1 /******************************************************************************
2 *
3 * Copyright (C) 2015 The Android Open Source Project
4 *
5 * Licensed under the Apache License, Version 2.0 (the "License");
6 * you may not use this file except in compliance with the License.
7 * You may obtain a copy of the License at:
8 *
9 * http://www.apache.org/licenses/LICENSE-2.0
10 *
11 * Unless required by applicable law or agreed to in writing, software
12 * distributed under the License is distributed on an "AS IS" BASIS,
13 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14 * See the License for the specific language governing permissions and
15 * limitations under the License.
16 *
17 *****************************************************************************
18 * Originally developed and contributed by Ittiam Systems Pvt. Ltd, Bangalore
19 */
20 /**
21 *******************************************************************************
22 * @file
23 * ih264_chroma_intra_pred_filters.c
24 *
25 * @brief
26 * Contains function definitions for chroma intra prediction filters
27 *
28 * @author
29 * Ittiam
30 *
31 * @par List of Functions:
32 * -ih264_intra_pred_chroma_8x8_mode_dc
33 * -ih264_intra_pred_chroma_8x8_mode_horz
34 * -ih264_intra_pred_chroma_8x8_mode_vert
35 * -ih264_intra_pred_chroma_8x8_mode_plane
36 *
37 * @remarks
38 * None
39 *
40 *******************************************************************************
41 */
42
43 /*****************************************************************************/
44 /* File Includes */
45 /*****************************************************************************/
46
47 /* System include files */
48 #include <stdio.h>
49 #include <stddef.h>
50 #include <string.h>
51
52 /* User include files */
53 #include "ih264_defs.h"
54 #include "ih264_typedefs.h"
55 #include "ih264_macros.h"
56 #include "ih264_platform_macros.h"
57 #include "ih264_intra_pred_filters.h"
58
59 /* Global variables used only in assembly files*/
60 const WORD8 ih264_gai1_intrapred_chroma_plane_coeffs1[] =
61 { 0x01,0x00,0x01,0x00,
62 0x02,0x00,0x02,0x00,
63 0x03,0x00,0x03,0x00,
64 0x04,0x00,0x04,0x00
65 };
66 const WORD8 ih264_gai1_intrapred_chroma_plane_coeffs2[] =
67 { 0xfd,0xff,0xfe,0xff,
68 0xff,0xff,0x00,0x00,
69 0x01,0x00,0x02,0x00,
70 0x03,0x00,0x04,0x00,
71 };
72
73 /*****************************************************************************/
74 /* Chroma Intra prediction 8x8 filters */
75 /*****************************************************************************/
76
77 /**
78 *******************************************************************************
79 *
80 * ih264_intra_pred_chroma_8x8_mode_dc
81 *
82 * @brief
83 * Perform Intra prediction for chroma_8x8 mode:DC
84 *
85 * @par Description:
86 * Perform Intra prediction for chroma_8x8 mode:DC ,described in sec 8.3.4.1
87 *
88 * @param[in] pu1_src
89 * UWORD8 pointer to the source containing alternate U and V samples
90 *
91 * @param[out] pu1_dst
92 * UWORD8 pointer to the destination with alternate U and V samples
