1 | /* The copyright in this software is being made available under the BSD |
---|
2 | * License, included below. This software may be subject to other third party |
---|
3 | * and contributor rights, including patent rights, and no such rights are |
---|
4 | * granted under this license. |
---|
5 | * |
---|
6 | * Copyright (c) 2010-2013, ITU/ISO/IEC |
---|
7 | * All rights reserved. |
---|
8 | * |
---|
9 | * Redistribution and use in source and binary forms, with or without |
---|
10 | * modification, are permitted provided that the following conditions are met: |
---|
11 | * |
---|
12 | * * Redistributions of source code must retain the above copyright notice, |
---|
13 | * this list of conditions and the following disclaimer. |
---|
14 | * * Redistributions in binary form must reproduce the above copyright notice, |
---|
15 | * this list of conditions and the following disclaimer in the documentation |
---|
16 | * and/or other materials provided with the distribution. |
---|
17 | * * Neither the name of the ITU/ISO/IEC nor the names of its contributors may |
---|
18 | * be used to endorse or promote products derived from this software without |
---|
19 | * specific prior written permission. |
---|
20 | * |
---|
21 | * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" |
---|
22 | * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
---|
23 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
---|
24 | * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS |
---|
25 | * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR |
---|
26 | * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF |
---|
27 | * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS |
---|
28 | * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN |
---|
29 | * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) |
---|
30 | * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF |
---|
31 | * THE POSSIBILITY OF SUCH DAMAGE. |
---|
32 | */ |
---|
33 | |
---|
34 | /** \file TComTrQuant.cpp |
---|
35 | \brief transform and quantization class |
---|
36 | */ |
---|
37 | |
---|
38 | #include <stdlib.h> |
---|
39 | #include <math.h> |
---|
40 | #include <memory.h> |
---|
41 | #include "TComTrQuant.h" |
---|
42 | #include "TComPic.h" |
---|
43 | #include "ContextTables.h" |
---|
44 | #define MAYBE_BUGFIX 1 |
---|
45 | typedef struct |
---|
46 | { |
---|
47 | Int iNNZbeforePos0; |
---|
48 | Double d64CodedLevelandDist; // distortion and level cost only |
---|
49 | Double d64UncodedDist; // all zero coded block distortion |
---|
50 | Double d64SigCost; |
---|
51 | Double d64SigCost_0; |
---|
52 | #if MAYBE_BUGFIX |
---|
53 | Void init() |
---|
54 | { |
---|
55 | iNNZbeforePos0 = 0; |
---|
56 | d64CodedLevelandDist = 0; |
---|
57 | d64UncodedDist = 0; |
---|
58 | d64SigCost = 0; |
---|
59 | d64SigCost_0 = 0; |
---|
60 | } |
---|
61 | #endif |
---|
62 | } coeffGroupRDStats; |
---|
63 | |
---|
64 | //! \ingroup TLibCommon |
---|
65 | //! \{ |
---|
66 | |
---|
67 | // ==================================================================================================================== |
---|
68 | // Constants |
---|
69 | // ==================================================================================================================== |
---|
70 | |
---|
71 | #define RDOQ_CHROMA 1 ///< use of RDOQ in chroma |
---|
72 | |
---|
73 | // ==================================================================================================================== |
---|
74 | // Tables |
---|
75 | // ==================================================================================================================== |
---|
76 | |
---|
77 | // RDOQ parameter |
---|
78 | |
---|
79 | // ==================================================================================================================== |
---|
80 | // Qp class member functions |
---|
81 | // ==================================================================================================================== |
---|
82 | |
---|
83 | QpParam::QpParam() |
---|
84 | { |
---|
85 | } |
---|
86 | |
---|
87 | // ==================================================================================================================== |
---|
88 | // TComTrQuant class member functions |
---|
89 | // ==================================================================================================================== |
---|
90 | |
---|
91 | TComTrQuant::TComTrQuant() |
---|
92 | { |
---|
93 | m_cQP.clear(); |
---|
94 | |
---|
95 | // allocate temporary buffers |
---|
96 | m_plTempCoeff = new Int[ MAX_CU_SIZE*MAX_CU_SIZE ]; |
---|
97 | |
---|
98 | // allocate bit estimation class (for RDOQ) |
---|
99 | m_pcEstBitsSbac = new estBitsSbacStruct; |
---|
100 | initScalingList(); |
---|
101 | } |
---|
102 | |
---|
103 | TComTrQuant::~TComTrQuant() |
---|
104 | { |
---|
105 | // delete temporary buffers |
---|
106 | if ( m_plTempCoeff ) |
---|
107 | { |
---|
108 | delete [] m_plTempCoeff; |
---|
109 | m_plTempCoeff = NULL; |
---|
110 | } |
---|
111 | |
---|
112 | // delete bit estimation class |
---|
113 | if ( m_pcEstBitsSbac ) |
---|
114 | { |
---|
115 | delete m_pcEstBitsSbac; |
---|
116 | } |
---|
117 | destroyScalingList(); |
---|
118 | } |
---|
119 | |
---|
120 | #if ADAPTIVE_QP_SELECTION |
---|
121 | Void TComTrQuant::storeSliceQpNext(TComSlice* pcSlice) |
---|
122 | { |
---|
123 | Int qpBase = pcSlice->getSliceQpBase(); |
---|
124 | Int sliceQpused = pcSlice->getSliceQp(); |
---|
125 | Int sliceQpnext; |
---|
126 | Double alpha = qpBase < 17 ? 0.5 : 1; |
---|
127 | |
---|
128 | Int cnt=0; |
---|
129 | for(Int u=1; u<=LEVEL_RANGE; u++) |
---|
130 | { |
---|
131 | cnt += m_sliceNsamples[u] ; |
---|
132 | } |
---|
133 | |
---|
134 | if( !m_useRDOQ ) |
---|
135 | { |
---|
136 | sliceQpused = qpBase; |
---|
137 | alpha = 0.5; |
---|
138 | } |
---|
139 | |
---|
140 | if( cnt > 120 ) |
---|
141 | { |
---|
142 | Double sum = 0; |
---|
143 | Int k = 0; |
---|
144 | for(Int u=1; u<LEVEL_RANGE; u++) |
---|
145 | { |
---|
146 | sum += u*m_sliceSumC[u]; |
---|
147 | k += u*u*m_sliceNsamples[u]; |
---|
148 | } |
---|
149 | |
---|
150 | Int v; |
---|
151 | Double q[MAX_QP+1] ; |
---|
152 | for(v=0; v<=MAX_QP; v++) |
---|
153 | { |
---|
154 | q[v] = (Double)(g_invQuantScales[v%6] * (1<<(v/6)))/64 ; |
---|
155 | } |
---|
156 | |
---|
157 | Double qnext = sum/k * q[sliceQpused] / (1<<ARL_C_PRECISION); |
---|
158 | |
---|
159 | for(v=0; v<MAX_QP; v++) |
---|
160 | { |
---|
161 | if(qnext < alpha * q[v] + (1 - alpha) * q[v+1] ) |
---|
162 | { |
---|
163 | break; |
---|
164 | } |
---|
165 | } |
---|
166 | sliceQpnext = Clip3(sliceQpused - 3, sliceQpused + 3, v); |
---|
167 | } |
---|
168 | else |
---|
169 | { |
---|
170 | sliceQpnext = sliceQpused; |
---|
171 | } |
---|
172 | |
---|
173 | m_qpDelta[qpBase] = sliceQpnext - qpBase; |
---|
174 | } |
---|
175 | |
---|
176 | Void TComTrQuant::initSliceQpDelta() |
---|
177 | { |
---|
178 | for(Int qp=0; qp<=MAX_QP; qp++) |
---|
179 | { |
---|
180 | m_qpDelta[qp] = qp < 17 ? 0 : 1; |
---|
181 | } |
---|
182 | } |
---|
183 | |
---|
184 | Void TComTrQuant::clearSliceARLCnt() |
---|
185 | { |
---|
186 | memset(m_sliceSumC, 0, sizeof(Double)*(LEVEL_RANGE+1)); |
---|
187 | memset(m_sliceNsamples, 0, sizeof(Int)*(LEVEL_RANGE+1)); |
---|
188 | } |
---|
189 | #endif |
---|
190 | |
---|
191 | |
---|
192 | /** Set qP for Quantization. |
---|
193 | * \param qpy QPy |
---|
194 | * \param bLowpass |
---|
195 | * \param eSliceType |
---|
196 | * \param eTxtType |
---|
197 | * \param qpBdOffset |
---|
198 | * \param chromaQPOffset |
---|
199 | * |
---|
200 | * return void |
---|
201 | */ |
---|
202 | Void TComTrQuant::setQPforQuant( Int qpy, TextType eTxtType, Int qpBdOffset, Int chromaQPOffset) |
---|
203 | { |
---|
204 | Int qpScaled; |
---|
205 | |
---|
206 | if(eTxtType == TEXT_LUMA) |
---|
207 | { |
---|
208 | qpScaled = qpy + qpBdOffset; |
---|
209 | } |
---|
210 | else |
---|
211 | { |
---|
