|
2
|
1 |
/*
|
|
5506
|
2 |
* Copyright (c) 2000, 2003, Oracle and/or its affiliates. All rights reserved.
|
|
2
|
3 |
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
|
|
|
4 |
*
|
|
|
5 |
* This code is free software; you can redistribute it and/or modify it
|
|
|
6 |
* under the terms of the GNU General Public License version 2 only, as
|
|
5506
|
7 |
* published by the Free Software Foundation. Oracle designates this
|
|
2
|
8 |
* particular file as subject to the "Classpath" exception as provided
|
|
5506
|
9 |
* by Oracle in the LICENSE file that accompanied this code.
|
|
2
|
10 |
*
|
|
|
11 |
* This code is distributed in the hope that it will be useful, but WITHOUT
|
|
|
12 |
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
|
|
|
13 |
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
|
|
|
14 |
* version 2 for more details (a copy is included in the LICENSE file that
|
|
|
15 |
* accompanied this code).
|
|
|
16 |
*
|
|
|
17 |
* You should have received a copy of the GNU General Public License version
|
|
|
18 |
* 2 along with this work; if not, write to the Free Software Foundation,
|
|
|
19 |
* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
|
|
|
20 |
*
|
|
5506
|
21 |
* Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
|
|
|
22 |
* or visit www.oracle.com if you need additional information or have any
|
|
|
23 |
* questions.
|
|
2
|
24 |
*/
|
|
|
25 |
|
|
|
26 |
|
|
|
27 |
|
|
|
28 |
#include <stdlib.h>
|
|
|
29 |
#include "mlib_image.h"
|
|
|
30 |
#include "mlib_ImageCheck.h"
|
|
|
31 |
|
|
|
32 |
typedef union {
|
|
|
33 |
double d64;
|
|
|
34 |
struct {
|
|
|
35 |
float f0;
|
|
|
36 |
float f1;
|
|
|
37 |
} f32s;
|
|
|
38 |
} d64_2_f32;
|
|
|
39 |
|
|
|
40 |
/***************************************************************/
|
|
|
41 |
|
|
|
42 |
void mlib_v_ImageChannelExtract_U8_2_1(mlib_u8 *sl, mlib_s32 slb,
|
|
|
43 |
mlib_u8 *dl, mlib_s32 dlb,
|
|
|
44 |
mlib_s32 width, mlib_s32 height)
|
|
|
45 |
{
|
|
|
46 |
mlib_u8 *sp = sl;
|
|
|
47 |
mlib_u8 *dp = dl;
|
|
|
48 |
int i, j;
|
|
|
49 |
|
|
|
50 |
for (j = 0; j < height; j++) {
|
|
|
51 |
mlib_u8 *dend = dl + width;
|
|
|
52 |
mlib_u32 *sp2;
|
|
|
53 |
while (((mlib_addr)sp & 7) > 1) {
|
|
|
54 |
*dp++ = *sp;
|
|
|
55 |
sp += 2;
|
|
|
56 |
if (dp >= dend) break;
|
|
|
57 |
}
|
|
|
58 |
if ((mlib_addr)sp & 7) {
|
|
|
59 |
sp2 = (mlib_u32 *)(sp - 1);
|
|
|
60 |
#pragma pipeloop(0)
|
|
|
61 |
for (; dp <= (dend-2); dp += 2) {
|
|
|
62 |
mlib_u32 s0;
|
|
|
63 |
s0 = *sp2++;
|
|
|
64 |
dp[0] = s0 >> 16;
|
|
|
65 |
dp[1] = s0;
|
|
|
66 |
}
|
|
|
67 |
if (dp < dend) {
|
|
|
68 |
dp[0] = sp2[0] >> 16;
|
|
|
69 |
}
|
|
|
70 |
} else {
|
|
|
71 |
sp2 = (mlib_u32 *)sp;
|
|
|
72 |
#pragma pipeloop(0)
|
|
|
73 |
for (; dp <= (dend-2); dp += 2) {
|
|
|
74 |
mlib_u32 s0;
|
|
|
75 |
s0 = *sp2++;
|
|
|
76 |
dp[0] = s0 >> 24;
|
|
|
77 |
dp[1] = s0 >> 8;
|
|
|
78 |
}
|
|
|
79 |
if (dp < dend) {
|
|
|
80 |
dp[0] = sp2[0] >> 24;
|
|
|
81 |
}
|
|
|
