summaryrefslogtreecommitdiff
path: root/libavcodec/dct.c
blob: dab94c30a005b78610eda4fbfd70b2a8e0e355bd (plain)
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
/*
 * (I)DCT Transforms
 * Copyright (c) 2009 Peter Ross <pross@xvid.org>
 * Copyright (c) 2010 Alex Converse <alex.converse@gmail.com>
 * Copyright (c) 2010 Vitor Sessak
 *
 * This file is part of FFmpeg.
 *
 * FFmpeg is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * FFmpeg is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with FFmpeg; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 */

/**
 * @file
 * (Inverse) Discrete Cosine Transforms. These are also known as the
 * type II and type III DCTs respectively.
 */

#include <math.h>
#include "libavutil/mathematics.h"
#include "fft.h"
#include "x86/fft.h"

#define DCT32_FLOAT
#include "dct32.c"

/* sin((M_PI * x / (2*n)) */
#define SIN(s,n,x) (s->costab[(n) - (x)])

/* cos((M_PI * x / (2*n)) */
#define COS(s,n,x) (s->costab[x])

static void ff_dst_calc_I_c(DCTContext *ctx, FFTSample *data)
{
    int n = 1 << ctx->nbits;
    int i;

    data[0] = 0;
    for(i = 1; i < n/2; i++) {
        float tmp1 = data[i    ];
        float tmp2 = data[n - i];
        float s = SIN(ctx, n, 2*i);

        s *= tmp1 + tmp2;
        tmp1 = (tmp1 - tmp2) * 0.5f;
        data[i    ] = s + tmp1;
        data[n - i] = s - tmp1;
    }

    data[n/2] *= 2;
    ff_rdft_calc(&ctx->rdft, data);

    data[0] *= 0.5f;

    for(i = 1; i < n-2; i += 2) {
        data[i + 1] += data[i - 1];
        data[i    ] = -data[i + 2];
    }

    data[n-1] = 0;
}

static void ff_dct_calc_I_c(DCTContext *ctx, FFTSample *data)
{
    int n = 1 << ctx->nbits;
    int i;
    float next = -0.5f * (data[0] - data[n]);

    for(i = 0; i < n/2; i++) {
        float tmp1 = data[i    ];
        float tmp2 = data[n - i];
        float s = SIN(ctx, n, 2*i);
        float c = COS(ctx, n, 2*i);

        c *= tmp1 - tmp2;
        s *= tmp1 - tmp2;

        next += c;

        tmp1 = (tmp1 + tmp2) * 0.5f;
        data[i    ] = tmp1 - s;
        data[n - i] = tmp1 + s;
    }

    ff_rdft_calc(&ctx->rdft, data);
    data[n] = data[1];
    data[1] = next;

    for(i = 3; i <= n; i += 2)
        data[i] = data[i - 2] - data[i];
}

static void ff_dct_calc_III_c(DCTContext *ctx, FFTSample *data)
{
    int n = 1 << ctx->nbits;
    int i;

    float next = data[n - 1];
    float inv_n = 1.0f / n;

    for (i = n - 2; i >= 2; i -= 2) {
        float val1 = data[i    ];
        float val2 = data[i - 1] - data[i + 1];
        float c = COS(ctx, n, i);
        float s = SIN(ctx, n, i);

        data[i    ] = c * val1 + s * val2;
        data[i + 1] = s * val1 - c * val2;
    }

    data[1] = 2 * next;

    ff_rdft_calc(&ctx->rdft, data);

    for (i = 0; i < n / 2; i++) {
        float tmp1 = data[i        ] * inv_n;
        float tmp2 = data[n - i - 1] * inv_n;
        float csc = ctx->csc2[i] * (tmp1 - tmp2);

        tmp1 += tmp2;
        data[i        ] = tmp1 + csc;
        data[n - i - 1] = tmp1 - csc;
    }
}

static void ff_dct_calc_II_c(DCTContext *ctx, FFTSample *data)
{
    int n = 1 << ctx->nbits;
    int i;
    float next;

    for (i=0; i < n/2; i++) {
        float tmp1 = data[i        ];
        float tmp2 = data[n - i - 1];
        float s = SIN(ctx, n, 2*i + 1);

        s *= tmp1 - tmp2;
        tmp1 = (tmp1 + tmp2) * 0.5f;

        data[i    ] = tmp1 + s;
        data[n-i-1] = tmp1 - s;
    }

    ff_rdft_calc(&ctx->rdft, data);

    next = data[1] * 0.5;
    data[1] *= -1;

    for (i = n - 2; i >= 0; i -= 2) {
        float inr = data[i    ];
        float ini = data[i + 1];
        float c = COS(ctx, n, i);
        float s = SIN(ctx, n, i);

        data[i  ] = c * inr + s * ini;

        data[i+1] = next;

        next +=     s * inr - c * ini;
    }
}

static void dct32_func(DCTContext *ctx, FFTSample *data)
{
    ctx->dct32(data, data);
}

void ff_dct_calc(DCTContext *s, FFTSample *data)
{
    s->dct_calc(s, data);
}

av_cold int ff_dct_init(DCTContext *s, int nbits, enum DCTTransformType inverse)
{
    int n = 1 << nbits;
    int i;

    s->nbits    = nbits;
    s->inverse  = inverse;

    ff_init_ff_cos_tabs(nbits+2);

    s->costab = ff_cos_tabs[nbits+2];

    s->csc2 = av_malloc(n/2 * sizeof(FFTSample));

    if (ff_rdft_init(&s->rdft, nbits, inverse == DCT_III) < 0) {
        av_free(s->csc2);
        return -1;
    }

    for (i = 0; i < n/2; i++)
        s->csc2[i] = 0.5 / sin((M_PI / (2*n) * (2*i + 1)));

    switch(inverse) {
    case DCT_I  : s->dct_calc = ff_dct_calc_I_c; break;
    case DCT_II : s->dct_calc = ff_dct_calc_II_c ; break;
    case DCT_III: s->dct_calc = ff_dct_calc_III_c; break;
    case DST_I  : s->dct_calc = ff_dst_calc_I_c; break;
    }

    if (inverse == DCT_II && nbits == 5)
        s->dct_calc = dct32_func;

    s->dct32 = dct32;
    if (HAVE_MMX)     ff_dct_init_mmx(s);

    return 0;
}

av_cold void ff_dct_end(DCTContext *s)
{
    ff_rdft_end(&s->rdft);
    av_free(s->csc2);
}