i++) { |
for (i = 0; i < 8; i++) { |
137 |
int pixel = (ptr_ref1[i] + ptr_ref2[i] + 1) / 2; |
int pixel = (ptr_ref1[i] + ptr_ref2[i] + 1) / 2; |
138 |
|
sad += abs(ptr_cur[i] - pixel); |
|
if (pixel < 0) { |
|
|
pixel = 0; |
|
|
} else if (pixel > 255) { |
|
|
pixel = 255; |
|
|
} |
|
|
|
|
|
sad += ABS(ptr_cur[i] - pixel); |
|
139 |
} |
} |
140 |
|
|
141 |
ptr_cur += stride; |
ptr_cur += stride; |
156 |
const uint32_t stride) |
const uint32_t stride) |
157 |
{ |
{ |
158 |
uint32_t sad = 0; |
uint32_t sad = 0; |
159 |
uint32_t i, j; |
uint32_t j; |
160 |
uint8_t const *ptr_cur = cur; |
uint8_t const *ptr_cur = cur; |
161 |
uint8_t const *ptr_ref = ref; |
uint8_t const *ptr_ref = ref; |
162 |
|
|
163 |
for (j = 0; j < 8; j++) { |
for (j = 0; j < 8; j++) { |
164 |
|
|
165 |
for (i = 0; i < 8; i++) { |
sad += abs(ptr_cur[0] - ptr_ref[0]); |
166 |
sad += ABS(*(ptr_cur + i) - *(ptr_ref + i)); |
sad += abs(ptr_cur[1] - ptr_ref[1]); |
167 |
} |
sad += abs(ptr_cur[2] - ptr_ref[2]); |
168 |
|
sad += abs(ptr_cur[3] - ptr_ref[3]); |
169 |
|
sad += abs(ptr_cur[4] - ptr_ref[4]); |
170 |
|
sad += abs(ptr_cur[5] - ptr_ref[5]); |
171 |
|
sad += abs(ptr_cur[6] - ptr_ref[6]); |
172 |
|
sad += abs(ptr_cur[7] - ptr_ref[7]); |
173 |
|
|
174 |
ptr_cur += stride; |
ptr_cur += stride; |
175 |
ptr_ref += stride; |
ptr_ref += stride; |
180 |
} |
} |
181 |
|
|
182 |
|
|
|
|
|
|
|
|
183 |
/* average deviation from mean */ |
/* average deviation from mean */ |
184 |
|
|
185 |
uint32_t |
uint32_t |
207 |
for (j = 0; j < 16; j++) { |
for (j = 0; j < 16; j++) { |
208 |
|
|
209 |
for (i = 0; i < 16; i++) |
for (i = 0; i < 16; i++) |
210 |
dev += ABS(*(ptr_cur + i) - (int32_t) mean); |
dev += abs(*(ptr_cur + i) - (int32_t) mean); |
211 |
|
|
212 |
ptr_cur += stride; |
ptr_cur += stride; |
213 |
|
|
215 |
|
|
216 |
return dev; |
return dev; |
217 |
} |
} |
218 |
|
|
219 |
|
uint32_t sad16v_c(const uint8_t * const cur, |
220 |
|
const uint8_t * const ref, |
221 |
|
const uint32_t stride, |
222 |
|
int32_t *sad) |
223 |
|
{ |
224 |
|
sad[0] = sad8(cur, ref, stride); |
225 |
|
sad[1] = sad8(cur + 8, ref + 8, stride); |
226 |
|
sad[2] = sad8(cur + 8*stride, ref + 8*stride, stride); |
227 |
|
sad[3] = sad8(cur + 8*stride + 8, ref + 8*stride + 8, stride); |
228 |
|
|
229 |
|
return sad[0]+sad[1]+sad[2]+sad[3]; |
230 |
|
} |
231 |
|
|
232 |
|
uint32_t sad32v_c(const uint8_t * const cur, |
233 |
|
const uint8_t * const ref, |
234 |
|
const uint32_t stride, |
235 |
|
int32_t *sad) |
236 |
|
{ |
237 |
|
sad[0] = sad16(cur, ref, stride, 256*4096); |
238 |
|
sad[1] = sad16(cur + 16, ref + 16, stride, 256*4096); |
239 |
|
sad[2] = sad16(cur + 16*stride, ref + 16*stride, stride, 256*4096); |
240 |
|
sad[3] = sad16(cur + 16*stride + 16, ref + 16*stride + 16, stride, 256*4096); |
241 |
|
|