82 |
}
|
|
|
83 |
sp = sl += slb;
|
|
|
84 |
dp = dl += dlb;
|
|
|
85 |
}
|
|
|
86 |
}
|
|
|
87 |
|
|
|
88 |
/***************************************************************/
|
|
|
89 |
|
|
|
90 |
void mlib_v_ImageChannelExtract_U8_3_2(mlib_u8 *sl, mlib_s32 slb,
|
|
|
91 |
mlib_u8 *dl, mlib_s32 dlb,
|
|
|
92 |
mlib_s32 width, mlib_s32 height,
|
|
|
93 |
mlib_s32 count1)
|
|
|
94 |
{
|
|
|
95 |
mlib_u8 *sp = sl;
|
|
|
96 |
mlib_u8 *dp = dl;
|
|
|
97 |
mlib_u32 *sp2;
|
|
|
98 |
mlib_u16 *dp2;
|
|
|
99 |
mlib_u16 *d2end;
|
|
|
100 |
mlib_u32 s0, s1, s2, s3;
|
|
|
101 |
int i, j, off, count_off;
|
|
|
102 |
|
|
|
103 |
for (j = 0; j < height; j++) {
|
|
|
104 |
mlib_u8 *dend = dl + 2*width;
|
|
|
105 |
|
|
|
106 |
if (count1 == 1) {
|
|
|
107 |
if (dp < dend) *dp++ = sp[0];
|
|
|
108 |
sp += 2;
|
|
|
109 |
}
|
|
|
110 |
|
|
|
111 |
if ((mlib_addr)dp & 1) {
|
|
|
112 |
#pragma pipeloop(0)
|
|
|
113 |
for (; dp <= (dend-2); dp += 2) {
|
|
|
114 |
dp[0] = sp[0];
|
|
|
115 |
dp[1] = sp[1];
|
|
|
116 |
sp += 3;
|
|
|
117 |
}
|
|
|
118 |
if (dp < dend) {
|
|
|
119 |
dp[0] = sp[0];
|
|
|
120 |
}
|
|
|
121 |
sp = sl += slb;
|
|
|
122 |
dp = dl += dlb;
|
|
|
123 |
continue;
|
|
|
124 |
}
|
|
|
125 |
|
|
|
126 |
dp2 = (mlib_u16*)dp;
|
|
|
127 |
d2end = (mlib_u16*)((mlib_addr)dend &~ 1);
|
|
|
128 |
off = (mlib_addr)sp & 3;
|
|
|
129 |
sp2 = (mlib_u32 *)(sp - off);
|
|
|
130 |
|
|
|
131 |
switch (off) {
|
|
|
132 |
|
|
|
133 |
case 0:
|
|
|
134 |
#pragma pipeloop(0)
|
|
|
135 |
for (; dp2 <= (d2end-4); dp2 += 4) {
|
|
|
136 |
s0 = sp2[0];
|
|
|
137 |
s1 = sp2[1];
|
|
|
138 |
s2 = sp2[2];
|
|
|
139 |
dp2[0] = s0 >> 16;
|
|
|
140 |
dp2[1] = (s0 << 8) | (s1 >> 24);
|
|
|
141 |
dp2[2] = s1;
|
|
|
142 |
dp2[3] = s2 >> 8;
|
|
|
143 |
sp2 += 3;
|
|
|
144 |
}
|
|
|
145 |
break;
|
|
|
146 |
|
|
|
147 |
case 1:
|
|
|
148 |
#pragma pipeloop(0)
|
|
|
149 |
for (; dp2 <= (d2end-4); dp2 += 4) {
|
|
|
150 |
s0 = sp2[0];
|
|
|
151 |
s1 = sp2[1];
|
|
|
152 |
s2 = sp2[2];
|
|
|
153 |
dp2[0] = s0 >> 8;
|
|
|
154 |
dp2[1] = s1 >> 16;
|
|
|
155 |
dp2[2] = (s1 << 8) | (s2 >> 24);
|
|
|
156 |
dp2[3] = s2;
|
|
|
157 |
sp2 += 3;
|
|
|
158 |
}
|
|
|
159 |
break;
|
|
|
160 |
|
|
|
161 |
case 2:
|
|
|
162 |
#pragma pipeloop(0)
|
|
|
163 |
s3 = sp2[0];
|
|
|
164 |
for (; dp2 <= (d2end-4); dp2 += 4) {
|
|
|
165 |
s0 = s3;
|
|
|
166 |
s1 = sp2[1];
|
|
|
167 |
s2 = sp2[2];
|
|
|
168 |
s3 = sp2[3];
|
|
|
169 |
dp2[0] = s0;
|
|
|
170 |
dp2[1] = s1 >> 8;
|
|
|
171 |
dp2[2] = s2 >> 16;
|
|
|
172 |
dp2[3] = (s2 << 8) | (s3 >> 24);
|
|
|
173 |
sp2 += 3;
|
|
|
174 |
}
|
|
|
175 |
break;
|
|
|
176 |
|
|
|
177 |
case 3:
|
|
|
178 |
#pragma pipeloop(0)
|
|
|
179 |
s3 = sp2[0];
|
|
|
180 |
for (; dp2 <= (d2end-4); dp2 += 4) {
|
|
|
181 |
s0 = s3;
|
|
|
182 |
s1 = sp2[1];
|
|
|
183 |
s2 = sp2[2];
|
|
|
184 |
s3 = sp2[3];
|
|
|
185 |
dp2[0] = (s0 << 8) | (s1 >> 24);
|
|
|
186 |
dp2[1] = s1;
|
|
|
187 |
dp2[2] = s2 >> 8;
|
|
|
188 |
dp2[3] = s3 >> 16;
|
|
|
189 |
sp2 += 3;
|
|
|
