[cvs] / xvidcore / src / dct / idct.c Repository:
ViewVC logotype

Annotation of /xvidcore/src/dct/idct.c

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1.3 - (view) (download)

1 : chl 1.3 /*****************************************************************************
2 :     *
3 :     * XVID MPEG-4 VIDEO CODEC
4 :     * - inverse fast disrete cosine transformation - integer C version
5 :     *
6 :     * These routines are from Independent JPEG Group's free JPEG software
7 :     * Copyright (C) 1991-1998, Thomas G. Lane (see the file README.IJG)
8 :     *
9 :     * 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
11 :     * to use this software module in hardware or software products are
12 :     * advised that its use may infringe existing patents or copyrights, and
13 :     * any such use would be at such party's own risk. The original
14 :     * developer of this software module and his/her company, and subsequent
15 :     * editors and their companies, will have no liability for use of this
16 :     * software or modifications or derivatives thereof.
17 :     *
18 :     * This program is free software; you can redistribute it and/or modify
19 :     * it under the terms of the GNU General Public License as published by
20 :     * the Free Software Foundation; either version 2 of the License, or
21 :     * (at your option) any later version.
22 :     *
23 :     * This program is distributed in the hope that it will be useful,
24 :     * but WITHOUT ANY WARRANTY; without even the implied warranty of
25 :     * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
26 :     * GNU General Public License for more details.
27 :     *
28 :     * You should have received a copy of the GNU General Public License
29 :     * along with this program; if not, write to the Free Software
30 :     * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
31 : Isibaar 1.1 *
32 : chl 1.3 *************************************************************************/
33 : Isibaar 1.1
34 :     /**********************************************************/
35 :     /* inverse two dimensional DCT, Chen-Wang algorithm */
36 :     /* (cf. IEEE ASSP-32, pp. 803-816, Aug. 1984) */
37 :     /* 32-bit integer arithmetic (8 bit coefficients) */
38 :     /* 11 mults, 29 adds per DCT */
39 :     /* sE, 18.8.91 */
40 :     /**********************************************************/
41 :     /* coefficients extended to 12 bit for IEEE1180-1990 */
42 :     /* compliance sE, 2.1.94 */
43 :     /**********************************************************/
44 :    
45 :     /* this code assumes >> to be a two's-complement arithmetic */
46 :     /* right shift: (-2)>>1 == -1 , (-3)>>1 == -2 */
47 :    
48 :     #include "idct.h"
49 :    
50 : edgomez 1.2 #define W1 2841 /* 2048*sqrt(2)*cos(1*pi/16) */
51 :     #define W2 2676 /* 2048*sqrt(2)*cos(2*pi/16) */
52 :     #define W3 2408 /* 2048*sqrt(2)*cos(3*pi/16) */
53 :     #define W5 1609 /* 2048*sqrt(2)*cos(5*pi/16) */
54 :     #define W6 1108 /* 2048*sqrt(2)*cos(6*pi/16) */
55 :     #define W7 565 /* 2048*sqrt(2)*cos(7*pi/16) */
56 : Isibaar 1.1
57 :    
58 :     /* global declarations */
59 :     //void init_idct_int32 (void);
60 :     //void idct_int32 (short *block);
61 :    
62 :     /* private data */