212 | qpScaled = Clip3( -qpBdOffset, 57, qpy + chromaQPOffset ); |
---|
213 | |
---|
214 | if(qpScaled < 0) |
---|
215 | { |
---|
216 | qpScaled = qpScaled + qpBdOffset; |
---|
217 | } |
---|
218 | else |
---|
219 | { |
---|
220 | qpScaled = g_aucChromaScale[ qpScaled ] + qpBdOffset; |
---|
221 | } |
---|
222 | } |
---|
223 | m_cQP.setQpParam( qpScaled ); |
---|
224 | } |
---|
225 | |
---|
226 | #if MATRIX_MULT |
---|
227 | /** NxN forward transform (2D) using brute force matrix multiplication (3 nested loops) |
---|
228 | * \param block pointer to input data (residual) |
---|
229 | * \param coeff pointer to output data (transform coefficients) |
---|
230 | * \param uiStride stride of input data |
---|
231 | * \param uiTrSize transform size (uiTrSize x uiTrSize) |
---|
232 | * \param uiMode is Intra Prediction mode used in Mode-Dependent DCT/DST only |
---|
233 | */ |
---|
234 | void xTr(Int bitDepth, Pel *block, Int *coeff, UInt uiStride, UInt uiTrSize, UInt uiMode) |
---|
235 | { |
---|
236 | Int i,j,k,iSum; |
---|
237 | Int tmp[32*32]; |
---|
238 | const Short *iT; |
---|
239 | UInt uiLog2TrSize = g_aucConvertToBit[ uiTrSize ] + 2; |
---|
240 | |
---|
241 | if (uiTrSize==4) |
---|
242 | { |
---|
243 | iT = g_aiT4[0]; |
---|
244 | } |
---|
245 | else if (uiTrSize==8) |
---|
246 | { |
---|
247 | iT = g_aiT8[0]; |
---|
248 | } |
---|
249 | else if (uiTrSize==16) |
---|
250 | { |
---|
251 | iT = g_aiT16[0]; |
---|
252 | } |
---|
253 | else if (uiTrSize==32) |
---|
254 | { |
---|
255 | iT = g_aiT32[0]; |
---|
256 | } |
---|
257 | else |
---|
258 | { |
---|
259 | assert(0); |
---|
260 | } |
---|
261 | |
---|
262 | Int shift_1st = uiLog2TrSize - 1 + bitDepth-8; // log2(N) - 1 + g_bitDepth-8 |
---|
263 | Int add_1st = 1<<(shift_1st-1); |
---|
264 | Int shift_2nd = uiLog2TrSize + 6; |
---|
265 | Int add_2nd = 1<<(shift_2nd-1); |
---|
266 | |
---|
267 | /* Horizontal transform */ |
---|
268 | |
---|
269 | if (uiTrSize==4) |
---|
270 | { |
---|
271 | if (uiMode != REG_DCT && g_aucDCTDSTMode_Hor[uiMode]) |
---|
272 | { |
---|
273 | iT = g_as_DST_MAT_4[0]; |
---|
274 | } |
---|
275 | } |
---|
276 | for (i=0; i<uiTrSize; i++) |
---|
277 | { |
---|
278 | for (j=0; j<uiTrSize; j++) |
---|
279 | { |
---|
280 | iSum = 0; |
---|
281 | for (k=0; k<uiTrSize; k++) |
---|
282 | { |
---|
283 | iSum += iT[i*uiTrSize+k]*block[j*uiStride+k]; |
---|
284 | } |
---|
285 | tmp[i*uiTrSize+j] = (iSum + add_1st)>>shift_1st; |
---|
286 | } |
---|
287 | } |
---|
288 | |
---|
289 | /* Vertical transform */ |
---|
290 | if (uiTrSize==4) |
---|
291 | { |
---|
292 | if (uiMode != REG_DCT && g_aucDCTDSTMode_Vert[uiMode]) |
---|
293 | { |
---|
294 | iT = g_as_DST_MAT_4[0]; |
---|
295 | } |
---|
296 | else |
---|
297 | { |
---|
298 | iT = g_aiT4[0]; |
---|
299 | } |
---|
300 | } |
---|
301 | for (i=0; i<uiTrSize; i++) |
---|
302 | { |
---|
303 | for (j=0; j<uiTrSize; j++) |
---|
304 | { |
---|
305 | iSum = 0; |
---|
306 | for (k=0; k<uiTrSize; k++) |
---|
307 | { |
---|
308 | iSum += iT[i*uiTrSize+k]*tmp[j*uiTrSize+k]; |
---|
309 | } |
---|
310 | coeff[i*uiTrSize+j] = (iSum + add_2nd)>>shift_2nd; |
---|
311 | } |
---|
312 | } |
---|
313 | } |
---|
314 | |
---|
315 | /** NxN inverse transform (2D) using brute force matrix multiplication (3 nested loops) |
---|
316 | * \param coeff pointer to input data (transform coefficients) |
---|
317 | * \param block pointer to output data (residual) |
---|
318 | * \param uiStride stride of output data |
---|
319 | * \param uiTrSize transform size (uiTrSize x uiTrSize) |
---|
320 | * \param uiMode is Intra Prediction mode used in Mode-Dependent DCT/DST only |
---|
321 | */ |
---|
322 | void xITr(Int *coeff, Pel *block, UInt uiStride, UInt uiTrSize, UInt uiMode) |
---|
323 | { |
---|
324 | Int i,j,k,iSum; |
---|
325 | Int tmp[32*32]; |
---|
326 | const Short *iT; |
---|
327 | |
---|
328 | if (uiTrSize==4) |
---|
329 | { |
---|
330 | iT = g_aiT4[0]; |
---|
331 | } |
---|
332 | else if (uiTrSize==8) |
---|
333 | { |
---|
334 | iT = g_aiT8[0]; |
---|
335 | } |
---|
336 | else if (uiTrSize==16) |
---|
337 | { |
---|
338 | iT = g_aiT16[0]; |
---|
339 | } |
---|
340 | else if (uiTrSize==32) |
---|
341 | { |
---|
342 | iT = g_aiT32[0]; |
---|
343 | } |
---|
344 | else |
---|
345 | { |
---|
346 | assert(0); |
---|
347 | } |
---|
348 | |
---|
349 | Int shift_1st = SHIFT_INV_1ST; |
---|
350 | Int add_1st = 1<<(shift_1st-1); |
---|
351 | Int shift_2nd = SHIFT_INV_2ND - g_bitDepth-8; |
---|
352 | Int add_2nd = 1<<(shift_2nd-1); |
---|
353 | if (uiTrSize==4) |
---|
354 | { |
---|
355 | if (uiMode != REG_DCT && g_aucDCTDSTMode_Vert[uiMode] ) // Check for DCT or DST |
---|
356 | { |
---|
357 | iT = g_as_DST_MAT_4[0]; |
---|
358 | } |
---|
359 | } |
---|
360 | |
---|
361 | /* Horizontal transform */ |
---|
362 | for (i=0; i<uiTrSize; i++) |
---|
363 | { |
---|
364 | for (j=0; j<uiTrSize; j++) |
---|
365 | { |
---|
366 | iSum = 0; |
---|
367 | for (k=0; k<uiTrSize; k++) |
---|
368 | { |
---|
369 | iSum += iT[k*uiTrSize+i]*coeff[k*uiTrSize+j]; |
---|
370 | } |
---|
371 | tmp[i*uiTrSize+j] = Clip3(-32768, 32767, (iSum + add_1st)>>shift_1st); // Clipping is normative |
---|
372 | } |
---|
373 | } |
---|
374 | |
---|
375 | if (uiTrSize==4) |
---|
376 | { |
---|
377 | if (uiMode != REG_DCT && g_aucDCTDSTMode_Hor[uiMode] ) // Check for DCT or DST |
---|
378 | { |
---|
379 | iT = g_as_DST_MAT_4[0]; |
---|
380 | } |
---|
381 | else |
---|
382 | { |
---|
383 | iT = g_aiT4[0]; |
---|
384 | } |
---|
385 | } |
---|
386 | |
---|
387 | /* Vertical transform */ |
---|
388 | for (i=0; i<uiTrSize; i++) |
---|
389 | { |
---|
390 | for (j=0; j<uiTrSize; j++) |
---|
391 | { |
---|
392 | iSum = 0; |
---|
393 | for (k=0; k<uiTrSize; k++) |
---|
394 | { |
---|
395 | iSum += iT[k*uiTrSize+j]*tmp[i*uiTrSize+k]; |
---|
396 | } |
---|
397 | block[i*uiStride+j] = Clip3(-32768, 32767, (iSum + add_2nd)>>shift_2nd); // Clipping is non-normative |
---|
398 | } |
---|
399 | } |
---|
400 | } |
---|
401 | |
---|
402 | #else //MATRIX_MULT |
---|
403 | |
---|
404 | /** 4x4 forward transform implemented using partial butterfly structure (1D) |
---|
405 | * \param src input data (residual) |
---|
406 | * \param dst output data (transform coefficients) |
---|
407 | * \param shift specifies right shift after 1D transform |
---|
408 | */ |
---|
409 | |
---|
410 | void partialButterfly4(Short *src,Short *dst,Int shift, Int line) |
---|
411 | { |
---|
412 | Int j; |
---|
413 | Int E[2],O[2]; |
---|
414 | Int add = 1<<(shift-1); |
---|
415 | |
---|
416 | for (j=0; j<line; j++) |
---|
417 | { |
---|
418 | /* E and O */ |
---|
419 | E[0] = src[0] + src[3]; |
---|
420 | O[0] = src[0] - src[3]; |
---|
421 | E[1] = src[1] + src[2]; |
---|
422 | O[1] = src[1] - src[2]; |
---|
423 | |
---|
424 | dst[0] = (g_aiT4[0][0]*E[0] + g_aiT4[0][1]*E[1] + add)>>shift; |
---|
425 | dst[2*line] = (g_aiT4[2][0]*E[0] + g_aiT4[2][1]*E[1] + add)>>shift; |
---|
426 | dst[line] = (g_aiT4[1][0]*O[0] + g_aiT4[1][1]*O[1] + add)>>shift; |
---|
427 | dst[3*line] = (g_aiT4[3][0]*O[0] + g_aiT4[3][1]*O[1] + add)>>shift; |
---|
428 | |
---|
429 | src += 4; |
---|
430 | dst ++; |
---|
431 | } |
---|
432 | } |
---|
433 | |
---|
434 | // Fast DST Algorithm. Full matrix multiplication for DST and Fast DST algorithm |
---|
435 | // give identical results |
---|
436 | void fastForwardDst(Short *block,Short *coeff,Int shift) // input block, output coeff |
---|
437 | { |
---|
438 | Int i, c[4]; |
---|
439 | Int rnd_factor = 1<<(shift-1); |
---|
440 | for (i=0; i<4; i++) |
---|
441 | { |
---|
442 | // Intermediate Variables |
---|
443 | c[0] = block[4*i+0] + block[4*i+3]; |
---|
444 | c[1] = block[4*i+1] + block[4*i+3]; |
---|
445 | c[2] = block[4*i+0] - block[4*i+1]; |
---|
446 | c[3] = 74* block[4*i+2]; |
---|
447 | |
---|
448 | coeff[ i] = ( 29 * c[0] + 55 * c[1] + c[3] + rnd_factor ) >> shift; |
---|
449 | coeff[ 4+i] = ( 74 * (block[4*i+0]+ block[4*i+1] - block[4*i+3]) + rnd_factor ) >> shift; |
---|
450 | coeff[ 8+i] = ( 29 * c[2] + 55 * c[0] - c[3] + rnd_factor ) >> shift; |
