blob: da43e8a3998fd98725ddafa3e2c7d5feb0241221 [file] [log] [blame]
Yushin Cho77bba8d2016-11-04 16:36:56 -07001/*
2 * Copyright (c) 2001-2016, Alliance for Open Media. All rights reserved
3 *
4 * This source code is subject to the terms of the BSD 2 Clause License and
5 * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
6 * was not distributed with this source code in the LICENSE file, you can
7 * obtain it at www.aomedia.org/license/software. If the Alliance for Open
8 * Media Patent License 1.0 was not distributed with this source code in the
9 * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
10 */
Nathan E. Egge1078dee2016-03-06 10:59:29 -050011
Sebastien Alaiwan0d52f122018-03-06 09:54:23 +010012#include <assert.h>
Yaowu Xu931bc2a2016-10-14 13:53:51 -070013#include "aom_dsp/entdec.h"
Thomas Daede837262b2017-11-06 20:07:01 -080014#include "aom_dsp/prob.h"
Nathan E. Egge1078dee2016-03-06 10:59:29 -050015
16/*A range decoder.
17 This is an entropy decoder based upon \cite{Mar79}, which is itself a
18 rediscovery of the FIFO arithmetic code introduced by \cite{Pas76}.
19 It is very similar to arithmetic encoding, except that encoding is done with
20 digits in any base, instead of with bits, and so it is faster when using
21 larger bases (i.e.: a byte).
22 The author claims an average waste of $\frac{1}{2}\log_b(2b)$ bits, where $b$
23 is the base, longer than the theoretical optimum, but to my knowledge there
24 is no published justification for this claim.
25 This only seems true when using near-infinite precision arithmetic so that
26 the process is carried out with no rounding errors.
27
28 An excellent description of implementation details is available at
29 http://www.arturocampos.com/ac_range.html
30 A recent work \cite{MNW98} which proposes several changes to arithmetic
31 encoding for efficiency actually re-discovers many of the principles
32 behind range encoding, and presents a good theoretical analysis of them.
33
34 End of stream is handled by writing out the smallest number of bits that
35 ensures that the stream will be correctly decoded regardless of the value of
36 any subsequent bits.
37 od_ec_dec_tell() can be used to determine how many bits were needed to decode
38 all the symbols thus far; other data can be packed in the remaining bits of
39 the input buffer.
40 @PHDTHESIS{Pas76,
41 author="Richard Clark Pasco",
42 title="Source coding algorithms for fast data compression",
43 school="Dept. of Electrical Engineering, Stanford University",
44 address="Stanford, CA",
45 month=May,
46 year=1976,
47 URL="http://www.richpasco.org/scaffdc.pdf"
48 }
49 @INPROCEEDINGS{Mar79,
50 author="Martin, G.N.N.",
51 title="Range encoding: an algorithm for removing redundancy from a digitised
52 message",
53 booktitle="Video & Data Recording Conference",
54 year=1979,
55 address="Southampton",
56 month=Jul,
57 URL="http://www.compressconsult.com/rangecoder/rngcod.pdf.gz"
58 }
59 @ARTICLE{MNW98,
60 author="Alistair Moffat and Radford Neal and Ian H. Witten",
61 title="Arithmetic Coding Revisited",
62 journal="{ACM} Transactions on Information Systems",
63 year=1998,
64 volume=16,
65 number=3,
66 pages="256--294",
67 month=Jul,
68 URL="http://researchcommons.waikato.ac.nz/bitstream/handle/10289/78/content.pdf"
69 }*/
70
Nathan E. Egge1078dee2016-03-06 10:59:29 -050071/*This is meant to be a large, positive constant that can still be efficiently
72 loaded as an immediate (on platforms like ARM, for example).