63 : edgomez 1.2 static short iclip[1024]; /* clipping table */
64 : Isibaar 1.1 static short *iclp;
65 :    
66 :     /* private prototypes */
67 :     //static void idctrow _ANSI_ARGS_((short *blk));
68 :     //static void idctcol _ANSI_ARGS_((short *blk));
69 :    
70 :     /* row (horizontal) IDCT
71 :     *
72 :     * 7 pi 1
73 :     * dst[k] = sum c[l] * src[l] * cos( -- * ( k + - ) * l )
74 :     * l=0 8 2
75 :     *
76 :     * where: c[0] = 128
77 :     * c[1..7] = 128*sqrt(2)
78 :     */
79 :    
80 :     /*
81 :     static void idctrow(blk)
82 :     short *blk;
83 :     {
84 :     int X0, X1, X2, X3, X4, X5, X6, X7, X8;
85 :    
86 :     // shortcut
87 :     if (!((X1 = blk[4]<<11) | (X2 = blk[6]) | (X3 = blk[2]) |
88 :     (X4 = blk[1]) | (X5 = blk[7]) | (X6 = blk[5]) | (X7 = blk[3])))
89 :     {
90 :     blk[0]=blk[1]=blk[2]=blk[3]=blk[4]=blk[5]=blk[6]=blk[7]=blk[0]<<3;
91 :     return;
92 :     }
93 :    
94 :     X0 = (blk[0]<<11) + 128; // for proper rounding in the fourth stage
95 :    
96 :     // first stage
97 :     X8 = W7*(X4+X5);
98 :     X4 = X8 + (W1-W7)*X4;
99 :     X5 = X8 - (W1+W7)*X5;
100 :     X8 = W3*(X6+X7);
101 :     X6 = X8 - (W3-W5)*X6;
102 :     X7 = X8 - (W3+W5)*X7;
103 :    
104 :     // second stage
105 :     X8 = X0 + X1;
106 :     X0 -= X1;
107 :     X1 = W6*(X3+X2);
108 :     X2 = X1 - (W2+W6)*X2;
109 :     X3 = X1 + (W2-W6)*X3;
110 :     X1 = X4 + X6;
111 :     X4 -= X6;
112 :     X6 = X5 + X7;
113 :     X5 -= X7;
114 :    
115 :     // third stage
116 :     X7 = X8 + X3;
117 :     X8 -= X3;
118 :     X3 = X0 + X2;
119 :     X0 -= X2;
120 :     X2 = (181*(X4+X5)+128)>>8;
121 :     X4 = (181*(X4-X5)+128)>>8;
122 :    
123 :     // fourth stage
124 :     blk[0] = (X7+X1)>>8;
125 :     blk[1] = (X3+X2)>>8;
126 :     blk[2] = (X0+X4)>>8;
127 :     blk[3] = (X8+X6)>>8;
128 :     blk[4] = (X8-X6)>>8;
129 :     blk[5] = (X0-X4)>>8;
130 :     blk[6] = (X3-X2)>>8;
131 :     blk[7] = (X7-X1)>>8;
132 :     }*/
133 :    
134 :     /* column (vertical) IDCT
135 :     *
136 :     * 7 pi 1
137 :     * dst[8*k] = sum c[l] * src[8*l] * cos( -- * ( k + - ) * l )
138 :     * l=0 8 2
139 :     *
140 :     * where: c[0] = 1/1024
141 :     * c[1..7] = (1/1024)*sqrt(2)
142 :     */
143 :     /*
144 :     static void idctcol(blk)
145 :     short *blk;
146 :     {
147 :     int X0, X1, X2, X3, X4, X5, X6, X7, X8;
148 :    
149 :     // shortcut
150 :     if (!((X1 = (blk[8*4]<<8)) | (X2 = blk[8*6]) | (X3 = blk[8*2]) |
151 :     (X4 = blk[8*1]) | (X5 = blk[8*7]) | (X6 = blk[8*5]) | (X7 = blk[8*3])))
152 :     {
153 :     blk[8*0]=blk[8*1]=blk[8*2]=blk[8*3]=blk[8*4]=blk[8*5]=blk[8*6]=blk[8*7]=
154 :     iclp[(blk[8*0]+32)>>6];
155 :     return;
156 :     }
157 :    
158 :     X0 = (blk[8*0]<<8) + 8192;
159 :    
160 :     // first stage
161 :     X8 = W7*(X4+X5) + 4;
162 :     X4 = (X8+(W1-W7)*X4)>>3;
163 :     X5 = (X8-(W1+W7)*X5)>>3;
164 :     X8 = W3*(X6+X7) + 4;
165 :     X6 = (X8-(W3-W5)*X6)>>3;
166 :     X7 = (X8-(W3+W5)*X7)>>3;
167 :    
168 :     // second stage
169 :     X8 = X0 + X1;
170 :     X0 -= X1;
171 :     X1 = W6*(X3+X2) + 4;
172 :     X2 = (X1-(W2+W6)*X2)>>3;
173 :     X3 = (X1+(W2-W6)*X3)>>3;
174 :     X1 = X4 + X6;
175 :     X4 -= X6;
176 :     X6 = X5 + X7;