---|
451 | coeff[12+i] = ( 55 * c[2] - 29 * c[1] + c[3] + rnd_factor ) >> shift; |
---|
452 | } |
---|
453 | } |
---|
454 | |
---|
455 | void fastInverseDst(Short *tmp,Short *block,Int shift) // input tmp, output block |
---|
456 | { |
---|
457 | Int i, c[4]; |
---|
458 | Int rnd_factor = 1<<(shift-1); |
---|
459 | for (i=0; i<4; i++) |
---|
460 | { |
---|
461 | // Intermediate Variables |
---|
462 | c[0] = tmp[ i] + tmp[ 8+i]; |
---|
463 | c[1] = tmp[8+i] + tmp[12+i]; |
---|
464 | c[2] = tmp[ i] - tmp[12+i]; |
---|
465 | c[3] = 74* tmp[4+i]; |
---|
466 | |
---|
467 | block[4*i+0] = Clip3( -32768, 32767, ( 29 * c[0] + 55 * c[1] + c[3] + rnd_factor ) >> shift ); |
---|
468 | block[4*i+1] = Clip3( -32768, 32767, ( 55 * c[2] - 29 * c[1] + c[3] + rnd_factor ) >> shift ); |
---|
469 | block[4*i+2] = Clip3( -32768, 32767, ( 74 * (tmp[i] - tmp[8+i] + tmp[12+i]) + rnd_factor ) >> shift ); |
---|
470 | block[4*i+3] = Clip3( -32768, 32767, ( 55 * c[0] + 29 * c[2] - c[3] + rnd_factor ) >> shift ); |
---|
471 | } |
---|
472 | } |
---|
473 | |
---|
474 | void partialButterflyInverse4(Short *src,Short *dst,Int shift, Int line) |
---|
475 | { |
---|
476 | Int j; |
---|
477 | Int E[2],O[2]; |
---|
478 | Int add = 1<<(shift-1); |
---|
479 | |
---|
480 | for (j=0; j<line; j++) |
---|
481 | { |
---|
482 | /* Utilizing symmetry properties to the maximum to minimize the number of multiplications */ |
---|
483 | O[0] = g_aiT4[1][0]*src[line] + g_aiT4[3][0]*src[3*line]; |
---|
484 | O[1] = g_aiT4[1][1]*src[line] + g_aiT4[3][1]*src[3*line]; |
---|
485 | E[0] = g_aiT4[0][0]*src[0] + g_aiT4[2][0]*src[2*line]; |
---|
486 | E[1] = g_aiT4[0][1]*src[0] + g_aiT4[2][1]*src[2*line]; |
---|
487 | |
---|
488 | /* Combining even and odd terms at each hierarchy levels to calculate the final spatial domain vector */ |
---|
489 | dst[0] = Clip3( -32768, 32767, (E[0] + O[0] + add)>>shift ); |
---|
490 | dst[1] = Clip3( -32768, 32767, (E[1] + O[1] + add)>>shift ); |
---|
491 | dst[2] = Clip3( -32768, 32767, (E[1] - O[1] + add)>>shift ); |
---|
492 | dst[3] = Clip3( -32768, 32767, (E[0] - O[0] + add)>>shift ); |
---|
493 | |
---|
494 | src ++; |
---|
495 | dst += 4; |
---|
496 | } |
---|
497 | } |
---|
498 | |
---|
499 | |
---|
500 | void partialButterfly8(Short *src,Short *dst,Int shift, Int line) |
---|
501 | { |
---|
502 | Int j,k; |
---|
503 | Int E[4],O[4]; |
---|
504 | Int EE[2],EO[2]; |
---|
505 | Int add = 1<<(shift-1); |
---|
506 | |
---|
507 | for (j=0; j<line; j++) |
---|
508 | { |
---|
509 | /* E and O*/ |
---|
510 | for (k=0;k<4;k++) |
---|
511 | { |
---|
512 | E[k] = src[k] + src[7-k]; |
---|
513 | O[k] = src[k] - src[7-k]; |
---|
514 | } |
---|
515 | /* EE and EO */ |
---|
516 | EE[0] = E[0] + E[3]; |
---|
517 | EO[0] = E[0] - E[3]; |
---|
518 | EE[1] = E[1] + E[2]; |
---|
519 | EO[1] = E[1] - E[2]; |
---|
520 | |
---|
521 | dst[0] = (g_aiT8[0][0]*EE[0] + g_aiT8[0][1]*EE[1] + add)>>shift; |
---|
522 | dst[4*line] = (g_aiT8[4][0]*EE[0] + g_aiT8[4][1]*EE[1] + add)>>shift; |
---|
523 | dst[2*line] = (g_aiT8[2][0]*EO[0] + g_aiT8[2][1]*EO[1] + add)>>shift; |
---|
524 | dst[6*line] = (g_aiT8[6][0]*EO[0] + g_aiT8[6][1]*EO[1] + add)>>shift; |
---|
525 | |
---|
526 | dst[line] = (g_aiT8[1][0]*O[0] + g_aiT8[1][1]*O[1] + g_aiT8[1][2]*O[2] + g_aiT8[1][3]*O[3] + add)>>shift; |
---|
527 | dst[3*line] = (g_aiT8[3][0]*O[0] + g_aiT8[3][1]*O[1] + g_aiT8[3][2]*O[2] + g_aiT8[3][3]*O[3] + add)>>shift; |
---|
528 | dst[5*line] = (g_aiT8[5][0]*O[0] + g_aiT8[5][1]*O[1] + g_aiT8[5][2]*O[2] + g_aiT8[5][3]*O[3] + add)>>shift; |
---|
529 | dst[7*line] = (g_aiT8[7][0]*O[0] + g_aiT8[7][1]*O[1] + g_aiT8[7][2]*O[2] + g_aiT8[7][3]*O[3] + add)>>shift; |
---|
530 | |
---|
531 | src += 8; |
---|
532 | dst ++; |
---|
533 | } |
---|
534 | } |
---|
535 | |
---|
536 | |
---|
537 | void partialButterflyInverse8(Short *src,Short *dst,Int shift, Int line) |
---|
538 | { |
---|
539 | Int j,k; |
---|
540 | Int E[4],O[4]; |
---|
541 | Int EE[2],EO[2]; |
---|
542 | Int add = 1<<(shift-1); |
---|
543 | |
---|
544 | for (j=0; j<line; j++) |
---|
545 | { |
---|
546 | /* Utilizing symmetry properties to the maximum to minimize the number of multiplications */ |
---|
547 | for (k=0;k<4;k++) |
---|
548 | { |
---|
549 | O[k] = g_aiT8[ 1][k]*src[line] + g_aiT8[ 3][k]*src[3*line] + g_aiT8[ 5][k]*src[5*line] + g_aiT8[ 7][k]*src[7*line]; |
---|
550 | } |
---|
551 | |
---|
552 | EO[0] = g_aiT8[2][0]*src[ 2*line ] + g_aiT8[6][0]*src[ 6*line ]; |
---|
553 | EO[1] = g_aiT8[2][1]*src[ 2*line ] + g_aiT8[6][1]*src[ 6*line ]; |
---|
554 | EE[0] = g_aiT8[0][0]*src[ 0 ] + g_aiT8[4][0]*src[ 4*line ]; |
---|
555 | EE[1] = g_aiT8[0][1]*src[ 0 ] + g_aiT8[4][1]*src[ 4*line ]; |
---|
556 | |
---|
557 | /* Combining even and odd terms at each hierarchy levels to calculate the final spatial domain vector */ |
---|
558 | E[0] = EE[0] + EO[0]; |
---|
559 | E[3] = EE[0] - EO[0]; |
---|
560 | E[1] = EE[1] + EO[1]; |
---|
561 | E[2] = EE[1] - EO[1]; |
---|
562 | for (k=0;k<4;k++) |
---|
563 | { |
---|
564 | dst[ k ] = Clip3( -32768, 32767, (E[k] + O[k] + add)>>shift ); |
---|
565 | dst[ k+4 ] = Clip3( -32768, 32767, (E[3-k] - O[3-k] + add)>>shift ); |
---|
566 | } |
---|
567 | src ++; |
---|
568 | dst += 8; |
---|
569 | } |
---|
570 | } |
---|
571 | |
---|
572 | |
---|
573 | void partialButterfly16(Short *src,Short *dst,Int shift, Int line) |
---|
574 | { |
---|
575 | Int j,k; |
---|
576 | Int E[8],O[8]; |
---|
577 | Int EE[4],EO[4]; |
---|
578 | Int EEE[2],EEO[2]; |
---|
579 | Int add = 1<<(shift-1); |
---|
580 | |
---|
581 | for (j=0; j<line; j++) |
---|
582 | { |
---|
583 | /* E and O*/ |
---|
584 | for (k=0;k<8;k++) |
---|
585 | { |
---|
586 | E[k] = src[k] + src[15-k]; |
---|
587 | O[k] = src[k] - src[15-k]; |
---|
588 | } |
---|
589 | /* EE and EO */ |
---|
590 | for (k=0;k<4;k++) |
---|
591 | { |
---|
592 | EE[k] = E[k] + E[7-k]; |
---|
593 | EO[k] = E[k] - E[7-k]; |
---|
594 | } |
---|
595 | /* EEE and EEO */ |
---|
596 | EEE[0] = EE[0] + EE[3]; |
---|
597 | EEO[0] = EE[0] - EE[3]; |
---|
598 | EEE[1] = EE[1] + EE[2]; |
---|
599 | EEO[1] = EE[1] - EE[2]; |
---|
600 | |
---|
601 | dst[ 0 ] = (g_aiT16[ 0][0]*EEE[0] + g_aiT16[ 0][1]*EEE[1] + add)>>shift; |
---|
602 | dst[ 8*line ] = (g_aiT16[ 8][0]*EEE[0] + g_aiT16[ 8][1]*EEE[1] + add)>>shift; |
---|
603 | dst[ 4*line ] = (g_aiT16[ 4][0]*EEO[0] + g_aiT16[ 4][1]*EEO[1] + add)>>shift; |
---|
604 | dst[ 12*line] = (g_aiT16[12][0]*EEO[0] + g_aiT16[12][1]*EEO[1] + add)>>shift; |
---|
605 | |
---|
606 | for (k=2;k<16;k+=4) |
---|
607 | { |
---|
608 | dst[ k*line ] = (g_aiT16[k][0]*EO[0] + g_aiT16[k][1]*EO[1] + g_aiT16[k][2]*EO[2] + g_aiT16[k][3]*EO[3] + add)>>shift; |
---|
609 | } |
---|
610 | |
---|
611 | for (k=1;k<16;k+=2) |
---|
612 | { |
---|
613 | dst[ k*line ] = (g_aiT16[k][0]*O[0] + g_aiT16[k][1]*O[1] + g_aiT16[k][2]*O[2] + g_aiT16[k][3]*O[3] + |
---|
614 | g_aiT16[k][4]*O[4] + g_aiT16[k][5]*O[5] + g_aiT16[k][6]*O[6] + g_aiT16[k][7]*O[7] + add)>>shift; |
---|
615 | } |
---|
616 | |
---|
617 | src += 16; |
---|
618 | dst ++; |
---|
619 | |
---|
620 | } |
---|
621 | } |
---|
622 | |
---|
623 | |
---|
624 | void partialButterflyInverse16(Short *src,Short *dst,Int shift, Int line) |
---|
625 | { |
---|
626 | Int j,k; |
---|
627 | Int E[8],O[8]; |
---|
628 | Int EE[4],EO[4]; |
---|
629 | Int EEE[2],EEO[2]; |
---|
630 | Int add = 1<<(shift-1); |
---|
631 | |
---|
632 | for (j=0; j<line; j++) |
---|
633 | { |
---|
634 | /* Utilizing symmetry properties to the maximum to minimize the number of multiplications */ |
---|
635 | for (k=0;k<8;k++) |
---|
636 | { |
---|
637 | O[k] = g_aiT16[ 1][k]*src[ line] + g_aiT16[ 3][k]*src[ 3*line] + g_aiT16[ 5][k]*src[ 5*line] + g_aiT16[ 7][k]*src[ 7*line] + |
---|
638 | g_aiT16[ 9][k]*src[ 9*line] + g_aiT16[11][k]*src[11*line] + g_aiT16[13][k]*src[13*line] + g_aiT16[15][k]*src[15*line]; |
---|
639 | } |
---|
640 | for (k=0;k<4;k++) |
---|
641 | { |
---|
642 | EO[k] = g_aiT16[ 2][k]*src[ 2*line] + g_aiT16[ 6][k]*src[ 6*line] + g_aiT16[10][k]*src[10*line] + g_aiT16[14][k]*src[14*line]; |