93 *
94 * @param[in] src_strd
95 * integer source stride
96 *
97 * @param[in] dst_strd
98 * integer destination stride
99 *
100 ** @param[in] ngbr_avail
101 * availability of neighbouring pixels
102 *
103 * @returns
104 *
105 * @remarks
106 * None
107 *
108 ******************************************************************************
109 */
ih264_intra_pred_chroma_8x8_mode_dc(UWORD8 * pu1_src,UWORD8 * pu1_dst,WORD32 src_strd,WORD32 dst_strd,WORD32 ngbr_avail)110 void ih264_intra_pred_chroma_8x8_mode_dc(UWORD8 *pu1_src,
111 UWORD8 *pu1_dst,
112 WORD32 src_strd,
113 WORD32 dst_strd,
114 WORD32 ngbr_avail)
115 {
116 WORD32 left_avail, left_avail1, left_avail2; /* availability of left predictors (only for DC) */
117 WORD32 top_avail; /* availability of top predictors (only for DC) */
118 UWORD8 *pu1_left = NULL; /* Pointer to start of left predictors */
119 UWORD8 *pu1_top = NULL; /* Pointer to start of top predictors */
120
121 /* temporary variables to store accumulated first left half,second left half,
122 * first top half,second top half of U and V values*/
123 WORD32 val_u_l1 = 0, val_u_l2 = 0, val_u_t1 = 0, val_u_t2 = 0;
124 WORD32 val_v_l1 = 0, val_v_l2 = 0, val_v_t1 = 0, val_v_t2 = 0;
125
126 WORD32 val_u1 = 0, val_u2 = 0, val_v1 = 0, val_v2 = 0;
127
128 WORD32 col, row; /*loop variables*/
129 UNUSED(src_strd);
130
131 left_avail = ngbr_avail & 0x11;
132 left_avail1 = ngbr_avail & 1;
133 left_avail2 = (ngbr_avail >> 4) & 1;
134 top_avail = (ngbr_avail >> 2) & 1;
135
136 pu1_top = pu1_src + 2 * BLK8x8SIZE + 2;
137 pu1_left = pu1_src + 2 * BLK8x8SIZE - 2;
138
139 if(left_avail1)
140 { /* First 4x4 block*/
141 val_u_l1 += *pu1_left;
142 val_v_l1 += *(pu1_left + 1);
143 pu1_left -= 2;
144 val_u_l1 += *pu1_left;
145 val_v_l1 += *(pu1_left + 1);
146 pu1_left -= 2;
147 val_u_l1 += *pu1_left;
148 val_v_l1 += *(pu1_left + 1);
149 pu1_left -= 2;
150 val_u_l1 += *pu1_left + 2;
151 val_v_l1 += *(pu1_left + 1) + 2;
152 pu1_left -= 2;
153 }
154 else
155 pu1_left -= 2 * 4;
156
157 if(left_avail2)
158 {
159 /* Second 4x4 block*/
160 val_u_l2 += *pu1_left;
161 val_v_l2 += *(pu1_left + 1);
162 pu1_left -= 2;
163 val_u_l2 += *pu1_left;
164 val_v_l2 += *(pu1_left + 1);
165 pu1_left -= 2;
166 val_u_l2 += *pu1_left;
167 val_v_l2 += *(pu1_left + 1);
168 pu1_left -= 2;
169 val_u_l2 += *pu1_left + 2;
170 val_v_l2 += *(pu1_left + 1) + 2;
171 pu1_left -= 2;
172 }
173 else
174 pu1_left -= 2 * 4;
175
176 if(top_avail)
177 {
178 val_u_t1 += *pu1_top + *(pu1_top + 2) + *(pu1_top + 4)
179 + *(pu1_top + 6) + 2;
180 val_u_t2 += *(pu1_top + 8) + *(pu1_top + 10) + *(pu1_top + 12)
181 + *(pu1_top + 14) + 2;
182 val_v_t1 += *(pu1_top + 1) + *(pu1_top + 3) + *(pu1_top + 5)
183 + *(pu1_top + 7) + 2;
184 val_v_t2 += *(pu1_top + 9) + *(pu1_top + 11) + *(pu1_top + 13)
185 + *(pu1_top + 15) + 2;
186 }
187
188 if(left_avail + top_avail)
189 {
190 val_u1 = (left_avail1 + top_avail) ?
191 ((val_u_l1 + val_u_t1)
192 >> (1 + left_avail1 + top_avail)) :128;
193 val_v1 = (left_avail1 + top_avail) ?
