1 |
/************************************************************************** |
/***************************************************************************** |
2 |
* |
* |
3 |
* XVID MPEG-4 VIDEO CODEC |
* XVID MPEG-4 VIDEO CODEC |
4 |
* sum of absolute difference |
* - SAD calculation module (C part) - |
5 |
|
* |
6 |
|
* Copyright(C) 2002 Michael Militzer <isibaar@xvid.org> |
7 |
|
* 2002 Peter Ross <pross@xvid.org> |
8 |
* |
* |
9 |
* This program is an implementation of a part of one or more MPEG-4 |
* This program is an implementation of a part of one or more MPEG-4 |
10 |
* Video tools as specified in ISO/IEC 14496-2 standard. Those intending |
* Video tools as specified in ISO/IEC 14496-2 standard. Those intending |
27 |
* |
* |
28 |
* You should have received a copy of the GNU General Public License |
* You should have received a copy of the GNU General Public License |
29 |
* along with this program; if not, write to the Free Software |
* along with this program; if not, write to the Free Software |
30 |
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. |
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
|
* |
|
|
*************************************************************************/ |
|
|
|
|
|
/************************************************************************** |
|
|
* |
|
|
* History: |
|
31 |
* |
* |
32 |
* 14.02.2002 added sad16bi_c() |
****************************************************************************/ |
|
* 10.11.2001 initial version; (c)2001 peter ross <pross@cs.rmit.edu.au> |
|
|
* |
|
|
*************************************************************************/ |
|
|
|
|
33 |
|
|
34 |
#include "../portab.h" |
#include "../portab.h" |
35 |
#include "sad.h" |
#include "sad.h" |
39 |
sad16biFuncPtr sad16bi; |
sad16biFuncPtr sad16bi; |
40 |
sad8biFuncPtr sad8bi; // not really sad16, but no difference in prototype |
sad8biFuncPtr sad8bi; // not really sad16, but no difference in prototype |
41 |
dev16FuncPtr dev16; |
dev16FuncPtr dev16; |
|
sad16vFuncPtr sad16v; |
|
42 |
|
|
43 |
sadInitFuncPtr sadInit; |
sadInitFuncPtr sadInit; |
44 |
|
|
45 |
#define ABS(X) (((X)>0)?(X):-(X)) |
#define ABS(X) (((X)>0)?(X):-(X)) |
46 |
|
|
47 |
|
#define MRSAD16_CORRFACTOR 8 |
48 |
|
uint32_t |
49 |
|
mrsad16_c(const uint8_t * const cur, |
50 |
|
const uint8_t * const ref, |
51 |
|
const uint32_t stride, |
52 |
|
const uint32_t best_sad) |
53 |
|
{ |
54 |
|
|
55 |
|
uint32_t sad = 0; |
56 |
|
int32_t mean = 0; |
57 |
|
uint32_t i, j; |
58 |
|
uint8_t const *ptr_cur = cur; |
59 |
|
uint8_t const *ptr_ref = ref; |
60 |
|
|
61 |
|
for (j = 0; j < 16; j++) { |
62 |
|
for (i = 0; i < 16; i++) { |
63 |
|
mean += ((int) *(ptr_cur + i) - (int) *(ptr_ref + i)); |
64 |
|
} |
65 |
|
ptr_cur += stride; |
66 |
|
ptr_ref += stride; |
67 |
|
|
68 |
|
} |
69 |
|
mean /= 256; |
70 |
|
|
71 |
|
for (j = 0; j < 16; j++) { |
72 |
|
|
73 |
|
ptr_cur -= stride; |
74 |
|
ptr_ref -= stride; |
75 |
|
|
76 |
|
for (i = 0; i < 16; i++) { |
77 |
|
|
78 |
|
sad += ABS(*(ptr_cur + i) - *(ptr_ref + i) - mean); |
79 |
|
if (sad >= best_sad) { |
80 |
|
return MRSAD16_CORRFACTOR * sad; |
81 |
|
} |
82 |
|
} |
83 |
|
} |
84 |
|
|
85 |
|
return MRSAD16_CORRFACTOR * sad; |
86 |
|
|
87 |
|
} |
88 |
|
|
89 |
|
|
90 |
uint32_t |
uint32_t |
91 |
sad16_c(const uint8_t * const cur, |
sad16_c(const uint8_t * const cur, |
92 |
const uint8_t * const ref, |
const uint8_t * const ref, |
95 |
{ |
{ |
96 |
|
|
97 |
uint32_t sad = 0; |
uint32_t sad = 0; |
98 |
uint32_t j; |
uint32_t i, j; |
99 |
uint8_t const *ptr_cur = cur; |