---|
643 | } |
---|
644 | EEO[0] = g_aiT16[4][0]*src[ 4*line ] + g_aiT16[12][0]*src[ 12*line ]; |
---|
645 | EEE[0] = g_aiT16[0][0]*src[ 0 ] + g_aiT16[ 8][0]*src[ 8*line ]; |
---|
646 | EEO[1] = g_aiT16[4][1]*src[ 4*line ] + g_aiT16[12][1]*src[ 12*line ]; |
---|
647 | EEE[1] = g_aiT16[0][1]*src[ 0 ] + g_aiT16[ 8][1]*src[ 8*line ]; |
---|
648 | |
---|
649 | /* Combining even and odd terms at each hierarchy levels to calculate the final spatial domain vector */ |
---|
650 | for (k=0;k<2;k++) |
---|
651 | { |
---|
652 | EE[k] = EEE[k] + EEO[k]; |
---|
653 | EE[k+2] = EEE[1-k] - EEO[1-k]; |
---|
654 | } |
---|
655 | for (k=0;k<4;k++) |
---|
656 | { |
---|
657 | E[k] = EE[k] + EO[k]; |
---|
658 | E[k+4] = EE[3-k] - EO[3-k]; |
---|
659 | } |
---|
660 | for (k=0;k<8;k++) |
---|
661 | { |
---|
662 | dst[k] = Clip3( -32768, 32767, (E[k] + O[k] + add)>>shift ); |
---|
663 | dst[k+8] = Clip3( -32768, 32767, (E[7-k] - O[7-k] + add)>>shift ); |
---|
664 | } |
---|
665 | src ++; |
---|
666 | dst += 16; |
---|
667 | } |
---|
668 | } |
---|
669 | |
---|
670 | |
---|
671 | void partialButterfly32(Short *src,Short *dst,Int shift, Int line) |
---|
672 | { |
---|
673 | Int j,k; |
---|
674 | Int E[16],O[16]; |
---|
675 | Int EE[8],EO[8]; |
---|
676 | Int EEE[4],EEO[4]; |
---|
677 | Int EEEE[2],EEEO[2]; |
---|
678 | Int add = 1<<(shift-1); |
---|
679 | |
---|
680 | for (j=0; j<line; j++) |
---|
681 | { |
---|
682 | /* E and O*/ |
---|
683 | for (k=0;k<16;k++) |
---|
684 | { |
---|
685 | E[k] = src[k] + src[31-k]; |
---|
686 | O[k] = src[k] - src[31-k]; |
---|
687 | } |
---|
688 | /* EE and EO */ |
---|
689 | for (k=0;k<8;k++) |
---|
690 | { |
---|
691 | EE[k] = E[k] + E[15-k]; |
---|
692 | EO[k] = E[k] - E[15-k]; |
---|
693 | } |
---|
694 | /* EEE and EEO */ |
---|
695 | for (k=0;k<4;k++) |
---|
696 | { |
---|
697 | EEE[k] = EE[k] + EE[7-k]; |
---|
698 | EEO[k] = EE[k] - EE[7-k]; |
---|
699 | } |
---|
700 | /* EEEE and EEEO */ |
---|
701 | EEEE[0] = EEE[0] + EEE[3]; |
---|
702 | EEEO[0] = EEE[0] - EEE[3]; |
---|
703 | EEEE[1] = EEE[1] + EEE[2]; |
---|
704 | EEEO[1] = EEE[1] - EEE[2]; |
---|
705 | |
---|
706 | dst[ 0 ] = (g_aiT32[ 0][0]*EEEE[0] + g_aiT32[ 0][1]*EEEE[1] + add)>>shift; |
---|
707 | dst[ 16*line ] = (g_aiT32[16][0]*EEEE[0] + g_aiT32[16][1]*EEEE[1] + add)>>shift; |
---|
708 | dst[ 8*line ] = (g_aiT32[ 8][0]*EEEO[0] + g_aiT32[ 8][1]*EEEO[1] + add)>>shift; |
---|
709 | dst[ 24*line ] = (g_aiT32[24][0]*EEEO[0] + g_aiT32[24][1]*EEEO[1] + add)>>shift; |
---|
710 | for (k=4;k<32;k+=8) |
---|
711 | { |
---|
712 | dst[ k*line ] = (g_aiT32[k][0]*EEO[0] + g_aiT32[k][1]*EEO[1] + g_aiT32[k][2]*EEO[2] + g_aiT32[k][3]*EEO[3] + add)>>shift; |
---|
713 | } |
---|
714 | for (k=2;k<32;k+=4) |
---|
715 | { |
---|
716 | dst[ k*line ] = (g_aiT32[k][0]*EO[0] + g_aiT32[k][1]*EO[1] + g_aiT32[k][2]*EO[2] + g_aiT32[k][3]*EO[3] + |
---|
717 | g_aiT32[k][4]*EO[4] + g_aiT32[k][5]*EO[5] + g_aiT32[k][6]*EO[6] + g_aiT32[k][7]*EO[7] + add)>>shift; |
---|
718 | } |
---|
719 | for (k=1;k<32;k+=2) |
---|
720 | { |
---|
721 | dst[ k*line ] = (g_aiT32[k][ 0]*O[ 0] + g_aiT32[k][ 1]*O[ 1] + g_aiT32[k][ 2]*O[ 2] + g_aiT32[k][ 3]*O[ 3] + |
---|
722 | g_aiT32[k][ 4]*O[ 4] + g_aiT32[k][ 5]*O[ 5] + g_aiT32[k][ 6]*O[ 6] + g_aiT32[k][ 7]*O[ 7] + |
---|
723 | g_aiT32[k][ 8]*O[ 8] + g_aiT32[k][ 9]*O[ 9] + g_aiT32[k][10]*O[10] + g_aiT32[k][11]*O[11] + |
---|
724 | g_aiT32[k][12]*O[12] + g_aiT32[k][13]*O[13] + g_aiT32[k][14]*O[14] + g_aiT32[k][15]*O[15] + add)>>shift; |
---|
725 | } |
---|
726 | src += 32; |
---|
727 | dst ++; |
---|
728 | } |
---|
729 | } |
---|
730 | |
---|
731 | |
---|
732 | void partialButterflyInverse32(Short *src,Short *dst,Int shift, Int line) |
---|
733 | { |
---|
734 | Int j,k; |
---|
735 | Int E[16],O[16]; |
---|
736 | Int EE[8],EO[8]; |
---|
737 | Int EEE[4],EEO[4]; |
---|
738 | Int EEEE[2],EEEO[2]; |
---|
739 | Int add = 1<<(shift-1); |
---|
740 | |
---|
741 | for (j=0; j<line; j++) |
---|
742 | { |
---|
743 | /* Utilizing symmetry properties to the maximum to minimize the number of multiplications */ |
---|
744 | for (k=0;k<16;k++) |
---|
745 | { |
---|
746 | O[k] = g_aiT32[ 1][k]*src[ line ] + g_aiT32[ 3][k]*src[ 3*line ] + g_aiT32[ 5][k]*src[ 5*line ] + g_aiT32[ 7][k]*src[ 7*line ] + |
---|
747 | g_aiT32[ 9][k]*src[ 9*line ] + g_aiT32[11][k]*src[ 11*line ] + g_aiT32[13][k]*src[ 13*line ] + g_aiT32[15][k]*src[ 15*line ] + |
---|
748 | g_aiT32[17][k]*src[ 17*line ] + g_aiT32[19][k]*src[ 19*line ] + g_aiT32[21][k]*src[ 21*line ] + g_aiT32[23][k]*src[ 23*line ] + |
---|
749 | g_aiT32[25][k]*src[ 25*line ] + g_aiT32[27][k]*src[ 27*line ] + g_aiT32[29][k]*src[ 29*line ] + g_aiT32[31][k]*src[ 31*line ]; |
---|
750 | } |
---|
751 | for (k=0;k<8;k++) |
---|
752 | { |
---|
753 | EO[k] = g_aiT32[ 2][k]*src[ 2*line ] + g_aiT32[ 6][k]*src[ 6*line ] + g_aiT32[10][k]*src[ 10*line ] + g_aiT32[14][k]*src[ 14*line ] + |
---|
754 | g_aiT32[18][k]*src[ 18*line ] + g_aiT32[22][k]*src[ 22*line ] + g_aiT32[26][k]*src[ 26*line ] + g_aiT32[30][k]*src[ 30*line ]; |
---|
755 | } |
---|
756 | for (k=0;k<4;k++) |
---|
757 | { |
---|
758 | EEO[k] = g_aiT32[4][k]*src[ 4*line ] + g_aiT32[12][k]*src[ 12*line ] + g_aiT32[20][k]*src[ 20*line ] + g_aiT32[28][k]*src[ 28*line ]; |
---|
759 | } |
---|
760 | EEEO[0] = g_aiT32[8][0]*src[ 8*line ] + g_aiT32[24][0]*src[ 24*line ]; |
---|
761 | EEEO[1] = g_aiT32[8][1]*src[ 8*line ] + g_aiT32[24][1]*src[ 24*line ]; |
---|
762 | EEEE[0] = g_aiT32[0][0]*src[ 0 ] + g_aiT32[16][0]*src[ 16*line ]; |
---|
763 | EEEE[1] = g_aiT32[0][1]*src[ 0 ] + g_aiT32[16][1]*src[ 16*line ]; |
---|
764 | |
---|
765 | /* Combining even and odd terms at each hierarchy levels to calculate the final spatial domain vector */ |
---|
766 | EEE[0] = EEEE[0] + EEEO[0]; |
---|
767 | EEE[3] = EEEE[0] - EEEO[0]; |
---|
768 | EEE[1] = EEEE[1] + EEEO[1]; |
---|
769 | EEE[2] = EEEE[1] - EEEO[1]; |
---|
770 | for (k=0;k<4;k++) |
---|
771 | { |
---|
772 | EE[k] = EEE[k] + EEO[k]; |
---|
773 | EE[k+4] = EEE[3-k] - EEO[3-k]; |
---|
774 | } |
---|
775 | for (k=0;k<8;k++) |
---|
776 | { |
---|
777 | E[k] = EE[k] + EO[k]; |
---|
778 | E[k+8] = EE[7-k] - EO[7-k]; |
---|
779 | } |
---|
780 | for (k=0;k<16;k++) |
---|
781 | { |
---|
782 | dst[k] = Clip3( -32768, 32767, (E[k] + O[k] + add)>>shift ); |
---|
783 | dst[k+16] = Clip3( -32768, 32767, (E[15-k] - O[15-k] + add)>>shift ); |
---|
784 | } |
---|
785 | src ++; |
---|
786 | dst += 32; |
---|
787 | } |
---|
788 | } |
---|
789 | |
---|
790 | /** MxN forward transform (2D) |
---|
791 | * \param block input data (residual) |
---|
792 | * \param coeff output data (transform coefficients) |
---|
793 | * \param iWidth input data (width of transform) |
---|
794 | * \param iHeight input data (height of transform) |
---|
795 | */ |
---|
796 | void xTrMxN(Int bitDepth, Short *block,Short *coeff, Int iWidth, Int iHeight, UInt uiMode) |
---|
797 | { |
---|
798 | Int shift_1st = g_aucConvertToBit[iWidth] + 1 + bitDepth-8; // log2(iWidth) - 1 + g_bitDepth - 8 |
---|
799 | Int shift_2nd = g_aucConvertToBit[iHeight] + 8; // log2(iHeight) + 6 |
---|
800 | |
---|
801 | Short tmp[ 64 * 64 ]; |
---|
802 | |
---|
803 | if( iWidth == 4 && iHeight == 4) |
---|
804 | { |
---|
805 | if (uiMode != REG_DCT) |
---|
806 | { |
---|
807 | fastForwardDst(block,tmp,shift_1st); // Forward DST BY FAST ALGORITHM, block input, tmp output |
---|
808 | fastForwardDst(tmp,coeff,shift_2nd); // Forward DST BY FAST ALGORITHM, tmp input, coeff output |
---|
809 | } |
---|
810 | else |
---|
811 | { |
---|
812 | partialButterfly4(block, tmp, shift_1st, iHeight); |
---|
813 | partialButterfly4(tmp, coeff, shift_2nd, iWidth); |
---|
814 | } |
---|
815 | |
---|
816 | } |
---|
817 | else if( iWidth == 8 && iHeight == 8) |
---|
818 | { |
---|
819 | partialButterfly8( block, tmp, shift_1st, iHeight ); |
---|
820 | partialButterfly8( tmp, coeff, shift_2nd, iWidth ); |
---|
821 | } |
---|
822 | else if( iWidth == 16 && iHeight == 16) |