190 |
}
|
|
|
191 |
}
|
|
|
192 |
|
|
|
193 |
sp = (mlib_u8 *)sp2 + off;
|
|
|
194 |
dp = (mlib_u8 *)dp2;
|
|
|
195 |
while (dp < dend) {
|
|
|
196 |
*dp++ = sp[0];
|
|
|
197 |
if (dp < dend) *dp++ = sp[1];
|
|
|
198 |
sp += 3;
|
|
|
199 |
}
|
|
|
200 |
|
|
|
201 |
sp = sl += slb;
|
|
|
202 |
dp = dl += dlb;
|
|
|
203 |
}
|
|
|
204 |
}
|
|
|
205 |
|
|
|
206 |
/***************************************************************/
|
|
|
207 |
|
|
|
208 |
void mlib_v_ImageChannelExtract_U8_4_2(mlib_u8 *sl, mlib_s32 slb,
|
|
|
209 |
mlib_u8 *dl, mlib_s32 dlb,
|
|
|
210 |
mlib_s32 width, mlib_s32 height,
|
|
|
211 |
mlib_s32 count1)
|
|
|
212 |
{
|
|
|
213 |
mlib_u8 *sp = sl;
|
|
|
214 |
mlib_u8 *dp = dl;
|
|
|
215 |
mlib_u32 *sp2;
|
|
|
216 |
mlib_u16 *dp2;
|
|
|
217 |
mlib_u16 *d2end;
|
|
|
218 |
mlib_u32 s0, s1, s2, s3;
|
|
|
219 |
int i, j, off, count_off;
|
|
|
220 |
|
|
|
221 |
for (j = 0; j < height; j++) {
|
|
|
222 |
mlib_u8 *dend = dl + 2*width;
|
|
|
223 |
|
|
|
224 |
if (count1 == 1) {
|
|
|
225 |
if (dp < dend) *dp++ = sp[0];
|
|
|
226 |
sp += 3;
|
|
|
227 |
}
|
|
|
228 |
|
|
|
229 |
off = (mlib_addr)sp & 3;
|
|
|
230 |
|
|
|
231 |
if (((mlib_addr)dp & 1) || (off == 3)) {
|
|
|
232 |
#pragma pipeloop(0)
|
|
|
233 |
for (; dp <= (dend-2); dp += 2) {
|
|
|
234 |
dp[0] = sp[0];
|
|
|
235 |
dp[1] = sp[1];
|
|
|
236 |
sp += 4;
|
|
|
237 |
}
|
|
|
238 |
if (dp < dend) {
|
|
|
239 |
dp[0] = sp[0];
|
|
|
240 |
}
|
|
|
241 |
sp = sl += slb;
|
|
|
242 |
dp = dl += dlb;
|
|
|
243 |
continue;
|
|
|
244 |
}
|
|
|
245 |
|
|
|
246 |
dp2 = (mlib_u16*)dp;
|
|
|
247 |
d2end = (mlib_u16*)((mlib_addr)dend &~ 1);
|
|
|
248 |
sp2 = (mlib_u32 *)(sp - off);
|
|
|
249 |
|
|
|
250 |
switch (off) {
|
|
|
251 |
|
|
|
252 |
case 0:
|
|
|
253 |
#pragma pipeloop(0)
|
|
|
254 |
for (; dp2 < d2end; dp2++) {
|
|
|
255 |
s0 = sp2[0];
|
|
|
256 |
dp2[0] = s0 >> 16;
|
|
|
257 |
sp2++;
|
|
|
258 |
}
|
|
|
259 |
break;
|
|
|
260 |
|
|
|
261 |
case 1:
|
|
|
262 |
#pragma pipeloop(0)
|
|
|
263 |
for (; dp2 < d2end; dp2++) {
|
|
|
264 |
s0 = sp2[0];
|
|
|
265 |
dp2[0] = s0 >> 8;
|
|
|
266 |
sp2++;
|
|
|
267 |
}
|
|
|
268 |
break;
|
|
|
269 |
|
|
|
270 |
case 2:
|
|
|
271 |
#pragma pipeloop(0)
|
|
|
272 |
for (; dp2 < d2end; dp2++) {
|
|
|
273 |
s0 = sp2[0];
|
|
|
274 |
dp2[0] = s0;
|
|
|
275 |
sp2++;
|
|
|
276 |
}
|
|
|
277 |
break;
|
|
|
278 |
}
|
|
|
279 |
|
|
|
280 |
sp = (mlib_u8 *)sp2 + off;
|
|
|
281 |
dp = (mlib_u8 *)dp2;
|
|
|
282 |
if (dp < dend) {
|
|
|
283 |
*dp++ = sp[0];
|
|
|
284 |
}
|
|
|
285 |
|
|
|
286 |
sp = sl += slb;
|
|
|
287 |
dp = dl += dlb;
|
|
|
288 |
}
|
|
|
289 |
}
|
|
|
290 |
|
|
|
291 |
/***************************************************************/
|
|
|
292 |
|
|
|
293 |
void mlib_v_ImageChannelExtract_32_2_1(mlib_f32 *sp, mlib_s32 slb,
|
|
|
294 |
mlib_f32 *dp, mlib_s32 dlb,
|
|
|
295 |
mlib_s32 width, mlib_s32 height)
|
|
|
296 |
{
|
|
|
297 |
mlib_d64 *sp2;
|
|
|
298 |
int i, j, off;
|
|
|
299 |
|
|
|
300 |
for (j = 0; j < height; j++) {