242 |
|
return sad[0]+sad[1]+sad[2]+sad[3]; |
243 |
|
} |
244 |
|
|
245 |
|
|
246 |
|
|
247 |
|
#define MRSAD16_CORRFACTOR 8 |
248 |
|
uint32_t |
249 |
|
mrsad16_c(const uint8_t * const cur, |
250 |
|
const uint8_t * const ref, |
251 |
|
const uint32_t stride, |
252 |
|
const uint32_t best_sad) |
253 |
|
{ |
254 |
|
|
255 |
|
uint32_t sad = 0; |
256 |
|
int32_t mean = 0; |
257 |
|
uint32_t i, j; |
258 |
|
uint8_t const *ptr_cur = cur; |
259 |
|
uint8_t const *ptr_ref = ref; |
260 |
|
|
261 |
|
for (j = 0; j < 16; j++) { |
262 |
|
for (i = 0; i < 16; i++) { |
263 |
|
mean += ((int) *(ptr_cur + i) - (int) *(ptr_ref + i)); |
264 |
|
} |
265 |
|
ptr_cur += stride; |
266 |
|
ptr_ref += stride; |
267 |
|
|
268 |
|
} |
269 |
|
mean /= 256; |
270 |
|
|
271 |
|
for (j = 0; j < 16; j++) { |
272 |
|
|
273 |
|
ptr_cur -= stride; |
274 |
|
ptr_ref -= stride; |
275 |
|
|
276 |
|
for (i = 0; i < 16; i++) { |
277 |
|
|
278 |
|
sad += abs(*(ptr_cur + i) - *(ptr_ref + i) - mean); |
279 |
|
if (sad >= best_sad) { |
280 |
|
return MRSAD16_CORRFACTOR * sad; |
281 |
|
} |
282 |
|
} |
283 |
|
} |
284 |
|
|
285 |
|
return MRSAD16_CORRFACTOR * sad; |
286 |
|
} |
287 |
|
|
288 |
|
uint32_t |
289 |
|
sse8_16bit_c(const int16_t * b1, |
290 |
|
const int16_t * b2, |
291 |
|
const uint32_t stride) |
292 |
|
{ |
293 |
|
int i; |
294 |
|
int sse = 0; |
295 |
|
|
296 |
|
for (i=0; i<8; i++) { |
297 |
|
sse += (b1[0] - b2[0])*(b1[0] - b2[0]); |
298 |
|
sse += (b1[1] - b2[1])*(b1[1] - b2[1]); |
299 |
|
sse += (b1[2] - b2[2])*(b1[2] - b2[2]); |
300 |
|
sse += (b1[3] - b2[3])*(b1[3] - b2[3]); |
301 |
|
sse += (b1[4] - b2[4])*(b1[4] - b2[4]); |
302 |
|
sse += (b1[5] - b2[5])*(b1[5] - b2[5]); |
303 |
|
sse += (b1[6] - b2[6])*(b1[6] - b2[6]); |
304 |
|
sse += (b1[7] - b2[7])*(b1[7] - b2[7]); |
305 |
|
|
306 |
|
b1 = (const int16_t*)((int8_t*)b1+stride); |
307 |
|
b2 = (const int16_t*)((int8_t*)b2+stride); |
308 |
|
} |
309 |
|
|
310 |
|
return(sse); |
311 |
|
} |
312 |
|
|
313 |
|
uint32_t |
314 |
|
sse8_8bit_c(const uint8_t * b1, |
315 |
|
const uint8_t * b2, |
316 |
|
const uint32_t stride) |
317 |
|
{ |
318 |
|
int i; |
319 |
|
int sse = 0; |
320 |
|
|
321 |
|
for (i=0; i<8; i++) { |
322 |
|
sse += (b1[0] - b2[0])*(b1[0] - b2[0]); |
323 |
|
sse += (b1[1] - b2[1])*(b1[1] - b2[1]); |
324 |
|
sse += (b1[2] - b2[2])*(b1[2] - b2[2]); |
325 |
|
sse += (b1[3] - b2[3])*(b1[3] - b2[3]); |
326 |
|
sse += (b1[4] - b2[4])*(b1[4] - b2[4]); |
327 |
|
sse += (b1[5] - b2[5])*(b1[5] - b2[5]); |
328 |
|
sse += (b1[6] - b2[6])*(b1[6] - b2[6]); |
329 |
|
sse += (b1[7] - b2[7])*(b1[7] - b2[7]); |
330 |
|
|
331 |
|
b1 = b1+stride; |
332 |
|
b2 = b2+stride; |
333 |
|
} |
334 |
|
|
335 |
|
return(sse); |
336 |
|
} |
337 |
|
|
338 |
|
|
339 |
|
/* PSNR-HVS-M helper functions */ |
340 |
|
|
341 |
|
