52 |
/* this code assumes >> to be a two's-complement arithmetic */ |
/* this code assumes >> to be a two's-complement arithmetic */ |
53 |
/* right shift: (-2)>>1 == -1 , (-3)>>1 == -2 */ |
/* right shift: (-2)>>1 == -1 , (-3)>>1 == -2 */ |
54 |
|
|
|
//#include <windows.h> |
|
55 |
#include "idct.h" |
#include "idct.h" |
56 |
|
|
57 |
#define W1 2841 /* 2048*sqrt(2)*cos(1*pi/16) */ |
#define W1 2841 /* 2048*sqrt(2)*cos(1*pi/16) */ |
61 |
#define W6 1108 /* 2048*sqrt(2)*cos(6*pi/16) */ |
#define W6 1108 /* 2048*sqrt(2)*cos(6*pi/16) */ |
62 |
#define W7 565 /* 2048*sqrt(2)*cos(7*pi/16) */ |
#define W7 565 /* 2048*sqrt(2)*cos(7*pi/16) */ |
63 |
|
|
|
|
|
|
/* global declarations */ |
|
|
//void init_idct_int32 (void); |
|
|
//void idct_int32 (short *block); |
|
|
|
|
64 |
/* private data */ |
/* private data */ |
65 |
static short iclip[1024]; /* clipping table */ |
static short iclip[1024]; /* clipping table */ |
66 |
static short *iclp; |
static short *iclp; |
67 |
|
|
68 |
/* private prototypes */ |
/* private prototypes */ |
|
//static void idctrow _ANSI_ARGS_((short *blk)); |
|
|
//static void idctcol _ANSI_ARGS_((short *blk)); |
|
69 |
|
|
70 |
/* row (horizontal) IDCT |
/* row (horizontal) IDCT |
71 |
* |
* |
77 |
* c[1..7] = 128*sqrt(2) |
* c[1..7] = 128*sqrt(2) |
78 |
*/ |
*/ |
79 |
|
|
80 |
/* |
#if 0 |
81 |
static void idctrow(blk) |
static void idctrow(blk) |
82 |
short *blk; |
short *blk; |
83 |
{ |
{ |
84 |
int X0, X1, X2, X3, X4, X5, X6, X7, X8; |
int X0, X1, X2, X3, X4, X5, X6, X7, X8; |
85 |
|
|
86 |
// shortcut |
/* shortcut */ |
87 |
if (!((X1 = blk[4]<<11) | (X2 = blk[6]) | (X3 = blk[2]) | |
if (!((X1 = blk[4]<<11) | (X2 = blk[6]) | (X3 = blk[2]) | |
88 |
(X4 = blk[1]) | (X5 = blk[7]) | (X6 = blk[5]) | (X7 = blk[3]))) |
(X4 = blk[1]) | (X5 = blk[7]) | (X6 = blk[5]) | (X7 = blk[3]))) |
89 |
{ |
{ |
91 |
return; |
return; |
92 |
} |
} |
93 |
|
|
94 |
X0 = (blk[0]<<11) + 128; // for proper rounding in the fourth stage |
X0 = (blk[0]<<11) + 128; /* for proper rounding in the fourth stage */ |
95 |
|
|
96 |
// first stage |
/* first stage */ |
97 |
X8 = W7*(X4+X5); |
X8 = W7*(X4+X5); |
98 |
X4 = X8 + (W1-W7)*X4; |
X4 = X8 + (W1-W7)*X4; |
99 |
X5 = X8 - (W1+W7)*X5; |
X5 = X8 - (W1+W7)*X5; |
101 |
X6 = X8 - (W3-W5)*X6; |
X6 = X8 - (W3-W5)*X6; |
102 |
X7 = X8 - (W3+W5)*X7; |
X7 = X8 - (W3+W5)*X7; |
103 |
|
|
104 |
// second stage |
/* second stage */ |
105 |
X8 = X0 + X1; |
X8 = X0 + X1; |
106 |
X0 -= X1; |
X0 -= X1; |
107 |
X1 = W6*(X3+X2); |
X1 = W6*(X3+X2); |
112 |
X6 = X5 + X7; |
X6 = X5 + X7; |
113 |
X5 -= X7; |
X5 -= X7; |
114 |
|
|
115 |
// third stage |
/* third stage */ |
116 |