|
|
|
301 |
|
|
|
302 |
if (((mlib_addr)sp & 7) == 0) {
|
|
|
303 |
sp2 = (mlib_d64 *)sp;
|
|
|
304 |
#pragma pipeloop(0)
|
|
|
305 |
for (i = 0; i < width; i++) {
|
|
|
306 |
d64_2_f32 d;
|
|
|
307 |
d.d64 = sp2[i];
|
|
|
308 |
dp[i] = d.f32s.f0;
|
|
|
309 |
}
|
|
|
310 |
} else {
|
|
|
311 |
sp2 = (mlib_d64 *)(sp - 1);
|
|
|
312 |
#pragma pipeloop(0)
|
|
|
313 |
for (i = 0; i < width; i++) {
|
|
|
314 |
d64_2_f32 d;
|
|
|
315 |
d.d64 = sp2[i];
|
|
|
316 |
dp[i] = d.f32s.f1;
|
|
|
317 |
}
|
|
|
318 |
}
|
|
|
319 |
|
|
|
320 |
sp += slb;
|
|
|
321 |
dp += dlb;
|
|
|
322 |
}
|
|
|
323 |
}
|
|
|
324 |
|
|
|
325 |
/***************************************************************/
|
|
|
326 |
|
|
|
327 |
void mlib_v_ImageChannelExtract_32_3_1(mlib_f32 *sl, mlib_s32 slb,
|
|
|
328 |
mlib_f32 *dl, mlib_s32 dlb,
|
|
|
329 |
mlib_s32 width, mlib_s32 height)
|
|
|
330 |
{
|
|
|
331 |
mlib_f32 *sp = sl;
|
|
|
332 |
mlib_f32 *dp = dl;
|
|
|
333 |
mlib_d64 *sp2;
|
|
|
334 |
d64_2_f32 d0;
|
|
|
335 |
int i, j, off;
|
|
|
336 |
|
|
|
337 |
for (j = 0; j < height; j++) {
|
|
|
338 |
mlib_f32 *dend = dl + width;
|
|
|
339 |
|
|
|
340 |
if ((mlib_addr)sp & 7) {
|
|
|
341 |
dp[0] = sp[0];
|
|
|
342 |
sp += 3;
|
|
|
343 |
dp ++;
|
|
|
344 |
}
|
|
|
345 |
|
|
|
346 |
sp2 = (mlib_d64 *)sp;
|
|
|
347 |
#pragma pipeloop(0)
|
|
|
348 |
for (; dp <= (dend-2); dp += 2) {
|
|
|
349 |
d64_2_f32 d0, d1;
|
|
|
350 |
d0.d64 = sp2[0];
|
|
|
351 |
d1.d64 = sp2[1];
|
|
|
352 |
dp[0] = d0.f32s.f0;
|
|
|
353 |
dp[1] = d1.f32s.f1;
|
|
|
354 |
sp2 += 3;
|
|
|
355 |
}
|
|
|
356 |
|
|
|
357 |
if (dp < dend) {
|
|
|
358 |
d0.d64 = sp2[0];
|
|
|
359 |
dp[0] = d0.f32s.f0;
|
|
|
360 |
}
|
|
|
361 |
|
|
|
362 |
sp = sl += slb;
|
|
|
363 |
dp = dl += dlb;
|
|
|
364 |
}
|
|
|
365 |
}
|
|
|
366 |
|
|
|
367 |
/***************************************************************/
|
|
|
368 |
|
|
|
369 |
void mlib_v_ImageChannelExtract_32_3_2(mlib_f32 *sl, mlib_s32 slb,
|
|
|
370 |
mlib_f32 *dl, mlib_s32 dlb,
|
|
|
371 |
mlib_s32 width, mlib_s32 height,
|
|
|
372 |
mlib_s32 count1)
|
|
|
373 |
{
|
|
|
374 |
mlib_f32 *sp = sl;
|
|
|
375 |
mlib_f32 *dp = dl;
|
|
|
376 |
mlib_d64 *sp2;
|
|
|
377 |
d64_2_f32 d0;
|
|
|
378 |
int i, j, off;
|
|
|
379 |
|
|
|
380 |
for (j = 0; j < height; j++) {
|
|
|
381 |
mlib_f32 *dend = dl + 2*width;
|
|
|
382 |
|
|
|
383 |
if (count1 == 1) {
|
|
|
384 |
if (dp < dend) *dp++ = sp[0];
|
|
|
385 |
sp += 2;
|
|
|
386 |
}
|
|
|
387 |
|
|
|
388 |
if ((mlib_addr)sp & 7) {
|
|
|
389 |
if (dp < dend) *dp++ = sp[0];
|
|
|
390 |
if (dp < dend) *dp++ = sp[1];
|
|
|
391 |
sp += 3;
|
|
|
392 |
}
|
|
|
393 |
|
|
|
394 |
sp2 = (mlib_d64 *)sp;
|
|
|
395 |
#pragma pipeloop(0)
|
|
|
396 |
for (; dp <= (dend-4); dp += 4) {
|
|
|
397 |
d64_2_f32 d0, d1, d2;
|
|
|
398 |
d0.d64 = sp2[0];
|
|
|
399 |
d1.d64 = sp2[1];
|
|
|
400 |
d2.d64 = sp2[2];
|
|
|
401 |
dp[0] = d0.f32s.f0;
|
|
|
402 |
dp[1] = d0.f32s.f1;
|
|
|
403 |
dp[2] = d1.f32s.f1;
|
|
|
404 |