---|
823 | { |
---|
824 | partialButterfly16( block, tmp, shift_1st, iHeight ); |
---|
825 | partialButterfly16( tmp, coeff, shift_2nd, iWidth ); |
---|
826 | } |
---|
827 | else if( iWidth == 32 && iHeight == 32) |
---|
828 | { |
---|
829 | partialButterfly32( block, tmp, shift_1st, iHeight ); |
---|
830 | partialButterfly32( tmp, coeff, shift_2nd, iWidth ); |
---|
831 | } |
---|
832 | } |
---|
833 | /** MxN inverse transform (2D) |
---|
834 | * \param coeff input data (transform coefficients) |
---|
835 | * \param block output data (residual) |
---|
836 | * \param iWidth input data (width of transform) |
---|
837 | * \param iHeight input data (height of transform) |
---|
838 | */ |
---|
839 | void xITrMxN(Int bitDepth, Short *coeff,Short *block, Int iWidth, Int iHeight, UInt uiMode) |
---|
840 | { |
---|
841 | Int shift_1st = SHIFT_INV_1ST; |
---|
842 | Int shift_2nd = SHIFT_INV_2ND - (bitDepth-8); |
---|
843 | |
---|
844 | Short tmp[ 64*64]; |
---|
845 | if( iWidth == 4 && iHeight == 4) |
---|
846 | { |
---|
847 | if (uiMode != REG_DCT) |
---|
848 | { |
---|
849 | fastInverseDst(coeff,tmp,shift_1st); // Inverse DST by FAST Algorithm, coeff input, tmp output |
---|
850 | fastInverseDst(tmp,block,shift_2nd); // Inverse DST by FAST Algorithm, tmp input, coeff output |
---|
851 | } |
---|
852 | else |
---|
853 | { |
---|
854 | partialButterflyInverse4(coeff,tmp,shift_1st,iWidth); |
---|
855 | partialButterflyInverse4(tmp,block,shift_2nd,iHeight); |
---|
856 | } |
---|
857 | } |
---|
858 | else if( iWidth == 8 && iHeight == 8) |
---|
859 | { |
---|
860 | partialButterflyInverse8(coeff,tmp,shift_1st,iWidth); |
---|
861 | partialButterflyInverse8(tmp,block,shift_2nd,iHeight); |
---|
862 | } |
---|
863 | else if( iWidth == 16 && iHeight == 16) |
---|
864 | { |
---|
865 | partialButterflyInverse16(coeff,tmp,shift_1st,iWidth); |
---|
866 | partialButterflyInverse16(tmp,block,shift_2nd,iHeight); |
---|
867 | } |
---|
868 | else if( iWidth == 32 && iHeight == 32) |
---|
869 | { |
---|
870 | partialButterflyInverse32(coeff,tmp,shift_1st,iWidth); |
---|
871 | partialButterflyInverse32(tmp,block,shift_2nd,iHeight); |
---|
872 | } |
---|
873 | } |
---|
874 | |
---|
875 | #endif //MATRIX_MULT |
---|
876 | |
---|
877 | // To minimize the distortion only. No rate is considered. |
---|
878 | Void TComTrQuant::signBitHidingHDQ( TCoeff* pQCoef, TCoeff* pCoef, UInt const *scan, Int* deltaU, Int width, Int height ) |
---|
879 | { |
---|
880 | Int lastCG = -1; |
---|
881 | Int absSum = 0 ; |
---|
882 | Int n ; |
---|
883 | |
---|
884 | for( Int subSet = (width*height-1) >> LOG2_SCAN_SET_SIZE; subSet >= 0; subSet-- ) |
---|
885 | { |
---|
886 | Int subPos = subSet << LOG2_SCAN_SET_SIZE; |
---|
887 | Int firstNZPosInCG=SCAN_SET_SIZE , lastNZPosInCG=-1 ; |
---|
888 | absSum = 0 ; |
---|
889 | |
---|
890 | for(n = SCAN_SET_SIZE-1; n >= 0; --n ) |
---|
891 | { |
---|
892 | if( pQCoef[ scan[ n + subPos ]] ) |
---|
893 | { |
---|
894 | lastNZPosInCG = n; |
---|
895 | break; |
---|
896 | } |
---|
897 | } |
---|
898 | |
---|
899 | for(n = 0; n <SCAN_SET_SIZE; n++ ) |
---|
900 | { |
---|
901 | if( pQCoef[ scan[ n + subPos ]] ) |
---|
902 | { |
---|
903 | firstNZPosInCG = n; |
---|
904 | break; |
---|
905 | } |
---|
906 | } |
---|
907 | |
---|
908 | for(n = firstNZPosInCG; n <=lastNZPosInCG; n++ ) |
---|
909 | { |
---|
910 | absSum += pQCoef[ scan[ n + subPos ]]; |
---|
911 | } |
---|
912 | |
---|
913 | if(lastNZPosInCG>=0 && lastCG==-1) |
---|
914 | { |
---|
915 | lastCG = 1 ; |
---|
916 | } |
---|
917 | |
---|
918 | if( lastNZPosInCG-firstNZPosInCG>=SBH_THRESHOLD ) |
---|
919 | { |
---|
920 | UInt signbit = (pQCoef[scan[subPos+firstNZPosInCG]]>0?0:1) ; |
---|
921 | if( signbit!=(absSum&0x1) ) //compare signbit with sum_parity |
---|
922 | { |
---|
923 | Int minCostInc = MAX_INT, minPos =-1, finalChange=0, curCost=MAX_INT, curChange=0; |
---|
924 | |
---|
925 | for( n = (lastCG==1?lastNZPosInCG:SCAN_SET_SIZE-1) ; n >= 0; --n ) |
---|
926 | { |
---|
927 | UInt blkPos = scan[ n+subPos ]; |
---|
928 | if(pQCoef[ blkPos ] != 0 ) |
---|
929 | { |
---|
930 | if(deltaU[blkPos]>0) |
---|
931 | { |
---|
932 | curCost = - deltaU[blkPos]; |
---|
933 | curChange=1 ; |
---|
934 | } |
---|
935 | else |
---|
936 | { |
---|
937 | //curChange =-1; |
---|
938 | if(n==firstNZPosInCG && abs(pQCoef[blkPos])==1) |
---|
939 | { |
---|
940 | curCost=MAX_INT ; |
---|
941 | } |
---|
942 | else |
---|
943 | { |
---|
944 | curCost = deltaU[blkPos]; |
---|
945 | curChange =-1; |
---|
946 | } |
---|
947 | } |
---|
948 | } |
---|
949 | else |
---|
950 | { |
---|
951 | if(n<firstNZPosInCG) |
---|
952 | { |
---|
953 | UInt thisSignBit = (pCoef[blkPos]>=0?0:1); |
---|
954 | if(thisSignBit != signbit ) |
---|
955 | { |
---|
956 | curCost = MAX_INT; |
---|
957 | } |
---|
958 | else |
---|
959 | { |
---|
960 | curCost = - (deltaU[blkPos]) ; |
---|
961 | curChange = 1 ; |
---|
962 | } |
---|
963 | } |
---|
964 | else |
---|
965 | { |
---|
966 | curCost = - (deltaU[blkPos]) ; |
---|
967 | curChange = 1 ; |
---|
968 | } |
---|
969 | } |
---|
970 | |
---|
971 | if( curCost<minCostInc) |
---|
972 | { |
---|
973 | minCostInc = curCost ; |
---|
974 | finalChange = curChange ; |
---|
975 | minPos = blkPos ; |
---|
976 | } |
---|
977 | } //CG loop |
---|
978 | |
---|
979 | if(pQCoef[minPos] == 32767 || pQCoef[minPos] == -32768) |
---|
980 | { |
---|
981 | finalChange = -1; |
---|
982 | } |
---|
983 | |
---|
984 | if(pCoef[minPos]>=0) |
---|
985 | { |
---|
986 | pQCoef[minPos] += finalChange ; |
---|
987 | } |
---|
988 | else |
---|
989 | { |
---|
990 | pQCoef[minPos] -= finalChange ; |
---|
991 | } |
---|
992 | } // Hide |
---|
993 | } |
---|
994 | if(lastCG==1) |
---|
995 | { |
---|
996 | lastCG=0 ; |
---|
997 | } |
---|
998 | } // TU loop |
---|
999 | |
---|
1000 | return; |
---|
1001 | } |
---|
1002 | |
---|
1003 | Void TComTrQuant::xQuant( TComDataCU* pcCU, |
---|
1004 | Int* pSrc, |
---|
1005 | TCoeff* pDes, |
---|
1006 | #if ADAPTIVE_QP_SELECTION |
---|
1007 | Int*& pArlDes, |
---|
1008 | #endif |
---|
1009 | Int iWidth, |
---|
1010 | Int iHeight, |
---|
1011 | UInt& uiAcSum, |
---|
1012 | TextType eTType, |
---|
1013 | UInt uiAbsPartIdx ) |
---|
1014 | { |
---|
1015 | Int* piCoef = pSrc; |
---|
1016 | TCoeff* piQCoef = pDes; |
---|
1017 | #if ADAPTIVE_QP_SELECTION |
---|
1018 | Int* piArlCCoef = pArlDes; |
---|
1019 | #endif |
---|
1020 | Int iAdd = 0; |
---|
1021 | |
---|
1022 | Bool useRDOQ = pcCU->getTransformSkip(uiAbsPartIdx,eTType) ? m_useRDOQTS:m_useRDOQ; |
---|
1023 | if ( useRDOQ && (eTType == TEXT_LUMA || RDOQ_CHROMA)) |
---|
1024 | { |
---|
1025 | #if ADAPTIVE_QP_SELECTION |
---|
1026 | xRateDistOptQuant( pcCU, piCoef, pDes, pArlDes, iWidth, iHeight, uiAcSum, eTType, uiAbsPartIdx ); |
---|
1027 | #else |
---|
1028 | xRateDistOptQuant( pcCU, piCoef, pDes, iWidth, iHeight, uiAcSum, eTType, uiAbsPartIdx ); |
---|
1029 | #endif |
---|
1030 | } |
---|
1031 | else |
---|
1032 | { |
---|
1033 | const UInt log2BlockSize = g_aucConvertToBit[ iWidth ] + 2; |
---|
1034 | |
---|
1035 | UInt scanIdx = pcCU->getCoefScanIdx(uiAbsPartIdx, iWidth, eTType==TEXT_LUMA, pcCU->isIntra(uiAbsPartIdx)); |
---|
1036 | const UInt *scan = g_auiSigLastScan[ scanIdx ][ log2BlockSize - 1 ]; |
---|
1037 | |
---|
1038 | Int deltaU[32*32] ; |
---|
1039 | |
---|
1040 | #if ADAPTIVE_QP_SELECTION |
---|
1041 | QpParam cQpBase; |
---|
1042 | Int iQpBase = pcCU->getSlice()->getSliceQpBase(); |
---|
1043 | |
---|
1044 | Int qpScaled; |
---|
1045 | #if REPN_FORMAT_IN_VPS |
---|
1046 | Int qpBDOffset = (eTType == TEXT_LUMA)? pcCU->getSlice()->getQpBDOffsetY() : pcCU->getSlice()->getQpBDOffsetC(); |
---|
1047 | #else |
---|
1048 | Int qpBDOffset = (eTType == TEXT_LUMA)? pcCU->getSlice()->getSPS()->getQpBDOffsetY() : pcCU->getSlice()->getSPS()->getQpBDOffsetC(); |
---|
1049 | #endif |
---|
1050 | |
---|
1051 | if(eTType == TEXT_LUMA) |
---|
1052 | { |
---|
1053 | qpScaled = iQpBase + qpBDOffset; |
---|
1054 | } |
---|
1055 | else |