uint8_t const *ptr_cur = cur; |
100 |
uint8_t const *ptr_ref = ref; |
uint8_t const *ptr_ref = ref; |
101 |
|
|
102 |
for (j = 0; j < 16; j++) { |
for (j = 0; j < 16; j++) { |
|
sad += ABS(ptr_cur[0] - ptr_ref[0]); |
|
|
sad += ABS(ptr_cur[1] - ptr_ref[1]); |
|
|
sad += ABS(ptr_cur[2] - ptr_ref[2]); |
|
|
sad += ABS(ptr_cur[3] - ptr_ref[3]); |
|
|
sad += ABS(ptr_cur[4] - ptr_ref[4]); |
|
|
sad += ABS(ptr_cur[5] - ptr_ref[5]); |
|
|
sad += ABS(ptr_cur[6] - ptr_ref[6]); |
|
|
sad += ABS(ptr_cur[7] - ptr_ref[7]); |
|
|
sad += ABS(ptr_cur[8] - ptr_ref[8]); |
|
|
sad += ABS(ptr_cur[9] - ptr_ref[9]); |
|
|
sad += ABS(ptr_cur[10] - ptr_ref[10]); |
|
|
sad += ABS(ptr_cur[11] - ptr_ref[11]); |
|
|
sad += ABS(ptr_cur[12] - ptr_ref[12]); |
|
|
sad += ABS(ptr_cur[13] - ptr_ref[13]); |
|
|
sad += ABS(ptr_cur[14] - ptr_ref[14]); |
|
|
sad += ABS(ptr_cur[15] - ptr_ref[15]); |
|
103 |
|
|
104 |
if (sad >= best_sad) |
for (i = 0; i < 16; i++) { |
105 |
|
|
106 |
|
sad += ABS(*(ptr_cur + i) - *(ptr_ref + i)); |
107 |
|
|
108 |
|
if (sad >= best_sad) { |
109 |
return sad; |
return sad; |
110 |
|
} |
111 |
|
|
112 |
|
|
113 |
|
} |
114 |
|
|
115 |
ptr_cur += stride; |
ptr_cur += stride; |
116 |
ptr_ref += stride; |
ptr_ref += stride; |
121 |
|
|
122 |
} |
} |
123 |
|
|
124 |
|
|
125 |
|
|
126 |
uint32_t |
uint32_t |
127 |
sad16bi_c(const uint8_t * const cur, |
sad16bi_c(const uint8_t * const cur, |
128 |
const uint8_t * const ref1, |
const uint8_t * const ref1, |
205 |
const uint32_t stride) |
const uint32_t stride) |
206 |
{ |
{ |
207 |
uint32_t sad = 0; |
uint32_t sad = 0; |
208 |
uint32_t j; |
uint32_t i, j; |
209 |
uint8_t const *ptr_cur = cur; |
uint8_t const *ptr_cur = cur; |
210 |
uint8_t const *ptr_ref = ref; |
uint8_t const *ptr_ref = ref; |
211 |
|
|
212 |
for (j = 0; j < 8; j++) { |
for (j = 0; j < 8; j++) { |
213 |
|
|
214 |
sad += ABS(ptr_cur[0] - ptr_ref[0]); |
for (i = 0; i < 8; i++) { |
215 |
sad += ABS(ptr_cur[1] - ptr_ref[1]); |
sad += ABS(*(ptr_cur + i) - *(ptr_ref + i)); |
216 |
sad += ABS(ptr_cur[2] - ptr_ref[2]); |
} |
|
sad += ABS(ptr_cur[3] - ptr_ref[3]); |
|
|
sad += ABS(ptr_cur[4] - ptr_ref[4]); |
|
|
sad += ABS(ptr_cur[5] - ptr_ref[5]); |
|
|
sad += ABS(ptr_cur[6] - ptr_ref[6]); |
|
|
sad += ABS(ptr_cur[7] - ptr_ref[7]); |
|
217 |
|
|
218 |
ptr_cur += stride; |
ptr_cur += stride; |
219 |
ptr_ref += stride; |
ptr_ref += stride; |
224 |
} |
} |
225 |
|
|
226 |
|
|
227 |
|
|
228 |
|
|
229 |
/* average deviation from mean */ |
/* average deviation from mean */ |
230 |
|
|
231 |
uint32_t |
uint32_t |
261 |
|
|
262 |
return dev; |
return dev; |
263 |
} |
} |
|
|
|
|
uint32_t sad16v_c(const uint8_t * const cur, |
|
|
const uint8_t * const ref, |
|
|
const uint32_t stride, |
|
|
int32_t *sad) |
|
|
{ |
|
|
sad[0] = sad8(cur, ref, stride); |
|
|
sad[1] = sad8(cur + 8, ref + 8, stride); |
|
|
sad[2] = sad8(cur + 8*stride, ref + 8*stride, stride); |
|
|
sad[3] = sad8(cur + 8*stride + 8, ref + 8*stride + 8, stride); |
|
|
|
|
|
return sad[0]+sad[1]+sad[2]+sad[3]; |
|
|
} |
|
|
|
|
|
uint32_t sad32v_c(const uint8_t * const cur, |
|
|
const uint8_t * const ref, |
|
|
const uint32_t stride, |
|
|
int32_t *sad) |
|
|
{ |
|
|
sad[0] = sad16(cur, ref, stride, 256*4096); |
|
|
sad[1] = sad16(cur + 8, ref + 8, stride, 256*4096); |
|
|
sad[2] = sad16(cur + 8*stride, ref + 8*stride, stride, 256*4096); |
|
|
sad[3] = sad16(cur + 8*stride + 8, ref + 8*stride + 8, stride, 256*4096); |
|
|
|
|
|
return sad[0]+sad[1]+sad[2]+sad[3]; |
|
|
} |
|
|
|
|
|
|
|
|
|
|
|
#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; |
|
|
|
|
|
} |
|
|
|
|
|
|
|