dp[3] = d2.f32s.f0;
|
|
|
405 |
sp2 += 3;
|
|
|
406 |
}
|
|
|
407 |
|
|
|
408 |
if (dp < dend) {
|
|
|
409 |
sp = (mlib_f32 *)sp2;
|
|
|
410 |
*dp++ = sp[0];
|
|
|
411 |
if (dp < dend) *dp++ = sp[1];
|
|
|
412 |
if (dp < dend) *dp++ = sp[3];
|
|
|
413 |
}
|
|
|
414 |
|
|
|
415 |
sp = sl += slb;
|
|
|
416 |
dp = dl += dlb;
|
|
|
417 |
}
|
|
|
418 |
}
|
|
|
419 |
|
|
|
420 |
/***************************************************************/
|
|
|
421 |
|
|
|
422 |
void mlib_v_ImageChannelExtract_32_4_1(mlib_f32 *sp, mlib_s32 slb,
|
|
|
423 |
mlib_f32 *dp, mlib_s32 dlb,
|
|
|
424 |
mlib_s32 width, mlib_s32 height)
|
|
|
425 |
{
|
|
|
426 |
mlib_d64 *sp2;
|
|
|
427 |
int i, j, off;
|
|
|
428 |
|
|
|
429 |
for (j = 0; j < height; j++) {
|
|
|
430 |
|
|
|
431 |
if (((mlib_addr)sp & 7) == 0) {
|
|
|
432 |
sp2 = (mlib_d64 *)sp;
|
|
|
433 |
#pragma pipeloop(0)
|
|
|
434 |
for (i = 0; i < width; i++) {
|
|
|
435 |
d64_2_f32 d;
|
|
|
436 |
d.d64 = sp2[2*i];
|
|
|
437 |
dp[i] = d.f32s.f0;
|
|
|
438 |
}
|
|
|
439 |
} else {
|
|
|
440 |
sp2 = (mlib_d64 *)(sp - 1);
|
|
|
441 |
#pragma pipeloop(0)
|
|
|
442 |
for (i = 0; i < width; i++) {
|
|
|
443 |
d64_2_f32 d;
|
|
|
444 |
d.d64 = sp2[2*i];
|
|
|
445 |
dp[i] = d.f32s.f1;
|
|
|
446 |
}
|
|
|
447 |
}
|
|
|
448 |
|
|
|
449 |
sp += slb;
|
|
|
450 |
dp += dlb;
|
|
|
451 |
}
|
|
|
452 |
}
|
|
|
453 |
|
|
|
454 |
/***************************************************************/
|
|
|
455 |
|
|
|
456 |
void mlib_v_ImageChannelExtract_32_4_2(mlib_f32 *sl, mlib_s32 slb,
|
|
|
457 |
mlib_f32 *dl, mlib_s32 dlb,
|
|
|
458 |
mlib_s32 width, mlib_s32 height,
|
|
|
459 |
mlib_s32 count1)
|
|
|
460 |
{
|
|
|
461 |
mlib_f32 *sp = sl;
|
|
|
462 |
mlib_f32 *dp = dl;
|
|
|
463 |
mlib_d64 *sp2;
|
|
|
464 |
int i, j, off;
|
|
|
465 |
d64_2_f32 d0, d1;
|
|
|
466 |
|
|
|
467 |
for (j = 0; j < height; j++) {
|
|
|
468 |
mlib_f32 *dend = dl + 2*width;
|
|
|
469 |
|
|
|
470 |
if (count1 == 1) {
|
|
|
471 |
dp[0] = sp[0];
|
|
|
472 |
sp += 3;
|
|
|
473 |
dp ++;
|
|
|
474 |
}
|
|
|
475 |
|
|
|
476 |
if (((mlib_addr)sp & 7) == 0) {
|
|
|
477 |
sp2 = (mlib_d64 *)sp;
|
|
|
478 |
#pragma pipeloop(0)
|
|
|
479 |
for (; dp <= (dend-2); dp += 2) {
|
|
|
480 |
d64_2_f32 d;
|
|
|
481 |
d.d64 = sp2[0];
|
|
|
482 |
dp[0] = d.f32s.f0;
|
|
|
483 |
dp[1] = d.f32s.f1;
|
|
|
484 |
sp2 += 2;
|
|
|
485 |
}
|
|
|
486 |
if (dp < dend) {
|
|
|
487 |
d0.d64 = sp2[0];
|
|
|
488 |
dp[0] = d0.f32s.f0;
|
|
|
489 |
}
|
|
|
490 |
} else {
|
|
|
491 |
sp2 = (mlib_d64 *)(sp - 1);
|
|
|
492 |
#pragma pipeloop(0)
|
|
|
493 |
for (; dp <= (dend-2); dp += 2) {
|
|
|
494 |
d64_2_f32 d0, d1;
|
|
|
495 |
d0.d64 = sp2[0];
|
|
|
496 |
d1.d64 = sp2[1];
|
|
|
497 |
dp[0] = d0.f32s.f1;
|
|
|
498 |
dp[1] = d1.f32s.f0;
|
|
|
499 |
sp2 += 2;
|
|
|
500 |
}
|
|
|
501 |
if (dp < dend) {
|
|
|
502 |
d0.d64 = sp2[0];
|
|
|
503 |
dp[0] = d0.f32s.f1;
|
|
|
504 |
}
|
|
|
505 |
}
|
|
|
506 |
|
|
|
507 |
sp = sl += slb;
|
|
|
508 |
dp = dl += dlb;
|
|
|
509 |
}
|
|
|
510 |