194 ((val_v_l1 + val_v_t1)
195 >> (1 + left_avail1 + top_avail)) :128;
196 if(top_avail)
197 {
198 val_u2 = val_u_t2 >> 2;
199 val_v2 = val_v_t2 >> 2;
200 }
201 else if(left_avail1)
202 {
203 val_u2 = val_u_l1 >> 2;
204 val_v2 = val_v_l1 >> 2;
205 }
206 else
207 {
208 val_u2 = val_v2 = 128;
209 }
210
211 for(row = 0; row < 4; row++)
212 {
213 /*top left 4x4 block*/
214 for(col = 0; col < 8; col += 2)
215 {
216 *(pu1_dst + row * dst_strd + col) = val_u1;
217 *(pu1_dst + row * dst_strd + col + 1) = val_v1;
218 }
219 /*top right 4x4 block*/
220 for(col = 8; col < 16; col += 2)
221 {
222 *(pu1_dst + row * dst_strd + col) = val_u2;
223 *(pu1_dst + row * dst_strd + col + 1) = val_v2;
224 }
225 }
226
227 if(left_avail2)
228 {
229 val_u1 = val_u_l2 >> 2;
230 val_v1 = val_v_l2 >> 2;
231 }
232 else if(top_avail)
233 {
234 val_u1 = val_u_t1 >> 2;
235 val_v1 = val_v_t1 >> 2;
236 }
237 else
238 {
239 val_u1 = val_v1 = 128;
240 }
241 val_u2 = (left_avail2 + top_avail) ?
242 ((val_u_l2 + val_u_t2)
243 >> (1 + left_avail2 + top_avail)) : 128;
244 val_v2 = (left_avail2 + top_avail) ?
245 ((val_v_l2 + val_v_t2)
246 >> (1 + left_avail2 + top_avail)) : 128;
247
248 for(row = 4; row < 8; row++)
249 { /*bottom left 4x4 block*/
250 for(col = 0; col < 8; col += 2)
251 {
252 *(pu1_dst + row * dst_strd + col) = val_u1;
253 *(pu1_dst + row * dst_strd + col + 1) = val_v1;
254 }
255 /*bottom right 4x4 block*/
256 for(col = 8; col < 16; col += 2)
257 {
258 *(pu1_dst + row * dst_strd + col) = val_u2;
259 *(pu1_dst + row * dst_strd + col + 1) = val_v2;
260 }
261 }
262 }
263 else
264 {
265 /* Both left and top are unavailable, set the block to 128 */
266 for(row = 0; row < 8; row++)
267 {
268 memset(pu1_dst + row * dst_strd, 128, 8 * sizeof(UWORD16));
269 }
270 }
271 }
272
273 /**
274 *******************************************************************************
275 *
276 *ih264_intra_pred_chroma_8x8_mode_horz
277 *
278 * @brief
279 * Perform Intra prediction for chroma_8x8 mode:Horizontal
280 *
281 * @par Description:
282 * Perform Intra prediction for chroma_8x8 mode:Horizontal ,described in sec 8.3.4.2
283 *
284 * @param[in] pu1_src
285 * UWORD8 pointer to the source containing alternate U and V samples
286 *
287 * @param[out] pu1_dst
288 * UWORD8 pointer to the destination with alternate U and V samples
289 *
290 * @param[in] src_strd
291 * integer source stride
292 *
293 * @param[in] dst_strd
294 * integer destination stride
295 *
296 * @param[in] ngbr_avail
297 * availability of neighbouring pixels(Not used in this function)
298 *
299 * @returns
300 *
301 * @remarks
302 * None
303 *
304 ******************************************************************************
305 */
ih264_intra_pred_chroma_8x8_mode_horz(UWORD8 * pu1_src,UWORD8 * pu1_dst,WORD32 src_strd,WORD32 dst_strd,WORD32 ngbr_avail)306 void ih264_intra_pred_chroma_8x8_mode_horz(UWORD8 *pu1_src,