X7 = X8 + X3; |
X7 = X8 + X3; |
117 |
X8 -= X3; |
X8 -= X3; |
118 |
X3 = X0 + X2; |
X3 = X0 + X2; |
120 |
X2 = (181*(X4+X5)+128)>>8; |
X2 = (181*(X4+X5)+128)>>8; |
121 |
X4 = (181*(X4-X5)+128)>>8; |
X4 = (181*(X4-X5)+128)>>8; |
122 |
|
|
123 |
// fourth stage |
/* fourth stage */ |
124 |
blk[0] = (X7+X1)>>8; |
blk[0] = (X7+X1)>>8; |
125 |
blk[1] = (X3+X2)>>8; |
blk[1] = (X3+X2)>>8; |
126 |
blk[2] = (X0+X4)>>8; |
blk[2] = (X0+X4)>>8; |
129 |
blk[5] = (X0-X4)>>8; |
blk[5] = (X0-X4)>>8; |
130 |
blk[6] = (X3-X2)>>8; |
blk[6] = (X3-X2)>>8; |
131 |
blk[7] = (X7-X1)>>8; |
blk[7] = (X7-X1)>>8; |
132 |
}*/ |
} |
133 |
|
#endif |
134 |
|
|
135 |
/* column (vertical) IDCT |
/* column (vertical) IDCT |
136 |
* |
* |
141 |
* where: c[0] = 1/1024 |
* where: c[0] = 1/1024 |
142 |
* c[1..7] = (1/1024)*sqrt(2) |
* c[1..7] = (1/1024)*sqrt(2) |
143 |
*/ |
*/ |
144 |
/* |
|
145 |
|
#if 0 |
146 |
static void idctcol(blk) |
static void idctcol(blk) |
147 |
short *blk; |
short *blk; |
148 |
{ |
{ |
149 |
int X0, X1, X2, X3, X4, X5, X6, X7, X8; |
int X0, X1, X2, X3, X4, X5, X6, X7, X8; |
150 |
|
|
151 |
// shortcut |
/* shortcut */ |
152 |
if (!((X1 = (blk[8*4]<<8)) | (X2 = blk[8*6]) | (X3 = blk[8*2]) | |
if (!((X1 = (blk[8*4]<<8)) | (X2 = blk[8*6]) | (X3 = blk[8*2]) | |
153 |
(X4 = blk[8*1]) | (X5 = blk[8*7]) | (X6 = blk[8*5]) | (X7 = blk[8*3]))) |
(X4 = blk[8*1]) | (X5 = blk[8*7]) | (X6 = blk[8*5]) | (X7 = blk[8*3]))) |
154 |
{ |
{ |
159 |
|
|
160 |
X0 = (blk[8*0]<<8) + 8192; |
X0 = (blk[8*0]<<8) + 8192; |
161 |
|
|
162 |
// first stage |
/* first stage */ |
163 |
X8 = W7*(X4+X5) + 4; |
X8 = W7*(X4+X5) + 4; |
164 |
X4 = (X8+(W1-W7)*X4)>>3; |
X4 = (X8+(W1-W7)*X4)>>3; |
165 |
X5 = (X8-(W1+W7)*X5)>>3; |
X5 = (X8-(W1+W7)*X5)>>3; |
167 |
X6 = (X8-(W3-W5)*X6)>>3; |
X6 = (X8-(W3-W5)*X6)>>3; |
168 |
X7 = (X8-(W3+W5)*X7)>>3; |
X7 = (X8-(W3+W5)*X7)>>3; |
169 |
|
|
170 |
// second stage |
/* second stage */ |
171 |
X8 = X0 + X1; |
X8 = X0 + X1; |
172 |
X0 -= X1; |
X0 -= X1; |
173 |
X1 = W6*(X3+X2) + 4; |
X1 = W6*(X3+X2) + 4; |
178 |
X6 = X5 + X7; |
X6 = X5 + X7; |
179 |
X5 -= X7; |
X5 -= X7; |
180 |
|
|
181 |
// third stage |
/* third stage */ |
182 |
X7 = X8 + X3; |
X7 = X8 + X3; |
183 |
X8 -= X3; |
X8 -= X3; |
184 |
X3 = X0 + X2; |
X3 = X0 + X2; |
186 |
X2 = (181*(X4+X5)+128)>>8; |
X2 = (181*(X4+X5)+128)>>8; |
187 |
X4 = (181*(X4-X5)+128)>>8; |
X4 = (181*(X4-X5)+128)>>8; |
188 |
|
|
189 |
// fourth stage |
/* fourth stage */ |
190 |
blk[8*0] = iclp[(X7+X1)>>14]; |
blk[8*0] = iclp[(X7+X1)>>14]; |
191 |
blk[8*1] = iclp[(X3+X2)>>14]; |
blk[8*1] = iclp[(X3+X2)>>14]; |
192 |
blk[8*2] = iclp[(X0+X4)>>14]; |
blk[8*2] = iclp[(X0+X4)>>14]; |