177 :     X5 -= X7;
178 :    
179 :     // third stage
180 :     X7 = X8 + X3;
181 :     X8 -= X3;
182 :     X3 = X0 + X2;
183 :     X0 -= X2;
184 :     X2 = (181*(X4+X5)+128)>>8;
185 :     X4 = (181*(X4-X5)+128)>>8;
186 :    
187 :     // fourth stage
188 :     blk[8*0] = iclp[(X7+X1)>>14];
189 :     blk[8*1] = iclp[(X3+X2)>>14];
190 :     blk[8*2] = iclp[(X0+X4)>>14];
191 :     blk[8*3] = iclp[(X8+X6)>>14];
192 :     blk[8*4] = iclp[(X8-X6)>>14];
193 :     blk[8*5] = iclp[(X0-X4)>>14];
194 :     blk[8*6] = iclp[(X3-X2)>>14];
195 :     blk[8*7] = iclp[(X7-X1)>>14];
196 :     }*/
197 :    
198 :     // function pointer
199 :     idctFuncPtr idct;
200 :    
201 :     /* two dimensional inverse discrete cosine transform */
202 :     //void j_rev_dct(block)
203 :     //short *block;
204 : edgomez 1.2 void
205 :     idct_int32(short *const block)
206 : Isibaar 1.1 {
207 :    
208 : edgomez 1.2 // idct_int32_init() must be called before the first call to this function!
209 : Isibaar 1.1
210 :    
211 : edgomez 1.2 /*int i;
212 :     long i;
213 : Isibaar 1.1
214 : edgomez 1.2 for (i=0; i<8; i++)
215 :     idctrow(block+8*i);
216 : Isibaar 1.1
217 : edgomez 1.2 for (i=0; i<8; i++)
218 :     idctcol(block+i); */
219 :     static short *blk;
220 :     static long i;
221 :     static long X0, X1, X2, X3, X4, X5, X6, X7, X8;
222 : Isibaar 1.1
223 :    
224 : edgomez 1.2 for (i = 0; i < 8; i++) // idct rows
225 : Isibaar 1.1 {
226 : edgomez 1.2 blk = block + (i << 3);
227 :     if (!
228 :     ((X1 = blk[4] << 11) | (X2 = blk[6]) | (X3 = blk[2]) | (X4 =
229 :     blk[1]) |
230 :     (X5 = blk[7]) | (X6 = blk[5]) | (X7 = blk[3]))) {
231 :     blk[0] = blk[1] = blk[2] = blk[3] = blk[4] = blk[5] = blk[6] =
232 :     blk[7] = blk[0] << 3;
233 :     continue;
234 :     }
235 :    
236 :     X0 = (blk[0] << 11) + 128; // for proper rounding in the fourth stage
237 :    
238 :     // first stage
239 :     X8 = W7 * (X4 + X5);
240 :     X4 = X8 + (W1 - W7) * X4;
241 :     X5 = X8 - (W1 + W7) * X5;
242 :     X8 = W3 * (X6 + X7);
243 :     X6 = X8 - (W3 - W5) * X6;
244 :     X7 = X8 - (W3 + W5) * X7;
245 :    
246 :     // second stage
247 :     X8 = X0 + X1;
248 :     X0 -= X1;
249 :     X1 = W6 * (X3 + X2);
250 :     X2 = X1 - (W2 + W6) * X2;
251 :     X3 = X1 + (W2 - W6) * X3;
252 :     X1 = X4 + X6;
253 :     X4 -= X6;
254 :     X6 = X5 + X7;
255 :     X5 -= X7;
256 :    
257 :     // third stage
258 :     X7 = X8 + X3;
259 :     X8 -= X3;
260 :     X3 = X0 + X2;
261 :     X0 -= X2;
262 :     X2 = (181 * (X4 + X5) + 128) >> 8;
263 :     X4 = (181 * (X4 - X5) + 128) >> 8;
264 :    
265 :     // fourth stage
266 :    
267 :     blk[0] = (short) ((X7 + X1) >> 8);
268 :     blk[1] = (short) ((X3 + X2) >> 8);
269 :     blk[2] = (short) ((X0 + X4) >> 8);
270 :     blk[3] = (short) ((X8 + X6) >> 8);
271 :     blk[4] = (short) ((X8 - X6) >> 8);
272 :     blk[5] = (short) ((X0 - X4) >> 8);
273 :     blk[6] = (short) ((X3 - X2) >> 8);
274 :     blk[7] = (short) ((X7 - X1) >> 8);
275 : Isibaar 1.1
276 : edgomez 1.2 } // end for ( i = 0; i < 8; ++i ) IDCT-rows
277 : Isibaar 1.1
278 :    
279 :    
280 : edgomez 1.2 for (i = 0; i < 8; i++) // idct columns
281 : Isibaar 1.1 {
282 : edgomez 1.2 blk = block + i;
283 :     // shortcut
284 :     if (!
