FFmpeg
dctref.c
Go to the documentation of this file.
1 /*
2  * reference discrete cosine transform (double precision)
3  * Copyright (C) 2009 Dylan Yudaken
4  *
5  * This file is part of FFmpeg.
6  *
7  * FFmpeg is free software; you can redistribute it and/or
8  * modify it under the terms of the GNU Lesser General Public
9  * License as published by the Free Software Foundation; either
10  * version 2.1 of the License, or (at your option) any later version.
11  *
12  * FFmpeg is distributed in the hope that it will be useful,
13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15  * Lesser General Public License for more details.
16  *
17  * You should have received a copy of the GNU Lesser General Public
18  * License along with FFmpeg; if not, write to the Free Software
19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20  */
21 
22 /**
23  * @file
24  * reference discrete cosine transform (double precision)
25  *
26  * @author Dylan Yudaken (dyudaken at gmail)
27  *
28  * @note This file could be optimized a lot, but is for
29  * reference and so readability is better.
30  */
31 
32 #include "libavutil/mathematics.h"
33 #include "dctref.h"
34 
35 static double coefficients[8 * 8];
36 
37 /**
38  * Initialize the double precision discrete cosine transform
39  * functions fdct & idct.
40  */
42 {
43  unsigned int i, j;
44 
45  for (j = 0; j < 8; ++j) {
46  coefficients[j] = sqrt(0.125);
47  for (i = 8; i < 64; i += 8) {
48  coefficients[i + j] = 0.5 * cos(i * (j + 0.5) * M_PI / 64.0);
49  }
50  }
51 }
52 
53 /**
54  * Transform 8x8 block of data with a double precision forward DCT <br>
55  * This is a reference implementation.
56  *
57  * @param block pointer to 8x8 block of data to transform
58  */
59 void ff_ref_fdct(short *block)
60 {
61  /* implement the equation: block = coefficients * block * coefficients' */
62 
63  unsigned int i, j, k;
64  double out[8 * 8];
65 
66  /* out = coefficients * block */
67  for (i = 0; i < 64; i += 8) {
68  for (j = 0; j < 8; ++j) {
69  double tmp = 0;
70  for (k = 0; k < 8; ++k) {
71  tmp += coefficients[i + k] * block[k * 8 + j];
72  }
73  out[i + j] = tmp * 8;
74  }
75  }
76 
77  /* block = out * (coefficients') */
78  for (j = 0; j < 8; ++j) {
79  for (i = 0; i < 64; i += 8) {
80  double tmp = 0;
81  for (k = 0; k < 8; ++k) {
82  tmp += out[i + k] * coefficients[j * 8 + k];
83  }
84  block[i + j] = floor(tmp + 0.499999999999);
85  }
86  }
87 }
88 
89 /**
90  * Transform 8x8 block of data with a double precision inverse DCT <br>
91  * This is a reference implementation.
92  *
93  * @param block pointer to 8x8 block of data to transform
94  */
95 void ff_ref_idct(short *block)
96 {
97  /* implement the equation: block = (coefficients') * block * coefficients */
98 
99  unsigned int i, j, k;
100  double out[8 * 8];
101 
102  /* out = block * coefficients */
103  for (i = 0; i < 64; i += 8) {
104  for (j = 0; j < 8; ++j) {
105  double tmp = 0;
106  for (k = 0; k < 8; ++k) {
107  tmp += block[i + k] * coefficients[k * 8 + j];
108  }
109  out[i + j] = tmp;
110  }
111  }
112 
113  /* block = (coefficients') * out */
114  for (i = 0; i < 8; ++i) {
115  for (j = 0; j < 8; ++j) {
116  double tmp = 0;
117  for (k = 0; k < 64; k += 8) {
118  tmp += coefficients[k + i] * out[k + j];
119  }
120  block[i * 8 + j] = floor(tmp + 0.5);
121  }
122  }
123 }
out
FILE * out
Definition: movenc.c:54
ff_ref_dct_init
av_cold void ff_ref_dct_init(void)
Initialize the double precision discrete cosine transform functions fdct & idct.
Definition: dctref.c:41
tmp
static uint8_t tmp[11]
Definition: aes_ctr.c:26
mathematics.h
av_cold
#define av_cold
Definition: attributes.h:84
coefficients
static double coefficients[8 *8]
Definition: dctref.c:35
M_PI
#define M_PI
Definition: mathematics.h:52
i
#define i(width, name, range_min, range_max)
Definition: cbs_h2645.c:259
ff_ref_fdct
void ff_ref_fdct(short *block)
Transform 8x8 block of data with a double precision forward DCT This is a reference implementation.
Definition: dctref.c:59
dctref.h
ff_ref_idct
void ff_ref_idct(short *block)
Transform 8x8 block of data with a double precision inverse DCT This is a reference implementation.
Definition: dctref.c:95
block
The exact code depends on how similar the blocks are and how related they are to the block
Definition: filter_design.txt:207