---|
1056 | { |
---|
1057 | Int chromaQPOffset; |
---|
1058 | if(eTType == TEXT_CHROMA_U) |
---|
1059 | { |
---|
1060 | chromaQPOffset = pcCU->getSlice()->getPPS()->getChromaCbQpOffset() + pcCU->getSlice()->getSliceQpDeltaCb(); |
---|
1061 | } |
---|
1062 | else |
---|
1063 | { |
---|
1064 | chromaQPOffset = pcCU->getSlice()->getPPS()->getChromaCrQpOffset() + pcCU->getSlice()->getSliceQpDeltaCr(); |
---|
1065 | } |
---|
1066 | iQpBase = iQpBase + chromaQPOffset; |
---|
1067 | |
---|
1068 | qpScaled = Clip3( -qpBDOffset, 57, iQpBase); |
---|
1069 | |
---|
1070 | if(qpScaled < 0) |
---|
1071 | { |
---|
1072 | qpScaled = qpScaled + qpBDOffset; |
---|
1073 | } |
---|
1074 | else |
---|
1075 | { |
---|
1076 | qpScaled = g_aucChromaScale[ qpScaled ] + qpBDOffset; |
---|
1077 | } |
---|
1078 | } |
---|
1079 | cQpBase.setQpParam(qpScaled); |
---|
1080 | #endif |
---|
1081 | |
---|
1082 | UInt uiLog2TrSize = g_aucConvertToBit[ iWidth ] + 2; |
---|
1083 | Int scalingListType = (pcCU->isIntra(uiAbsPartIdx) ? 0 : 3) + g_eTTable[(Int)eTType]; |
---|
1084 | assert(scalingListType < 6); |
---|
1085 | Int *piQuantCoeff = 0; |
---|
1086 | piQuantCoeff = getQuantCoeff(scalingListType,m_cQP.m_iRem,uiLog2TrSize-2); |
---|
1087 | |
---|
1088 | UInt uiBitDepth = eTType == TEXT_LUMA ? g_bitDepthY : g_bitDepthC; |
---|
1089 | Int iTransformShift = MAX_TR_DYNAMIC_RANGE - uiBitDepth - uiLog2TrSize; // Represents scaling through forward transform |
---|
1090 | |
---|
1091 | #if ADAPTIVE_QP_SELECTION |
---|
1092 | Int iQBits = QUANT_SHIFT + cQpBase.m_iPer + iTransformShift; |
---|
1093 | iAdd = (pcCU->getSlice()->getSliceType()==I_SLICE ? 171 : 85) << (iQBits-9); |
---|
1094 | Int iQBitsC = QUANT_SHIFT + cQpBase.m_iPer + iTransformShift - ARL_C_PRECISION; |
---|
1095 | Int iAddC = 1 << (iQBitsC-1); |
---|
1096 | #else |
---|
1097 | Int iQBits = QUANT_SHIFT + m_cQP.m_iPer + iTransformShift; // Right shift of non-RDOQ quantizer; level = (coeff*uiQ + offset)>>q_bits |
---|
1098 | iAdd = (pcCU->getSlice()->getSliceType()==I_SLICE ? 171 : 85) << (iQBits-9); |
---|
1099 | #endif |
---|
1100 | |
---|
1101 | Int qBits8 = iQBits-8; |
---|
1102 | for( Int n = 0; n < iWidth*iHeight; n++ ) |
---|
1103 | { |
---|
1104 | Int iLevel; |
---|
1105 | Int iSign; |
---|
1106 | UInt uiBlockPos = n; |
---|
1107 | iLevel = piCoef[uiBlockPos]; |
---|
1108 | iSign = (iLevel < 0 ? -1: 1); |
---|
1109 | |
---|
1110 | #if ADAPTIVE_QP_SELECTION |
---|
1111 | Int64 tmpLevel = (Int64)abs(iLevel) * piQuantCoeff[uiBlockPos]; |
---|
1112 | if( m_bUseAdaptQpSelect ) |
---|
1113 | { |
---|
1114 | piArlCCoef[uiBlockPos] = (Int)((tmpLevel + iAddC ) >> iQBitsC); |
---|
1115 | } |
---|
1116 | iLevel = (Int)((tmpLevel + iAdd ) >> iQBits); |
---|
1117 | deltaU[uiBlockPos] = (Int)((tmpLevel - (iLevel<<iQBits) )>> qBits8); |
---|
1118 | #else |
---|
1119 | iLevel = ((Int64)abs(iLevel) * piQuantCoeff[uiBlockPos] + iAdd ) >> iQBits; |
---|
1120 | deltaU[uiBlockPos] = (Int)( ((Int64)abs(piCoef[uiBlockPos]) * piQuantCoeff[uiBlockPos] - (iLevel<<iQBits) )>> qBits8 ); |
---|
1121 | #endif |
---|
1122 | uiAcSum += iLevel; |
---|
1123 | iLevel *= iSign; |
---|
1124 | piQCoef[uiBlockPos] = Clip3( -32768, 32767, iLevel ); |
---|
1125 | } // for n |
---|
1126 | if( pcCU->getSlice()->getPPS()->getSignHideFlag() ) |
---|
1127 | { |
---|
1128 | if(uiAcSum>=2) |
---|
1129 | { |
---|
1130 | signBitHidingHDQ( piQCoef, piCoef, scan, deltaU, iWidth, iHeight ) ; |
---|
1131 | } |
---|
1132 | } |
---|
1133 | } //if RDOQ |
---|
1134 | //return; |
---|
1135 | |
---|
1136 | } |
---|
1137 | |
---|
1138 | Void TComTrQuant::xDeQuant(Int bitDepth, const TCoeff* pSrc, Int* pDes, Int iWidth, Int iHeight, Int scalingListType ) |
---|
1139 | { |
---|
1140 | |
---|
1141 | const TCoeff* piQCoef = pSrc; |
---|
1142 | Int* piCoef = pDes; |
---|
1143 | |
---|
1144 | if ( iWidth > (Int)m_uiMaxTrSize ) |
---|
1145 | { |
---|
1146 | iWidth = m_uiMaxTrSize; |
---|
1147 | iHeight = m_uiMaxTrSize; |
---|
1148 | } |
---|
1149 | |
---|
1150 | Int iShift,iAdd,iCoeffQ; |
---|
1151 | UInt uiLog2TrSize = g_aucConvertToBit[ iWidth ] + 2; |
---|
1152 | |
---|
1153 | Int iTransformShift = MAX_TR_DYNAMIC_RANGE - bitDepth - uiLog2TrSize; |
---|
1154 | |
---|
1155 | iShift = QUANT_IQUANT_SHIFT - QUANT_SHIFT - iTransformShift; |
---|
1156 | |
---|
1157 | TCoeff clipQCoef; |
---|
1158 | |
---|
1159 | if(getUseScalingList()) |
---|
1160 | { |
---|
1161 | iShift += 4; |
---|
1162 | Int *piDequantCoef = getDequantCoeff(scalingListType,m_cQP.m_iRem,uiLog2TrSize-2); |
---|
1163 | |
---|
1164 | if(iShift > m_cQP.m_iPer) |
---|
1165 | { |
---|
1166 | iAdd = 1 << (iShift - m_cQP.m_iPer - 1); |
---|
1167 | |
---|
1168 | for( Int n = 0; n < iWidth*iHeight; n++ ) |
---|
1169 | { |
---|
1170 | clipQCoef = Clip3( -32768, 32767, piQCoef[n] ); |
---|
1171 | iCoeffQ = ((clipQCoef * piDequantCoef[n]) + iAdd ) >> (iShift - m_cQP.m_iPer); |
---|
1172 | piCoef[n] = Clip3(-32768,32767,iCoeffQ); |
---|
1173 | } |
---|
1174 | } |
---|
1175 | else |
---|
1176 | { |
---|
1177 | for( Int n = 0; n < iWidth*iHeight; n++ ) |
---|
1178 | { |
---|
1179 | clipQCoef = Clip3( -32768, 32767, piQCoef[n] ); |
---|
1180 | iCoeffQ = Clip3( -32768, 32767, clipQCoef * piDequantCoef[n] ); // Clip to avoid possible overflow in following shift left operation |
---|
1181 | piCoef[n] = Clip3( -32768, 32767, iCoeffQ << ( m_cQP.m_iPer - iShift ) ); |
---|
1182 | } |
---|
1183 | } |
---|
1184 | } |
---|
1185 | else |
---|
1186 | { |
---|
1187 | iAdd = 1 << (iShift-1); |
---|
1188 | Int scale = g_invQuantScales[m_cQP.m_iRem] << m_cQP.m_iPer; |
---|
1189 | |
---|
1190 | for( Int n = 0; n < iWidth*iHeight; n++ ) |
---|
1191 | { |
---|
1192 | clipQCoef = Clip3( -32768, 32767, piQCoef[n] ); |
---|
1193 | iCoeffQ = ( clipQCoef * scale + iAdd ) >> iShift; |
---|
1194 | piCoef[n] = Clip3(-32768,32767,iCoeffQ); |
---|
1195 | } |
---|
1196 | } |
---|
1197 | } |
---|
1198 | |
---|
1199 | Void TComTrQuant::init( UInt uiMaxTrSize, |
---|
1200 | Bool bUseRDOQ, |
---|
1201 | Bool bUseRDOQTS, |
---|
1202 | Bool bEnc, Bool useTransformSkipFast |
---|
1203 | #if ADAPTIVE_QP_SELECTION |
---|
1204 | , Bool bUseAdaptQpSelect |
---|
1205 | #endif |
---|
1206 | ) |
---|
1207 | { |
---|
1208 | m_uiMaxTrSize = uiMaxTrSize; |
---|
1209 | m_bEnc = bEnc; |
---|
1210 | m_useRDOQ = bUseRDOQ; |
---|
1211 | m_useRDOQTS = bUseRDOQTS; |
---|
1212 | #if ADAPTIVE_QP_SELECTION |
---|
1213 | m_bUseAdaptQpSelect = bUseAdaptQpSelect; |
---|
1214 | #endif |
---|
1215 | m_useTransformSkipFast = useTransformSkipFast; |
---|
1216 | } |
---|
1217 | |
---|
1218 | Void TComTrQuant::transformNxN( TComDataCU* pcCU, |
---|
1219 | Pel* pcResidual, |
---|
1220 | UInt uiStride, |
---|
1221 | TCoeff* rpcCoeff, |
---|
1222 | #if ADAPTIVE_QP_SELECTION |
---|
1223 | Int*& rpcArlCoeff, |
---|
1224 | #endif |
---|
1225 | UInt uiWidth, |
---|
1226 | UInt uiHeight, |
---|
1227 | UInt& uiAbsSum, |
---|
1228 | TextType eTType, |
---|
1229 | UInt uiAbsPartIdx, |
---|
1230 | Bool useTransformSkip |
---|
1231 | ) |
---|
1232 | { |
---|
1233 | if (pcCU->getCUTransquantBypass(uiAbsPartIdx)) |
---|
1234 | { |
---|
1235 | uiAbsSum=0; |
---|
1236 | for (UInt k = 0; k<uiHeight; k++) |
---|
1237 | { |
---|
1238 | for (UInt j = 0; j<uiWidth; j++) |
---|
1239 | { |
---|
1240 | rpcCoeff[k*uiWidth+j]= pcResidual[k*uiStride+j]; |
---|
1241 | uiAbsSum += abs(pcResidual[k*uiStride+j]); |
---|
1242 | } |
---|
1243 | } |
---|
1244 | return; |
---|
1245 | } |
---|
1246 | UInt uiMode; //luma intra pred |
---|
1247 | if(eTType == TEXT_LUMA && pcCU->getPredictionMode(uiAbsPartIdx) == MODE_INTRA ) |
---|
1248 | { |
---|
1249 | uiMode = pcCU->getLumaIntraDir( uiAbsPartIdx ); |
---|
1250 | } |
---|
1251 | else |
---|
1252 | { |
---|
1253 | uiMode = REG_DCT; |
---|
1254 | } |
---|
1255 | |
---|
1256 | uiAbsSum = 0; |
---|
1257 | assert( (pcCU->getSlice()->getSPS()->getMaxTrSize() >= uiWidth) ); |
---|
1258 | Int bitDepth = eTType == TEXT_LUMA ? g_bitDepthY : g_bitDepthC; |
---|