}
|
|
|
511 |
|
|
|
512 |
/***************************************************************/
|
|
|
513 |
|
|
|
514 |
void mlib_v_ImageChannelExtract_32_4_3(mlib_f32 *sl, mlib_s32 slb,
|
|
|
515 |
mlib_f32 *dl, mlib_s32 dlb,
|
|
|
516 |
mlib_s32 width, mlib_s32 height,
|
|
|
517 |
mlib_s32 count1)
|
|
|
518 |
{
|
|
|
519 |
mlib_f32 *sp = sl;
|
|
|
520 |
mlib_f32 *dp = dl;
|
|
|
521 |
mlib_d64 *sp2;
|
|
|
522 |
int i, j, k;
|
|
|
523 |
d64_2_f32 d0, d1;
|
|
|
524 |
|
|
|
525 |
for (j = 0; j < height; j++) {
|
|
|
526 |
mlib_f32 *dend = dl + 3*width;
|
|
|
527 |
|
|
|
528 |
for (k = 0; k < count1; k++) {
|
|
|
529 |
if (dp < dend) *dp++ = *sp++;
|
|
|
530 |
}
|
|
|
531 |
sp++;
|
|
|
532 |
|
|
|
533 |
if (((mlib_addr)sp & 7) == 0) {
|
|
|
534 |
sp2 = (mlib_d64 *)sp;
|
|
|
535 |
#pragma pipeloop(0)
|
|
|
536 |
for (; dp <= (dend-3); dp += 3) {
|
|
|
537 |
d64_2_f32 d0, d1;
|
|
|
538 |
d0.d64 = sp2[0];
|
|
|
539 |
d1.d64 = sp2[1];
|
|
|
540 |
dp[0] = d0.f32s.f0;
|
|
|
541 |
dp[1] = d0.f32s.f1;
|
|
|
542 |
dp[2] = d1.f32s.f0;
|
|
|
543 |
sp2 += 2;
|
|
|
544 |
}
|
|
|
545 |
if (dp < dend) {
|
|
|
546 |
d0.d64 = sp2[0];
|
|
|
547 |
*dp++ = d0.f32s.f0;
|
|
|
548 |
if (dp < dend) *dp++ = d0.f32s.f1;
|
|
|
549 |
}
|
|
|
550 |
} else {
|
|
|
551 |
sp2 = (mlib_d64 *)(sp - 1);
|
|
|
552 |
#pragma pipeloop(0)
|
|
|
553 |
for (; dp <= (dend-3); dp += 3) {
|
|
|
554 |
d64_2_f32 d0, d1;
|
|
|
555 |
d0.d64 = sp2[0];
|
|
|
556 |
d1.d64 = sp2[1];
|
|
|
557 |
dp[0] = d0.f32s.f1;
|
|
|
558 |
dp[1] = d1.f32s.f0;
|
|
|
559 |
dp[2] = d1.f32s.f1;
|
|
|
560 |
sp2 += 2;
|
|
|
561 |
}
|
|
|
562 |
if (dp < dend) {
|
|
|
563 |
d0.d64 = sp2[0];
|
|
|
564 |
d1.d64 = sp2[1];
|
|
|
565 |
*dp++ = d0.f32s.f1;
|
|
|
566 |
if (dp < dend) *dp++ = d1.f32s.f0;
|
|
|
567 |
}
|
|
|
568 |
}
|
|
|
569 |
|
|
|
570 |
sp = sl += slb;
|
|
|
571 |
dp = dl += dlb;
|
|
|
572 |
}
|
|
|
573 |
}
|
|
|
574 |
|
|
|
575 |
/***************************************************************/
|
|
|
576 |
/* general channel extraction: slower due to the inner loop */
|
|
|
577 |
|
|
|
578 |
void mlib_v_ImageChannelExtract_U8(mlib_u8 *src, mlib_s32 slb,
|
|
|
579 |
mlib_u8 *dst, mlib_s32 dlb,
|
|
|
580 |
mlib_s32 channels, mlib_s32 channeld,
|
|
|
581 |
mlib_s32 width, mlib_s32 height,
|
|
|
582 |
mlib_s32 cmask)
|
|
|
583 |
{
|
|
|
584 |
mlib_u8 *sp; /* pointer for pixel in src */
|
|
|
585 |
mlib_u8 *sl; /* pointer for line in src */
|
|
|
586 |
mlib_u8 *dp; /* pointer for pixel in dst */
|
|
|
587 |
mlib_u8 *dl; /* pointer for line in dst */
|
|
|
588 |
int i, j, k; /* indices for x, y, channel */
|
|
|
589 |
int deltac[5] = { 0, 1, 1, 1, 1 };
|
|
|
590 |
int inc0, inc1, inc2, inc3;
|
|
|
591 |
mlib_u8 s0, s1, s2, s3;
|
|
|
592 |
|
|
|
593 |
deltac[channeld] = 1;
|
|
|
594 |
for (i = (channels - 1), k = 0; i >= 0; i--) {
|
|
|
595 |
if ((cmask & (1 << i)) == 0)
|
|
|
596 |
deltac[k]++;
|
|
|
597 |
else
|
|
|
598 |
k++;
|
|
|
599 |
}
|
|
|
600 |
|
|
|
601 |
deltac[channeld] = channels;