307 UWORD8 *pu1_dst,
308 WORD32 src_strd,
309 WORD32 dst_strd,
310 WORD32 ngbr_avail)
311 {
312
313 UWORD8 *pu1_left = NULL; /* Pointer to start of top predictors */
314 WORD32 rows, cols; /* loop variables*/
315 UNUSED(src_strd);
316 UNUSED(ngbr_avail);
317 pu1_left = pu1_src + 2 * BLK8x8SIZE - 2;
318 for(rows = 0; rows < 8; rows++)
319 {
320 for(cols = 0; cols < 16; cols += 2)
321 {
322 *(pu1_dst + rows * dst_strd + cols) = *pu1_left;
323
324 *(pu1_dst + rows * dst_strd + cols + 1) = *(pu1_left + 1);
325 }
326 pu1_left -= 2;
327 }
328
329 }
330
331 /**
332 *******************************************************************************
333 *
334 *ih264_intra_pred_chroma_8x8_mode_vert
335 *
336 * @brief
337 * Perform Intra prediction for chroma_8x8 mode:vertical
338 *
339 * @par Description:
340 * Perform Intra prediction for chroma_8x8 mode:vertical ,described in sec 8.3.4.3
341 *
342 * @param[in] pu1_src
343 * UWORD8 pointer to the source containing alternate U and V samples
344 *
345 * @param[out] pu1_dst
346 * UWORD8 pointer to the destination with alternate U and V samples
347 *
348 * @param[in] src_strd
349 * integer source stride
350 *
351 * @param[in] dst_strd
352 * integer destination stride
353 *
354 * @param[in] ngbr_avail
355 * availability of neighbouring pixels(Not used in this function)
356 *
357 * @returns
358 *
359 * @remarks
360 * None
361 *
362 *******************************************************************************
363 */
ih264_intra_pred_chroma_8x8_mode_vert(UWORD8 * pu1_src,UWORD8 * pu1_dst,WORD32 src_strd,WORD32 dst_strd,WORD32 ngbr_avail)364 void ih264_intra_pred_chroma_8x8_mode_vert(UWORD8 *pu1_src,
365 UWORD8 *pu1_dst,
366 WORD32 src_strd,
367 WORD32 dst_strd,
368 WORD32 ngbr_avail)
369 {
370
371 UWORD8 *pu1_top = NULL; /* Pointer to start of top predictors */
372 WORD32 row;/*loop variable*/
373 UNUSED(src_strd);
374 UNUSED(ngbr_avail);
375 pu1_top = pu1_src + 2 * BLK8x8SIZE + 2;
376
377 /* 8 bytes are copied from src to dst */
378 for(row = 0; row < 2; row++)
379 {
380 memcpy(pu1_dst, pu1_top, 16);
381
382 pu1_dst += dst_strd;
383 memcpy(pu1_dst, pu1_top, 16);
384
385 pu1_dst += dst_strd;
386 memcpy(pu1_dst, pu1_top, 16);
387
388 pu1_dst += dst_strd;
389 memcpy(pu1_dst, pu1_top, 16);
390
391 pu1_dst += dst_strd;
392 }
393 }
394
395 /**
396 *******************************************************************************
397 *
398 * ih264_intra_pred_chroma_8x8_mode_plane
399 *
400 * @brief
401 * Perform Intra prediction for chroma_8x8 mode:PLANE
402 *
403 * @par Description:
404 * Perform Intra prediction for chroma_8x8 mode:PLANE ,described in sec 8.3.4.4
405 *
406 * @param[in] pu1_src
407 * UWORD8 pointer to the source containing alternate U and V samples
408 *
409 * @param[out] pu1_dst