1259 | if(useTransformSkip) |
---|
1260 | { |
---|
1261 | xTransformSkip(bitDepth, pcResidual, uiStride, m_plTempCoeff, uiWidth, uiHeight ); |
---|
1262 | } |
---|
1263 | else |
---|
1264 | { |
---|
1265 | xT(bitDepth, uiMode, pcResidual, uiStride, m_plTempCoeff, uiWidth, uiHeight ); |
---|
1266 | } |
---|
1267 | xQuant( pcCU, m_plTempCoeff, rpcCoeff, |
---|
1268 | #if ADAPTIVE_QP_SELECTION |
---|
1269 | rpcArlCoeff, |
---|
1270 | #endif |
---|
1271 | uiWidth, uiHeight, uiAbsSum, eTType, uiAbsPartIdx ); |
---|
1272 | } |
---|
1273 | |
---|
1274 | Void TComTrQuant::invtransformNxN( Bool transQuantBypass, TextType eText, UInt uiMode,Pel* rpcResidual, UInt uiStride, TCoeff* pcCoeff, UInt uiWidth, UInt uiHeight, Int scalingListType, Bool useTransformSkip ) |
---|
1275 | { |
---|
1276 | if(transQuantBypass) |
---|
1277 | { |
---|
1278 | for (UInt k = 0; k<uiHeight; k++) |
---|
1279 | { |
---|
1280 | for (UInt j = 0; j<uiWidth; j++) |
---|
1281 | { |
---|
1282 | rpcResidual[k*uiStride+j] = pcCoeff[k*uiWidth+j]; |
---|
1283 | } |
---|
1284 | } |
---|
1285 | return; |
---|
1286 | } |
---|
1287 | Int bitDepth = eText == TEXT_LUMA ? g_bitDepthY : g_bitDepthC; |
---|
1288 | xDeQuant(bitDepth, pcCoeff, m_plTempCoeff, uiWidth, uiHeight, scalingListType); |
---|
1289 | if(useTransformSkip == true) |
---|
1290 | { |
---|
1291 | xITransformSkip(bitDepth, m_plTempCoeff, rpcResidual, uiStride, uiWidth, uiHeight ); |
---|
1292 | } |
---|
1293 | else |
---|
1294 | { |
---|
1295 | xIT(bitDepth, uiMode, m_plTempCoeff, rpcResidual, uiStride, uiWidth, uiHeight ); |
---|
1296 | } |
---|
1297 | } |
---|
1298 | |
---|
1299 | Void TComTrQuant::invRecurTransformNxN( TComDataCU* pcCU, UInt uiAbsPartIdx, TextType eTxt, Pel* rpcResidual, UInt uiAddr, UInt uiStride, UInt uiWidth, UInt uiHeight, UInt uiMaxTrMode, UInt uiTrMode, TCoeff* rpcCoeff ) |
---|
1300 | { |
---|
1301 | if( !pcCU->getCbf(uiAbsPartIdx, eTxt, uiTrMode) ) |
---|
1302 | { |
---|
1303 | return; |
---|
1304 | } |
---|
1305 | const UInt stopTrMode = pcCU->getTransformIdx( uiAbsPartIdx ); |
---|
1306 | |
---|
1307 | if( uiTrMode == stopTrMode ) |
---|
1308 | { |
---|
1309 | UInt uiDepth = pcCU->getDepth( uiAbsPartIdx ) + uiTrMode; |
---|
1310 | UInt uiLog2TrSize = g_aucConvertToBit[ pcCU->getSlice()->getSPS()->getMaxCUWidth() >> uiDepth ] + 2; |
---|
1311 | if( eTxt != TEXT_LUMA && uiLog2TrSize == 2 ) |
---|
1312 | { |
---|
1313 | UInt uiQPDiv = pcCU->getPic()->getNumPartInCU() >> ( ( uiDepth - 1 ) << 1 ); |
---|
1314 | if( ( uiAbsPartIdx % uiQPDiv ) != 0 ) |
---|
1315 | { |
---|
1316 | return; |
---|
1317 | } |
---|
1318 | uiWidth <<= 1; |
---|
1319 | uiHeight <<= 1; |
---|
1320 | } |
---|
1321 | Pel* pResi = rpcResidual + uiAddr; |
---|
1322 | Int scalingListType = (pcCU->isIntra(uiAbsPartIdx) ? 0 : 3) + g_eTTable[(Int)eTxt]; |
---|
1323 | assert(scalingListType < 6); |
---|
1324 | invtransformNxN( pcCU->getCUTransquantBypass(uiAbsPartIdx), eTxt, REG_DCT, pResi, uiStride, rpcCoeff, uiWidth, uiHeight, scalingListType, pcCU->getTransformSkip(uiAbsPartIdx, eTxt) ); |
---|
1325 | } |
---|
1326 | else |
---|
1327 | { |
---|
1328 | uiTrMode++; |
---|
1329 | uiWidth >>= 1; |
---|
1330 | uiHeight >>= 1; |
---|
1331 | Int trWidth = uiWidth, trHeight = uiHeight; |
---|
1332 | UInt uiAddrOffset = trHeight * uiStride; |
---|
1333 | UInt uiCoefOffset = trWidth * trHeight; |
---|
1334 | UInt uiPartOffset = pcCU->getTotalNumPart() >> ( uiTrMode << 1 ); |
---|
1335 | { |
---|
1336 | invRecurTransformNxN( pcCU, uiAbsPartIdx, eTxt, rpcResidual, uiAddr , uiStride, uiWidth, uiHeight, uiMaxTrMode, uiTrMode, rpcCoeff ); rpcCoeff += uiCoefOffset; uiAbsPartIdx += uiPartOffset; |
---|
1337 | invRecurTransformNxN( pcCU, uiAbsPartIdx, eTxt, rpcResidual, uiAddr + trWidth , uiStride, uiWidth, uiHeight, uiMaxTrMode, uiTrMode, rpcCoeff ); rpcCoeff += uiCoefOffset; uiAbsPartIdx += uiPartOffset; |
---|
1338 | invRecurTransformNxN( pcCU, uiAbsPartIdx, eTxt, rpcResidual, uiAddr + uiAddrOffset , uiStride, uiWidth, uiHeight, uiMaxTrMode, uiTrMode, rpcCoeff ); rpcCoeff += uiCoefOffset; uiAbsPartIdx += uiPartOffset; |
---|
1339 | invRecurTransformNxN( pcCU, uiAbsPartIdx, eTxt, rpcResidual, uiAddr + uiAddrOffset + trWidth, uiStride, uiWidth, uiHeight, uiMaxTrMode, uiTrMode, rpcCoeff ); |
---|
1340 | } |
---|
1341 | } |
---|
1342 | } |
---|
1343 | |
---|
1344 | // ------------------------------------------------------------------------------------------------ |
---|
1345 | // Logical transform |
---|
1346 | // ------------------------------------------------------------------------------------------------ |
---|
1347 | |
---|
1348 | /** Wrapper function between HM interface and core NxN forward transform (2D) |
---|
1349 | * \param piBlkResi input data (residual) |
---|
1350 | * \param psCoeff output data (transform coefficients) |
---|
1351 | * \param uiStride stride of input residual data |
---|
1352 | * \param iSize transform size (iSize x iSize) |
---|
1353 | * \param uiMode is Intra Prediction mode used in Mode-Dependent DCT/DST only |
---|
1354 | */ |
---|
1355 | Void TComTrQuant::xT(Int bitDepth, UInt uiMode, Pel* piBlkResi, UInt uiStride, Int* psCoeff, Int iWidth, Int iHeight ) |
---|
1356 | { |
---|
1357 | #if MATRIX_MULT |
---|
1358 | Int iSize = iWidth; |
---|
1359 | xTr(bitDepth, piBlkResi,psCoeff,uiStride,(UInt)iSize,uiMode); |
---|
1360 | #else |
---|
1361 | Int j; |
---|
1362 | Short block[ 32 * 32 ]; |
---|
1363 | Short coeff[ 32 * 32 ]; |
---|
1364 | for (j = 0; j < iHeight; j++) |
---|
1365 | { |
---|
1366 | memcpy( block + j * iWidth, piBlkResi + j * uiStride, iWidth * sizeof( Short ) ); |
---|
1367 | } |
---|
1368 | xTrMxN(bitDepth, block, coeff, iWidth, iHeight, uiMode ); |
---|
1369 | for ( j = 0; j < iHeight * iWidth; j++ ) |
---|
1370 | { |
---|
1371 | psCoeff[ j ] = coeff[ j ]; |
---|
1372 | } |
---|
1373 | #endif |
---|
1374 | } |
---|
1375 | |
---|
1376 | |
---|
1377 | /** Wrapper function between HM interface and core NxN inverse transform (2D) |
---|
1378 | * \param plCoef input data (transform coefficients) |
---|
1379 | * \param pResidual output data (residual) |
---|
1380 | * \param uiStride stride of input residual data |
---|
1381 | * \param iSize transform size (iSize x iSize) |
---|
1382 | * \param uiMode is Intra Prediction mode used in Mode-Dependent DCT/DST only |
---|
1383 | */ |
---|
1384 | Void TComTrQuant::xIT(Int bitDepth, UInt uiMode, Int* plCoef, Pel* pResidual, UInt uiStride, Int iWidth, Int iHeight ) |
---|
1385 | { |
---|
1386 | #if MATRIX_MULT |
---|
1387 | Int iSize = iWidth; |
---|
1388 | xITr(bitDepth, plCoef,pResidual,uiStride,(UInt)iSize,uiMode); |
---|
1389 | #else |
---|
1390 | Int j; |
---|
1391 | { |
---|
1392 | Short block[ 32 * 32 ]; |
---|
1393 | Short coeff[ 32 * 32 ]; |
---|
1394 | for ( j = 0; j < iHeight * iWidth; j++ ) |
---|
1395 | { |
---|
1396 | coeff[j] = (Short)plCoef[j]; |
---|
1397 | } |
---|
1398 | xITrMxN(bitDepth, coeff, block, iWidth, iHeight, uiMode ); |
---|
1399 | { |
---|
1400 | for ( j = 0; j < iHeight; j++ ) |
---|
1401 | { |
---|
1402 | memcpy( pResidual + j * uiStride, block + j * iWidth, iWidth * sizeof(Short) ); |
---|
1403 | } |
---|
1404 | } |
---|
1405 | return ; |
---|
1406 | } |
---|
1407 | #endif |
---|
1408 | } |
---|
1409 | |
---|
1410 | /** Wrapper function between HM interface and core 4x4 transform skipping |
---|
1411 | * \param piBlkResi input data (residual) |
---|
1412 | * \param psCoeff output data (transform coefficients) |
---|
1413 | * \param uiStride stride of input residual data |
---|
1414 | * \param iSize transform size (iSize x iSize) |
---|
1415 | */ |
---|
1416 | Void TComTrQuant::xTransformSkip(Int bitDepth, Pel* piBlkResi, UInt uiStride, Int* psCoeff, Int width, Int height ) |
---|
1417 | { |
---|
1418 | assert( width == height ); |
---|
1419 | UInt uiLog2TrSize = g_aucConvertToBit[ width ] + 2; |
---|
1420 | Int shift = MAX_TR_DYNAMIC_RANGE - bitDepth - uiLog2TrSize; |
---|
1421 | UInt transformSkipShift; |
---|
1422 | Int j,k; |