73 Even relatively modest values like 100 would work fine.*/
74#define OD_EC_LOTS_OF_BITS (0x4000)
75
Wan-Teh Changc8ae5682018-05-08 12:15:36 -070076/*The return value of od_ec_dec_tell does not change across an od_ec_dec_refill
77 call.*/
Nathan E. Egge1078dee2016-03-06 10:59:29 -050078static void od_ec_dec_refill(od_ec_dec *dec) {
79 int s;
80 od_ec_window dif;
81 int16_t cnt;
82 const unsigned char *bptr;
83 const unsigned char *end;
84 dif = dec->dif;
85 cnt = dec->cnt;
86 bptr = dec->bptr;
87 end = dec->end;
88 s = OD_EC_WINDOW_SIZE - 9 - (cnt + 15);
89 for (; s >= 0 && bptr < end; s -= 8, bptr++) {
Wan-Teh Changed997ab2018-08-14 12:35:17 -070090 /*Each time a byte is inserted into the window (dif), bptr advances and cnt
91 is incremented by 8, so the total number of consumed bits (the return
92 value of od_ec_dec_tell) does not change.*/
Sebastien Alaiwan0d52f122018-03-06 09:54:23 +010093 assert(s <= OD_EC_WINDOW_SIZE - 8);
Timothy B. Terriberry881f1092017-03-07 20:03:09 -080094 dif ^= (od_ec_window)bptr[0] << s;
Nathan E. Egge1078dee2016-03-06 10:59:29 -050095 cnt += 8;
96 }
97 if (bptr >= end) {
Wan-Teh Changed997ab2018-08-14 12:35:17 -070098 /*We've reached the end of the buffer. It is perfectly valid for us to need
99 to fill the window with additional bits past the end of the buffer (and
100 this happens in normal operation). These bits should all just be taken
101 as zero. But we cannot increment bptr past 'end' (this is undefined
102 behavior), so we start to increment dec->tell_offs. We also don't want
103 to keep testing bptr against 'end', so we set cnt to OD_EC_LOTS_OF_BITS
104 and adjust dec->tell_offs so that the total number of unconsumed bits in
105 the window (dec->cnt - dec->tell_offs) does not change. This effectively
106 puts lots of zero bits into the window, and means we won't try to refill
107 it from the buffer for a very long time (at which point we'll put lots
108 of zero bits into the window again).*/
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500109 dec->tell_offs += OD_EC_LOTS_OF_BITS - cnt;
110 cnt = OD_EC_LOTS_OF_BITS;
111 }
112 dec->dif = dif;
113 dec->cnt = cnt;
114 dec->bptr = bptr;
115}
116
117/*Takes updated dif and range values, renormalizes them so that
118 32768 <= rng < 65536 (reading more bytes from the stream into dif if
119 necessary), and stores them back in the decoder context.
120 dif: The new value of dif.
121 rng: The new value of the range.
122 ret: The value to return.
123 Return: ret.
124 This allows the compiler to jump to this function via a tail-call.*/
Michael Bebenita63b44c42016-08-23 16:03:39 -0700125static int od_ec_dec_normalize(od_ec_dec *dec, od_ec_window dif, unsigned rng,
126 int ret) {
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500127 int d;
Sebastien Alaiwan0d52f122018-03-06 09:54:23 +0100128 assert(rng <= 65535U);
Wan-Teh Changed997ab2018-08-14 12:35:17 -0700129 /*The number of leading zeros in the 16-bit binary representation of rng.*/
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500130 d = 16 - OD_ILOG_NZ(rng);
Wan-Teh Changed997ab2018-08-14 12:35:17 -0700131 /*d bits in dec->dif are consumed.*/
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500132 dec->cnt -= d;
Timothy B. Terriberry881f1092017-03-07 20:03:09 -0800133 /*This is equivalent to shifting in 1's instead of 0's.*/
134 dec->dif = ((dif + 1) << d) - 1;
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500135 dec->rng = rng << d;
136 if (dec->cnt < 0) od_ec_dec_refill(dec);
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500137 return ret;
138}
139
140/*Initializes the decoder.
141 buf: The input buffer to use.
Wan-Teh Changc0b290f2018-08-15 16:12:56 -0700142 storage: The size in bytes of the input buffer.*/
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500143void od_ec_dec_init(od_ec_dec *dec, const unsigned char *buf,
144 uint32_t storage) {
145 dec->buf = buf;
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500146 dec->tell_offs = 10 - (OD_EC_WINDOW_SIZE - 8);
147 dec->end = buf + storage;
148 dec->bptr = buf;
Timothy B. Terriberry881f1092017-03-07 20:03:09 -0800149 dec->dif = ((od_ec_window)1 << (OD_EC_WINDOW_SIZE - 1)) - 1;
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500150 dec->rng = 0x8000;
151 dec->cnt = -15;
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500152 od_ec_dec_refill(dec);
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500153}
154
Timothy B. Terriberryead52872017-03-07 20:27:34 -0800155/*Decode a single binary value.
Timothy B. Terriberryf9ef4f62017-08-25 11:24:18 -0700156 f: The probability that the bit is one, scaled by 32768.
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500157 Return: The value decoded (0 or 1).*/
Timothy B. Terriberryead52872017-03-07 20:27:34 -0800158int od_ec_decode_bool_q15(od_ec_dec *dec, unsigned f) {
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500159 od_ec_window dif;
160 od_ec_window vw;
161 unsigned r;
Nathan E. Egge5357dca2016-09-09 14:21:56 -0400162 unsigned r_new;
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500163 unsigned v;
164 int ret;
Sebastien Alaiwan0d52f122018-03-06 09:54:23 +0100165 assert(0 < f);
166 assert(f < 32768U);
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500167 dif = dec->dif;
168 r = dec->rng;
Sebastien Alaiwan0d52f122018-03-06 09:54:23 +0100169 assert(dif >> (OD_EC_WINDOW_SIZE - 16) < r);
170 assert(32768U <= r);
Thomas Davies736ddef2017-11-09 09:46:08 +0000171 v = ((r >> 8) * (uint32_t)(f >> EC_PROB_SHIFT) >> (7 - EC_PROB_SHIFT));
Jonathan Matthews9ade3942017-11-23 08:44:07 +0000172 v += EC_MIN_PROB;
Timothy B. Terriberry881f1092017-03-07 20:03:09 -0800173 vw = (od_ec_window)v << (OD_EC_WINDOW_SIZE - 16);
174 ret = 1;
175 r_new = v;
176 if (dif >= vw) {
177 r_new = r - v;
178 dif -= vw;
179 ret = 0;
180 }
Nathan E. Egge5357dca2016-09-09 14:21:56 -0400181 return od_ec_dec_normalize(dec, dif, r_new, ret);
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500182}
183
Timothy B. Terriberryf9ef4f62017-08-25 11:24:18 -0700184/*Decodes a symbol given an inverse cumulative distribution function (CDF)
185 table in Q15.