195 |
blk[8*5] = iclp[(X0-X4)>>14]; |
blk[8*5] = iclp[(X0-X4)>>14]; |
196 |
blk[8*6] = iclp[(X3-X2)>>14]; |
blk[8*6] = iclp[(X3-X2)>>14]; |
197 |
blk[8*7] = iclp[(X7-X1)>>14]; |
blk[8*7] = iclp[(X7-X1)>>14]; |
198 |
}*/ |
} |
199 |
|
#endif |
200 |
|
|
201 |
// function pointer |
/* function pointer */ |
202 |
idctFuncPtr idct; |
idctFuncPtr idct; |
203 |
|
|
204 |
/* two dimensional inverse discrete cosine transform */ |
/* two dimensional inverse discrete cosine transform */ |
|
//void j_rev_dct(block) |
|
|
//short *block; |
|
205 |
void |
void |
206 |
idct_int32(short *const block) |
idct_int32(short *const block) |
207 |
{ |
{ |
208 |
|
|
209 |
// idct_int32_init() must be called before the first call to this function! |
/* |
210 |
|
* idct_int32_init() must be called before the first call to this |
211 |
|
* function! |
212 |
|
*/ |
213 |
|
|
214 |
|
|
215 |
/*int i; |
#if 0 |
216 |
|
int i; |
217 |
long i; |
long i; |
218 |
|
|
219 |
for (i=0; i<8; i++) |
for (i=0; i<8; i++) |
220 |
idctrow(block+8*i); |
idctrow(block+8*i); |
221 |
|
|
222 |
for (i=0; i<8; i++) |
for (i=0; i<8; i++) |
223 |
idctcol(block+i); */ |
idctcol(block+i); |
224 |
|
#endif |
225 |
|
|
226 |
static short *blk; |
static short *blk; |
227 |
static long i; |
static long i; |
228 |
static long X0, X1, X2, X3, X4, X5, X6, X7, X8; |
static long X0, X1, X2, X3, X4, X5, X6, X7, X8; |
229 |
|
|
230 |
|
|
231 |
for (i = 0; i < 8; i++) // idct rows |
for (i = 0; i < 8; i++) /* idct rows */ |
232 |
{ |
{ |
233 |
blk = block + (i << 3); |
blk = block + (i << 3); |
234 |
if (! |
if (! |
240 |
continue; |
continue; |
241 |
} |
} |
242 |
|
|
243 |
X0 = (blk[0] << 11) + 128; // for proper rounding in the fourth stage |
X0 = (blk[0] << 11) + 128; /* for proper rounding in the fourth stage */ |
244 |
|
|
245 |
// first stage |
/* first stage */ |
246 |
X8 = W7 * (X4 + X5); |
X8 = W7 * (X4 + X5); |
247 |
X4 = X8 + (W1 - W7) * X4; |
X4 = X8 + (W1 - W7) * X4; |
248 |
X5 = X8 - (W1 + W7) * X5; |
X5 = X8 - (W1 + W7) * X5; |
250 |
X6 = X8 - (W3 - W5) * X6; |
X6 = X8 - (W3 - W5) * X6; |
251 |
X7 = X8 - (W3 + W5) * X7; |
X7 = X8 - (W3 + W5) * X7; |
252 |
|
|
253 |
// second stage |
/* second stage */ |
254 |
X8 = X0 + X1; |
X8 = X0 + X1; |
255 |
X0 -= X1; |
X0 -= X1; |
256 |
X1 = W6 * (X3 + X2); |
X1 = W6 * (X3 + X2); |
261 |
X6 = X5 + X7; |
X6 = X5 + X7; |
262 |
X5 -= X7; |
X5 -= X7; |
263 |
|
|
264 |
// third stage |
/* third stage */ |
265 |
X7 = X8 + X3; |
X7 = X8 + X3; |
266 |
X8 -= X3; |
X8 -= X3; |
267 |
X3 = X0 + X2; |
X3 = X0 + X2; |
269 |
X2 = (181 * (X4 + X5) + 128) >> 8; |
X2 = (181 * (X4 + X5) + 128) >> 8; |
270 |