|
|
|
602 |
for (i = 1; i < channeld; i++) {
|
|
|
603 |
deltac[channeld] -= deltac[i];
|
|
|
604 |
}
|
|
|
605 |
|
|
|
606 |
sp = sl = src + deltac[0];
|
|
|
607 |
dp = dl = dst;
|
|
|
608 |
|
|
|
609 |
/* Only THREE CHANNEL CASE could be executed here!!! */
|
|
|
610 |
|
|
|
611 |
inc0 = deltac[1];
|
|
|
612 |
inc1 = deltac[2] + inc0;
|
|
|
613 |
inc2 = deltac[3] + inc1;
|
|
|
614 |
for (j = 0; j < height; j++) {
|
|
|
615 |
for (i = 0; i < width; i++) {
|
|
|
616 |
#pragma pipeloop(0)
|
|
|
617 |
s0 = sp[0]; s1 = sp[inc0]; s2 = sp[inc1];
|
|
|
618 |
dp[0] = s0;
|
|
|
619 |
dp[1] = s1;
|
|
|
620 |
dp[2] = s2;
|
|
|
621 |
sp += inc2;
|
|
|
622 |
dp += 3;
|
|
|
623 |
}
|
|
|
624 |
sp = sl += slb;
|
|
|
625 |
dp = dl += dlb;
|
|
|
626 |
}
|
|
|
627 |
}
|
|
|
628 |
|
|
|
629 |
/***************************************************************/
|
|
|
630 |
/* general channel extraction: slower due to the inner loop */
|
|
|
631 |
|
|
|
632 |
void mlib_v_ImageChannelExtract_S16(mlib_u16 *src, mlib_s32 slb,
|
|
|
633 |
mlib_u16 *dst, mlib_s32 dlb,
|
|
|
634 |
mlib_s32 channels, mlib_s32 channeld,
|
|
|
635 |
mlib_s32 width, mlib_s32 height,
|
|
|
636 |
mlib_s32 cmask)
|
|
|
637 |
{
|
|
|
638 |
mlib_u16 *sp; /* pointer for pixel in src */
|
|
|
639 |
mlib_u16 *sl; /* pointer for line in src */
|
|
|
640 |
mlib_u16 *dp; /* pointer for pixel in dst */
|
|
|
641 |
mlib_u16 *dl; /* pointer for line in dst */
|
|
|
642 |
int i, j, k; /* indices for x, y, channel */
|
|
|
643 |
int deltac[5] = { 0, 1, 1, 1, 1 };
|
|
|
644 |
int inc0, inc1, inc2, inc3;
|
|
|
645 |
mlib_u16 s0, s1, s2, s3;
|
|
|
646 |
|
|
|
647 |
slb >>= 1;
|
|
|
648 |
dlb >>= 1;
|
|
|
649 |
|
|
|
650 |
deltac[channeld] = 1;
|
|
|
651 |
for (i = (channels - 1), k = 0; i >= 0; i--) {
|
|
|
652 |
if ((cmask & (1 << i)) == 0)
|
|
|
653 |
deltac[k]++;
|
|
|
654 |
else
|
|
|
655 |
k++;
|
|
|
656 |
}
|
|
|
657 |
|
|
|
658 |
deltac[channeld] = channels;
|
|
|
659 |
for (i = 1; i < channeld; i++) {
|
|
|
660 |
deltac[channeld] -= deltac[i];
|
|
|
661 |
}
|
|
|
662 |
|
|
|
663 |
sp = sl = src + deltac[0];
|
|
|
664 |
dp = dl = dst;
|
|
|
665 |
|
|
|
666 |
if (channeld == 2) {
|
|
|
667 |
inc0 = deltac[1];
|
|
|
668 |
inc1 = deltac[2] + inc0;
|
|
|
669 |
for (j = 0; j < height; j++) {
|
|
|
670 |
#pragma pipeloop(0)
|
|
|
671 |
for (i = 0; i < width; i++) {
|
|
|
672 |
s0 = sp[0]; s1 = sp[inc0];
|
|
|
673 |
dp[0] = s0;
|
|
|
674 |
dp[1] = s1;
|
|
|
675 |
sp += inc1;
|
|
|
676 |
dp += 2;
|
|
|
677 |
}
|
|
|
678 |
sp = sl = sl + slb;
|
|
|
679 |
dp = dl = dl + dlb;
|
|
|
680 |
}
|
|
|
681 |
} else
|
|
|
682 |
|
|
|
683 |
if (channeld == 3) {
|
|
|
684 |
inc0 = deltac[1];
|
|
|
685 |
inc1 = deltac[2] + inc0;
|
|
|
686 |
inc2 = deltac[3] + inc1;
|
|
|
687 |
for (j = 0; j < height; j++) {
|
|
|
688 |
for (i = 0; i < width; i++) {
|
|
|
689 |
#pragma pipeloop(0)
|
|
|
690 |
s0 = sp[0]; s1 = sp[inc0]; s2 = sp[inc1];
|
|
|
691 |
dp[0] = s0;
|
|
|
692 |
dp[1] = s1;
|
|
|
693 |
dp[2] = s2;