static const int16_t iMask_Coeff[64] = { |
342 |
|
0, 29788, 32767, 20479, 13653, 8192, 6425, 5372, |
343 |
|
27306, 27306, 23405, 17246, 12603, 5650, 5461, 5958, |
344 |
|
23405, 25205, 20479, 13653, 8192, 5749, 4749, 5851, |
345 |
|
23405, 19275, 14894, 11299, 6425, 3766, 4096, 5285, |
346 |
|
18204, 14894, 8856, 5851, 4819, 3006, 3181, 4255, |
347 |
|
13653, 9362, 5958, 5120, 4045, 3151, 2900, 3562, |
348 |
|
6687, 5120, 4201, 3766, 3181, 2708, 2730, 3244, |
349 |
|
4551, 3562, 3449, 3344, 2926, 3277, 3181, 3310 |
350 |
|
}; |
351 |
|
|
352 |
|
/* Calculate CSF weighted energy of DCT coefficients */ |
353 |
|
|
354 |
|
uint32_t |
355 |
|
coeff8_energy_c(const int16_t * dct) |
356 |
|
{ |
357 |
|
int x, y; |
358 |
|
uint32_t sum_a = 0; |
359 |
|
|
360 |
|
for (y = 0; y < 8; y += 2) { |
361 |
|
for (x = 0; x < 8; x += 2) { |
362 |
|
int16_t a0 = ((dct[y*8+x]<<4) * iMask_Coeff[y*8+x]) >> 16; |
363 |
|
int16_t a1 = ((dct[y*8+x+1]<<4) * iMask_Coeff[y*8+x+1]) >> 16; |
364 |
|
int16_t a2 = ((dct[(y+1)*8+x]<<4) * iMask_Coeff[(y+1)*8+x]) >> 16; |
365 |
|
int16_t a3 = ((dct[(y+1)*8+x+1]<<4) * iMask_Coeff[(y+1)*8+x+1]) >> 16; |
366 |
|
|
367 |
|
sum_a += ((a0*a0 + a1*a1 + a2*a2 + a3*a3) >> 3); |
368 |
|
} |
369 |
|
} |
370 |
|
|
371 |
|
return sum_a; |
372 |
|
} |
373 |
|
|
374 |
|
/* Calculate MSE of DCT coeffs reduced by masking effect */ |
375 |
|
|
376 |
|
uint32_t |
377 |
|
sseh8_16bit_c(const int16_t * cur, const int16_t * ref, uint16_t mask) |
378 |
|
{ |
379 |
|
int j, i; |
380 |
|
uint32_t mse_h = 0; |
381 |
|
|
382 |
|
for (j = 0; j < 8; j++) { |
383 |
|
for (i = 0; i < 8; i++) { |
384 |
|
uint32_t t = (mask * Inv_iMask_Coeff[j*8+i] + 32) >> 7; |
385 |
|
uint16_t u = abs(cur[j*8+i] - ref[j*8+i]) << 4; |
386 |
|
uint16_t thresh = (t < 65536) ? t : 65535; |
387 |
|
|
388 |
|
if (u < thresh) |
389 |
|
u = 0; /* The error is not perceivable */ |
390 |
|
else |
391 |
|
u -= thresh; |
392 |
|
|
393 |
|
u = ((u + iCSF_Round[j*8 + i]) * iCSF_Coeff[j*8 + i]) >> 16; |
394 |
|
|
395 |
|
mse_h += (uint32_t) (u * u); |
396 |
|
} |
397 |
|
} |
398 |
|
|
399 |
|
return mse_h; |
400 |
|
} |
401 |
|
|
402 |
|
/* Sums all pixels of 8x8 block */ |
403 |
|
|
404 |
|
uint32_t |
405 |
|
blocksum8_c(const int8_t * cur, int stride, uint16_t sums[4], uint32_t squares[4]) |
406 |
|
{ |
407 |
|
int i, j; |
408 |
|
uint32_t sum = 0; |
409 |
|
|
410 |
|
sums[0] = sums[1] = sums[2] = sums[3] = 0; |
411 |
|
squares[0] = squares[1] = squares[2] = squares[3] = 0; |
412 |
|
|
413 |
|
for (j = 0; j < 8; j++) { |
414 |
|
for (i = 0; i < 8; i++) { |
415 |
|
uint8_t p = cur[j*stride + i]; |
416 |
|
|
417 |
|
sums[(j>>2)*2 + (i>>2)] += p; |
418 |
|
squares[(j>>2)*2 + (i>>2)] += p*p; |
419 |
|
|
420 |
|
sum += p; |
421 |
|
} |
422 |
|
} |
423 |
|
|
424 |
|
return sum; |
425 |
|
} |