410 * UWORD8 pointer to the destination with alternate U and V samples
411 *
412 * @param[in] src_strd
413 * integer source stride
414 *
415 * @param[in] dst_strd
416 * integer destination stride
417 *
418 * @param[in] ngbr_avail
419 * availability of neighbouring pixels(Not used in this function)
420 *
421 * @returns
422 *
423 * @remarks
424 * None
425 *
426 ******************************************************************************
427 */
ih264_intra_pred_chroma_8x8_mode_plane(UWORD8 * pu1_src,UWORD8 * pu1_dst,WORD32 src_strd,WORD32 dst_strd,WORD32 ngbr_avail)428 void ih264_intra_pred_chroma_8x8_mode_plane(UWORD8 *pu1_src,
429 UWORD8 *pu1_dst,
430 WORD32 src_strd,
431 WORD32 dst_strd,
432 WORD32 ngbr_avail)
433 {
434
435 UWORD8 *pu1_left = NULL; /* Pointer to start of left predictors */
436 UWORD8 *pu1_top = NULL; /* Pointer to start of top predictors */
437 WORD32 val = 0;
438 WORD32 rows, cols; /* loop variables*/
439 WORD32 a_u, b_u, c_u, h_u, v_u; /* Implementing section 8.3.4.4 . The variables represent the corresponding variables in the section*/
440 WORD32 a_v, b_v, c_v, h_v, v_v;
441 UNUSED(src_strd);
442 UNUSED(ngbr_avail);
443 a_u = b_u = c_u = h_u = v_u = 0;
444 a_v = b_v = c_v = h_v = v_v = 0;
445 /* As chroma format 4:2:0 is used,xCF = 4 * ( chroma_format_idc = = 3 ) = 0 and
446 yCF = 4 * ( chroma_format_idc != 1 ) = 0 */
447 pu1_top = pu1_src + 2 * BLK8x8SIZE + 2;
448 pu1_left = pu1_src + 2 * BLK8x8SIZE - 2;
449 /* Implementing section 8.3.4.4 */
450 for(cols = 0; cols < 4; cols++)
451 {
452 h_u += (cols + 1) * (pu1_top[8 + 2 * cols] - pu1_top[4 - 2 * cols]);/*section 8.3.4.4 equation (8-144)*/
453 h_v += (cols + 1) * (pu1_top[8 + 2 * cols + 1] - pu1_top[4 - 2 * cols+ 1]);
454
455 v_u += (cols + 1) * (pu1_left[(4 + cols) * (-2)] - pu1_left[(2 - cols) * (-2)]);
456 v_v += (cols + 1) * (pu1_left[(4 + cols) * (-2) + 1] - pu1_left[(2 - cols) * (-2) + 1]);/*section 8.3.4.4 equation (8-145)*/
457 }
458 a_u = 16 * (pu1_left[7 * (-2)] + pu1_top[14]);
459 a_v = 16 * (pu1_left[7 * (-2) + 1] + pu1_top[15]);/*section 8.3.3.4 equation (8-141)*/
460 b_u = (34 * h_u + 32) >> 6;/*section 8.3.3.4 equation (8-142)*/
461 b_v = (34 * h_v + 32) >> 6;/*section 8.3.3.4 equation (8-142)*/
462 c_u = (34 * v_u + 32) >> 6;/*section 8.3.3.4 equation (8-143)*/
463 c_v = (34 * v_v + 32) >> 6;/*section 8.3.3.4 equation (8-143)*/
464
465 for(rows = 0; rows < 8; rows++)
466 {
467 for(cols = 0; cols < 8; cols++)
468 {
469 val = (a_u + b_u * (cols - 3) + c_u * (rows - 3) );/*section 8.3.4.4 equation (8-140)*/
470 val = (val + 16) >> 5;
471 *(pu1_dst + rows * dst_strd + 2 * cols) = CLIP_U8(val);
472 val = (a_v + b_v * (cols - 3) + c_v * (rows - 3) );/*section 8.3.4.4 equation (8-140)*/
473 val = (val + 16) >> 5;
474 *(pu1_dst + rows * dst_strd + 2 * cols + 1) = CLIP_U8(val);
475 }
476 }
477 }
478
479