X4 = (181 * (X4 - X5) + 128) >> 8; |
X4 = (181 * (X4 - X5) + 128) >> 8; |
271 |
|
|
272 |
// fourth stage |
/* fourth stage */ |
273 |
|
|
274 |
blk[0] = (short) ((X7 + X1) >> 8); |
blk[0] = (short) ((X7 + X1) >> 8); |
275 |
blk[1] = (short) ((X3 + X2) >> 8); |
blk[1] = (short) ((X3 + X2) >> 8); |
280 |
blk[6] = (short) ((X3 - X2) >> 8); |
blk[6] = (short) ((X3 - X2) >> 8); |
281 |
blk[7] = (short) ((X7 - X1) >> 8); |
blk[7] = (short) ((X7 - X1) >> 8); |
282 |
|
|
283 |
} // end for ( i = 0; i < 8; ++i ) IDCT-rows |
} /* end for ( i = 0; i < 8; ++i ) IDCT-rows */ |
284 |
|
|
285 |
|
|
286 |
|
|
287 |
for (i = 0; i < 8; i++) // idct columns |
for (i = 0; i < 8; i++) /* idct columns */ |
288 |
{ |
{ |
289 |
blk = block + i; |
blk = block + i; |
290 |
// shortcut |
/* shortcut */ |
291 |
if (! |
if (! |
292 |
((X1 = (blk[8 * 4] << 8)) | (X2 = blk[8 * 6]) | (X3 = |
((X1 = (blk[8 * 4] << 8)) | (X2 = blk[8 * 6]) | (X3 = |
293 |
blk[8 * |
blk[8 * |
303 |
|
|
304 |
X0 = (blk[8 * 0] << 8) + 8192; |
X0 = (blk[8 * 0] << 8) + 8192; |
305 |
|
|
306 |
// first stage |
/* first stage */ |
307 |
X8 = W7 * (X4 + X5) + 4; |
X8 = W7 * (X4 + X5) + 4; |
308 |
X4 = (X8 + (W1 - W7) * X4) >> 3; |
X4 = (X8 + (W1 - W7) * X4) >> 3; |
309 |
X5 = (X8 - (W1 + W7) * X5) >> 3; |
X5 = (X8 - (W1 + W7) * X5) >> 3; |
311 |
X6 = (X8 - (W3 - W5) * X6) >> 3; |
X6 = (X8 - (W3 - W5) * X6) >> 3; |
312 |
X7 = (X8 - (W3 + W5) * X7) >> 3; |
X7 = (X8 - (W3 + W5) * X7) >> 3; |
313 |
|
|
314 |
// second stage |
/* second stage */ |
315 |
X8 = X0 + X1; |
X8 = X0 + X1; |
316 |
X0 -= X1; |
X0 -= X1; |
317 |
X1 = W6 * (X3 + X2) + 4; |
X1 = W6 * (X3 + X2) + 4; |
322 |
X6 = X5 + X7; |
X6 = X5 + X7; |
323 |
X5 -= X7; |
X5 -= X7; |
324 |
|
|
325 |
// third stage |
/* third stage */ |
326 |
X7 = X8 + X3; |
X7 = X8 + X3; |
327 |
X8 -= X3; |
X8 -= X3; |
328 |
X3 = X0 + X2; |
X3 = X0 + X2; |
330 |
X2 = (181 * (X4 + X5) + 128) >> 8; |
X2 = (181 * (X4 + X5) + 128) >> 8; |
331 |
X4 = (181 * (X4 - X5) + 128) >> 8; |
X4 = (181 * (X4 - X5) + 128) >> 8; |
332 |
|
|
333 |
// fourth stage |
/* fourth stage */ |
334 |
blk[8 * 0] = iclp[(X7 + X1) >> 14]; |
blk[8 * 0] = iclp[(X7 + X1) >> 14]; |
335 |
blk[8 * 1] = iclp[(X3 + X2) >> 14]; |
blk[8 * 1] = iclp[(X3 + X2) >> 14]; |
336 |
blk[8 * 2] = iclp[(X0 + X4) >> 14]; |
blk[8 * 2] = iclp[(X0 + X4) >> 14]; |
341 |
blk[8 * 7] = iclp[(X7 - X1) >> 14]; |
blk[8 * 7] = iclp[(X7 - X1) >> 14]; |
342 |
} |
} |
343 |
|
|
344 |
} // end function idct_int32(block) |
} /* end function idct_int32(block) */ |
345 |
|
|
346 |
|
|
|
//void |
|
|
//idct_int32_init() |
|
347 |
void |
void |
348 |
idct_int32_init() |
idct_int32_init() |
349 |
{ |
{ |