|
|
|
694 |
sp += inc2;
|
|
|
695 |
dp += 3;
|
|
|
696 |
}
|
|
|
697 |
sp = sl = sl + slb;
|
|
|
698 |
dp = dl = dl + dlb;
|
|
|
699 |
}
|
|
|
700 |
}}
|
|
|
701 |
|
|
|
702 |
/***************************************************************/
|
|
|
703 |
/* general channel extraction: slower due to the inner loop */
|
|
|
704 |
|
|
|
705 |
void mlib_v_ImageChannelExtract_D64(mlib_d64 *src, mlib_s32 slb,
|
|
|
706 |
mlib_d64 *dst, mlib_s32 dlb,
|
|
|
707 |
mlib_s32 channels, mlib_s32 channeld,
|
|
|
708 |
mlib_s32 width, mlib_s32 height,
|
|
|
709 |
mlib_s32 cmask)
|
|
|
710 |
{
|
|
|
711 |
mlib_d64 *sp; /* pointer for pixel in src */
|
|
|
712 |
mlib_d64 *sl; /* pointer for line in src */
|
|
|
713 |
mlib_d64 *dp; /* pointer for pixel in dst */
|
|
|
714 |
mlib_d64 *dl; /* pointer for line in dst */
|
|
|
715 |
int i, j, k; /* indices for x, y, channel */
|
|
|
716 |
int deltac[5] = { 0, 1, 1, 1, 1 };
|
|
|
717 |
int inc0, inc1, inc2, inc3;
|
|
|
718 |
mlib_d64 s0, s1, s2, s3;
|
|
|
719 |
|
|
|
720 |
deltac[channeld] = 1;
|
|
|
721 |
for (i = (channels - 1), k = 0; i >= 0; i--) {
|
|
|
722 |
if ((cmask & (1 << i)) == 0)
|
|
|
723 |
deltac[k]++;
|
|
|
724 |
else
|
|
|
725 |
k++;
|
|
|
726 |
}
|
|
|
727 |
|
|
|
728 |
deltac[channeld] = channels;
|
|
|
729 |
for (i = 1; i < channeld; i++) {
|
|
|
730 |
deltac[channeld] -= deltac[i];
|
|
|
731 |
}
|
|
|
732 |
|
|
|
733 |
sp = sl = src + deltac[0];
|
|
|
734 |
dp = dl = dst;
|
|
|
735 |
|
|
|
736 |
if (channeld == 1) {
|
|
|
737 |
for (j = 0; j < height; j++) {
|
|
|
738 |
#pragma pipeloop(0)
|
|
|
739 |
for (i = 0; i < width; i++) {
|
|
|
740 |
s0 = sp[0];
|
|
|
741 |
dp[i] = s0;
|
|
|
742 |
sp += channels;
|
|
|
743 |
}
|
|
|
744 |
sp = sl = (mlib_d64 *)((mlib_u8 *)sl + slb);
|
|
|
745 |
dp = dl = (mlib_d64 *)((mlib_u8 *)dl + dlb);
|
|
|
746 |
}
|
|
|
747 |
} else
|
|
|
748 |
|
|
|
749 |
if (channeld == 2) {
|
|
|
750 |
inc0 = deltac[1];
|
|
|
751 |
inc1 = deltac[2] + inc0;
|
|
|
752 |
for (j = 0; j < height; j++) {
|
|
|
753 |
#pragma pipeloop(0)
|
|
|
754 |
for (i = 0; i < width; i++) {
|
|
|
755 |
s0 = sp[0]; s1 = sp[inc0];
|
|
|
756 |
dp[0] = s0;
|
|
|
757 |
dp[1] = s1;
|
|
|
758 |
sp += inc1;
|
|
|
759 |
dp += 2;
|
|
|
760 |
}
|
|
|
761 |
sp = sl = (mlib_d64 *)((mlib_u8 *)sl + slb);
|
|
|
762 |
dp = dl = (mlib_d64 *)((mlib_u8 *)dl + dlb);
|
|
|
763 |
}
|
|
|
764 |
} else
|
|
|
765 |
|
|
|
766 |
if (channeld == 3) {
|
|
|
767 |
inc0 = deltac[1];
|
|
|
768 |
inc1 = deltac[2] + inc0;
|
|
|
769 |
inc2 = deltac[3] + inc1;
|
|
|
770 |
for (j = 0; j < height; j++) {
|
|
|
771 |
for (i = 0; i < width; i++) {
|
|
|
772 |
#pragma pipeloop(0)
|
|
|
773 |
s0 = sp[0]; s1 = sp[inc0]; s2 = sp[inc1];
|
|
|
774 |
dp[0] = s0;
|
|
|
775 |
dp[1] = s1;
|
|
|
776 |
dp[2] = s2;
|
|
|
777 |
sp += inc2;
|
|
|
778 |
dp += 3;
|
|
|
779 |
}
|
|
|
780 |
sp = sl = (mlib_d64 *)((mlib_u8 *)sl + slb);
|
|
|
781 |
dp = dl = (mlib_d64 *)((mlib_u8 *)dl + dlb);
|
|
|
782 |
}
|
|
|
783 |
}
|
|
|
784 |
}
|