Thomas Daede837262b2017-11-06 20:07:01 -0800186 icdf: CDF_PROB_TOP minus the CDF, such that symbol s falls in the range
187 [s > 0 ? (CDF_PROB_TOP - icdf[s - 1]) : 0, CDF_PROB_TOP - icdf[s]).
Timothy B. Terriberryf9ef4f62017-08-25 11:24:18 -0700188 The values must be monotonically non-increasing, and icdf[nsyms - 1]
189 must be 0.
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500190 nsyms: The number of symbols in the alphabet.
191 This should be at most 16.
192 Return: The decoded symbol s.*/
Timothy B. Terriberryf9ef4f62017-08-25 11:24:18 -0700193int od_ec_decode_cdf_q15(od_ec_dec *dec, const uint16_t *icdf, int nsyms) {
Timothy B. Terriberry561eb7c2017-03-07 18:06:44 -0800194 od_ec_window dif;
195 unsigned r;
196 unsigned c;
197 unsigned u;
198 unsigned v;
199 int ret;
200 (void)nsyms;
201 dif = dec->dif;
202 r = dec->rng;
Thomas Davies736ddef2017-11-09 09:46:08 +0000203 const int N = nsyms - 1;
204
Sebastien Alaiwan0d52f122018-03-06 09:54:23 +0100205 assert(dif >> (OD_EC_WINDOW_SIZE - 16) < r);
206 assert(icdf[nsyms - 1] == OD_ICDF(CDF_PROB_TOP));
207 assert(32768U <= r);
208 assert(7 - EC_PROB_SHIFT - CDF_SHIFT >= 0);
Timothy B. Terriberry881f1092017-03-07 20:03:09 -0800209 c = (unsigned)(dif >> (OD_EC_WINDOW_SIZE - 16));
210 v = r;
211 ret = -1;
212 do {
213 u = v;
Thomas Davies736ddef2017-11-09 09:46:08 +0000214 v = ((r >> 8) * (uint32_t)(icdf[++ret] >> EC_PROB_SHIFT) >>
Thomas Daede837262b2017-11-06 20:07:01 -0800215 (7 - EC_PROB_SHIFT - CDF_SHIFT));
Thomas Davies736ddef2017-11-09 09:46:08 +0000216 v += EC_MIN_PROB * (N - ret);
Timothy B. Terriberry881f1092017-03-07 20:03:09 -0800217 } while (c < v);
Sebastien Alaiwan0d52f122018-03-06 09:54:23 +0100218 assert(v < u);
219 assert(u <= r);
Timothy B. Terriberry881f1092017-03-07 20:03:09 -0800220 r = u - v;
221 dif -= (od_ec_window)v << (OD_EC_WINDOW_SIZE - 16);
Timothy B. Terriberry561eb7c2017-03-07 18:06:44 -0800222 return od_ec_dec_normalize(dec, dif, r, ret);
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500223}
224
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500225/*Returns the number of bits "used" by the decoded symbols so far.
226 This same number can be computed in either the encoder or the decoder, and is
227 suitable for making coding decisions.
228 Return: The number of bits.
229 This will always be slightly larger than the exact value (e.g., all
230 rounding error is in the positive direction).*/
Nathan E. Egge19698a72016-08-18 02:34:53 -0400231int od_ec_dec_tell(const od_ec_dec *dec) {
Wan-Teh Changed997ab2018-08-14 12:35:17 -0700232 /*There is a window of bits stored in dec->dif. The difference
233 (dec->bptr - dec->buf) tells us how many bytes have been read into this
234 window. The difference (dec->cnt - dec->tell_offs) tells us how many of
235 the bits in that window remain unconsumed.*/
Wan-Teh Changc8ae5682018-05-08 12:15:36 -0700236 return (int)((dec->bptr - dec->buf) * 8 - dec->cnt + dec->tell_offs);
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500237}
238
239/*Returns the number of bits "used" by the decoded symbols so far.
240 This same number can be computed in either the encoder or the decoder, and is
241 suitable for making coding decisions.
242 Return: The number of bits scaled by 2**OD_BITRES.
243 This will always be slightly larger than the exact value (e.g., all
244 rounding error is in the positive direction).*/
Nathan E. Egge19698a72016-08-18 02:34:53 -0400245uint32_t od_ec_dec_tell_frac(const od_ec_dec *dec) {
Nathan E. Egge1078dee2016-03-06 10:59:29 -0500246 return od_ec_tell_frac(od_ec_dec_tell(dec), dec->rng);
247}