285 :     ((X1 = (blk[8 * 4] << 8)) | (X2 = blk[8 * 6]) | (X3 =
286 :     blk[8 *
287 :     2]) | (X4 =
288 :     blk[8 *
289 :     1])
290 :     | (X5 = blk[8 * 7]) | (X6 = blk[8 * 5]) | (X7 = blk[8 * 3]))) {
291 :     blk[8 * 0] = blk[8 * 1] = blk[8 * 2] = blk[8 * 3] = blk[8 * 4] =
292 :     blk[8 * 5] = blk[8 * 6] = blk[8 * 7] =
293 :     iclp[(blk[8 * 0] + 32) >> 6];
294 :     continue;
295 :     }
296 :    
297 :     X0 = (blk[8 * 0] << 8) + 8192;
298 :    
299 :     // first stage
300 :     X8 = W7 * (X4 + X5) + 4;
301 :     X4 = (X8 + (W1 - W7) * X4) >> 3;
302 :     X5 = (X8 - (W1 + W7) * X5) >> 3;
303 :     X8 = W3 * (X6 + X7) + 4;
304 :     X6 = (X8 - (W3 - W5) * X6) >> 3;
305 :     X7 = (X8 - (W3 + W5) * X7) >> 3;
306 :    
307 :     // second stage
308 :     X8 = X0 + X1;
309 :     X0 -= X1;
310 :     X1 = W6 * (X3 + X2) + 4;
311 :     X2 = (X1 - (W2 + W6) * X2) >> 3;
312 :     X3 = (X1 + (W2 - W6) * X3) >> 3;
313 :     X1 = X4 + X6;
314 :     X4 -= X6;
315 :     X6 = X5 + X7;
316 :     X5 -= X7;
317 :    
318 :     // third stage
319 :     X7 = X8 + X3;
320 :     X8 -= X3;
321 :     X3 = X0 + X2;
322 :     X0 -= X2;
323 :     X2 = (181 * (X4 + X5) + 128) >> 8;
324 :     X4 = (181 * (X4 - X5) + 128) >> 8;
325 :    
326 :     // fourth stage
327 :     blk[8 * 0] = iclp[(X7 + X1) >> 14];
328 :     blk[8 * 1] = iclp[(X3 + X2) >> 14];
329 :     blk[8 * 2] = iclp[(X0 + X4) >> 14];
330 :     blk[8 * 3] = iclp[(X8 + X6) >> 14];
331 :     blk[8 * 4] = iclp[(X8 - X6) >> 14];
332 :     blk[8 * 5] = iclp[(X0 - X4) >> 14];
333 :     blk[8 * 6] = iclp[(X3 - X2) >> 14];
334 :     blk[8 * 7] = iclp[(X7 - X1) >> 14];
335 : Isibaar 1.1 }
336 :    
337 : edgomez 1.2 } // end function idct_int32(block)
338 : Isibaar 1.1
339 :    
340 :     //void
341 :     //idct_int32_init()
342 : edgomez 1.2 void
343 :     idct_int32_init()
344 : Isibaar 1.1 {
345 : edgomez 1.2 int i;
346 : Isibaar 1.1
347 : edgomez 1.2 iclp = iclip + 512;
348 :     for (i = -512; i < 512; i++)
349 :     iclp[i] = (i < -256) ? -256 : ((i > 255) ? 255 : i);
350 : Isibaar 1.1 }

No admin address has been configured
ViewVC Help
Powered by ViewVC 1.0.4