---|
1423 | if(shift >= 0) |
---|
1424 | { |
---|
1425 | transformSkipShift = shift; |
---|
1426 | for (j = 0; j < height; j++) |
---|
1427 | { |
---|
1428 | for(k = 0; k < width; k ++) |
---|
1429 | { |
---|
1430 | psCoeff[j*height + k] = piBlkResi[j * uiStride + k] << transformSkipShift; |
---|
1431 | } |
---|
1432 | } |
---|
1433 | } |
---|
1434 | else |
---|
1435 | { |
---|
1436 | //The case when uiBitDepth > 13 |
---|
1437 | Int offset; |
---|
1438 | transformSkipShift = -shift; |
---|
1439 | offset = (1 << (transformSkipShift - 1)); |
---|
1440 | for (j = 0; j < height; j++) |
---|
1441 | { |
---|
1442 | for(k = 0; k < width; k ++) |
---|
1443 | { |
---|
1444 | psCoeff[j*height + k] = (piBlkResi[j * uiStride + k] + offset) >> transformSkipShift; |
---|
1445 | } |
---|
1446 | } |
---|
1447 | } |
---|
1448 | } |
---|
1449 | |
---|
1450 | /** Wrapper function between HM interface and core NxN transform skipping |
---|
1451 | * \param plCoef input data (coefficients) |
---|
1452 | * \param pResidual output data (residual) |
---|
1453 | * \param uiStride stride of input residual data |
---|
1454 | * \param iSize transform size (iSize x iSize) |
---|
1455 | */ |
---|
1456 | Void TComTrQuant::xITransformSkip(Int bitDepth, Int* plCoef, Pel* pResidual, UInt uiStride, Int width, Int height ) |
---|
1457 | { |
---|
1458 | assert( width == height ); |
---|
1459 | UInt uiLog2TrSize = g_aucConvertToBit[ width ] + 2; |
---|
1460 | Int shift = MAX_TR_DYNAMIC_RANGE - bitDepth - uiLog2TrSize; |
---|
1461 | UInt transformSkipShift; |
---|
1462 | Int j,k; |
---|
1463 | if(shift > 0) |
---|
1464 | { |
---|
1465 | Int offset; |
---|
1466 | transformSkipShift = shift; |
---|
1467 | offset = (1 << (transformSkipShift -1)); |
---|
1468 | for ( j = 0; j < height; j++ ) |
---|
1469 | { |
---|
1470 | for(k = 0; k < width; k ++) |
---|
1471 | { |
---|
1472 | pResidual[j * uiStride + k] = (plCoef[j*width+k] + offset) >> transformSkipShift; |
---|
1473 | } |
---|
1474 | } |
---|
1475 | } |
---|
1476 | else |
---|
1477 | { |
---|
1478 | //The case when uiBitDepth >= 13 |
---|
1479 | transformSkipShift = - shift; |
---|
1480 | for ( j = 0; j < height; j++ ) |
---|
1481 | { |
---|
1482 | for(k = 0; k < width; k ++) |
---|
1483 | { |
---|
1484 | pResidual[j * uiStride + k] = plCoef[j*width+k] << transformSkipShift; |
---|
1485 | } |
---|
1486 | } |
---|
1487 | } |
---|
1488 | } |
---|
1489 | |
---|
1490 | /** RDOQ with CABAC |
---|
1491 | * \param pcCU pointer to coding unit structure |
---|
1492 | * \param plSrcCoeff pointer to input buffer |
---|
1493 | * \param piDstCoeff reference to pointer to output buffer |
---|
1494 | * \param uiWidth block width |
---|
1495 | * \param uiHeight block height |
---|
1496 | * \param uiAbsSum reference to absolute sum of quantized transform coefficient |
---|
1497 | * \param eTType plane type / luminance or chrominance |
---|
1498 | * \param uiAbsPartIdx absolute partition index |
---|
1499 | * \returns Void |
---|
1500 | * Rate distortion optimized quantization for entropy |
---|
1501 | * coding engines using probability models like CABAC |
---|
1502 | */ |
---|
1503 | Void TComTrQuant::xRateDistOptQuant ( TComDataCU* pcCU, |
---|
1504 | Int* plSrcCoeff, |
---|
1505 | TCoeff* piDstCoeff, |
---|
1506 | #if ADAPTIVE_QP_SELECTION |
---|
1507 | Int*& piArlDstCoeff, |
---|
1508 | #endif |
---|
1509 | UInt uiWidth, |
---|
1510 | UInt uiHeight, |
---|
1511 | UInt& uiAbsSum, |
---|
1512 | TextType eTType, |
---|
1513 | UInt uiAbsPartIdx ) |
---|
1514 | { |
---|
1515 | UInt uiLog2TrSize = g_aucConvertToBit[ uiWidth ] + 2; |
---|
1516 | |
---|
1517 | UInt uiBitDepth = eTType == TEXT_LUMA ? g_bitDepthY : g_bitDepthC; |
---|
1518 | Int iTransformShift = MAX_TR_DYNAMIC_RANGE - uiBitDepth - uiLog2TrSize; // Represents scaling through forward transform |
---|
1519 | UInt uiGoRiceParam = 0; |
---|
1520 | Double d64BlockUncodedCost = 0; |
---|
1521 | const UInt uiLog2BlkSize = g_aucConvertToBit[ uiWidth ] + 2; |
---|
1522 | const UInt uiMaxNumCoeff = uiWidth * uiHeight; |
---|
1523 | Int scalingListType = (pcCU->isIntra(uiAbsPartIdx) ? 0 : 3) + g_eTTable[(Int)eTType]; |
---|
1524 | assert(scalingListType < 6); |
---|
1525 | |
---|
1526 | Int iQBits = QUANT_SHIFT + m_cQP.m_iPer + iTransformShift; // Right shift of non-RDOQ quantizer; level = (coeff*uiQ + offset)>>q_bits |
---|
1527 | Double *pdErrScaleOrg = getErrScaleCoeff(scalingListType,uiLog2TrSize-2,m_cQP.m_iRem); |
---|
1528 | Int *piQCoefOrg = getQuantCoeff(scalingListType,m_cQP.m_iRem,uiLog2TrSize-2); |
---|
1529 | Int *piQCoef = piQCoefOrg; |
---|
1530 | Double *pdErrScale = pdErrScaleOrg; |
---|
1531 | #if ADAPTIVE_QP_SELECTION |
---|
1532 | Int iQBitsC = iQBits - ARL_C_PRECISION; |
---|
1533 | Int iAddC = 1 << (iQBitsC-1); |
---|
1534 | #endif |
---|
1535 | UInt uiScanIdx = pcCU->getCoefScanIdx(uiAbsPartIdx, uiWidth, eTType==TEXT_LUMA, pcCU->isIntra(uiAbsPartIdx)); |
---|
1536 | |
---|
1537 | #if ADAPTIVE_QP_SELECTION |
---|
1538 | memset(piArlDstCoeff, 0, sizeof(Int) * uiMaxNumCoeff); |
---|
1539 | #endif |
---|
1540 | |
---|
1541 | Double pdCostCoeff [ 32 * 32 ]; |
---|
1542 | Double pdCostSig [ 32 * 32 ]; |
---|
1543 | Double pdCostCoeff0[ 32 * 32 ]; |
---|
1544 | ::memset( pdCostCoeff, 0, sizeof(Double) * uiMaxNumCoeff ); |
---|
1545 | ::memset( pdCostSig, 0, sizeof(Double) * uiMaxNumCoeff ); |
---|
1546 | Int rateIncUp [ 32 * 32 ]; |
---|
1547 | Int rateIncDown [ 32 * 32 ]; |
---|
1548 | Int sigRateDelta[ 32 * 32 ]; |
---|
1549 | Int deltaU [ 32 * 32 ]; |
---|
1550 | ::memset( rateIncUp, 0, sizeof(Int) * uiMaxNumCoeff ); |
---|
1551 | ::memset( rateIncDown, 0, sizeof(Int) * uiMaxNumCoeff ); |
---|
1552 | ::memset( sigRateDelta, 0, sizeof(Int) * uiMaxNumCoeff ); |
---|
1553 | ::memset( deltaU, 0, sizeof(Int) * uiMaxNumCoeff ); |
---|
1554 | |
---|
1555 | const UInt * scanCG; |
---|
1556 | { |
---|
1557 | scanCG = g_auiSigLastScan[ uiScanIdx ][ uiLog2BlkSize > 3 ? uiLog2BlkSize-2-1 : 0 ]; |
---|
1558 | if( uiLog2BlkSize == 3 ) |
---|
1559 | { |
---|
1560 | scanCG = g_sigLastScan8x8[ uiScanIdx ]; |
---|
1561 | } |
---|
1562 | else if( uiLog2BlkSize == 5 ) |
---|
1563 | { |
---|
1564 | scanCG = g_sigLastScanCG32x32; |
---|
1565 | } |
---|
1566 | } |
---|
1567 | const UInt uiCGSize = (1 << MLS_CG_SIZE); // 16 |
---|
1568 | Double pdCostCoeffGroupSig[ MLS_GRP_NUM ]; |
---|
1569 | UInt uiSigCoeffGroupFlag[ MLS_GRP_NUM ]; |
---|
1570 | UInt uiNumBlkSide = uiWidth / MLS_CG_SIZE; |
---|
1571 | Int iCGLastScanPos = -1; |
---|
1572 | |
---|
1573 | UInt uiCtxSet = 0; |
---|
1574 | Int c1 = 1; |
---|
1575 | Int c2 = 0; |
---|
1576 | Double d64BaseCost = 0; |
---|
1577 | Int iLastScanPos = -1; |
---|
1578 | |
---|
1579 | UInt c1Idx = 0; |
---|
1580 | UInt c2Idx = 0; |
---|
1581 | Int baseLevel; |
---|
1582 | |
---|
1583 | const UInt *scan = g_auiSigLastScan[ uiScanIdx ][ uiLog2BlkSize - 1 ]; |
---|
1584 | |
---|
1585 | ::memset( pdCostCoeffGroupSig, 0, sizeof(Double) * MLS_GRP_NUM ); |
---|
1586 | ::memset( uiSigCoeffGroupFlag, 0, sizeof(UInt) * MLS_GRP_NUM ); |
---|
1587 | |
---|
1588 | UInt uiCGNum = uiWidth * uiHeight >> MLS_CG_SIZE; |
---|
1589 | Int iScanPos; |
---|
1590 | coeffGroupRDStats rdStats; |
---|
1591 | |
---|
1592 | for (Int iCGScanPos = uiCGNum-1; iCGScanPos >= 0; iCGScanPos--) |
---|
1593 | { |
---|
1594 | UInt uiCGBlkPos = scanCG[ iCGScanPos ]; |
---|
1595 | UInt uiCGPosY = uiCGBlkPos / uiNumBlkSide; |
---|
1596 | UInt uiCGPosX = uiCGBlkPos - (uiCGPosY * uiNumBlkSide); |
---|
1597 | #if MAYBE_BUGFIX |
---|
1598 | rdStats.init(); |
---|
1599 | #else |
---|
1600 | ::memset( &rdStats, 0, sizeof (coeffGroupRDStats)); |
---|
1601 | #endif |
---|
1602 | const Int patternSigCtx = TComTrQuant::calcPatternSigCtx(uiSigCoeffGroupFlag, uiCGPosX, uiCGPosY, uiWidth, uiHeight); |
---|
1603 | for (Int iScanPosinCG = uiCGSize-1; iScanPosinCG >= 0; iScanPosinCG--) |
---|
1604 | { |
---|
1605 | iScanPos = iCGScanPos*uiCGSize + iScanPosinCG; |
---|
1606 | //===== quantization ===== |
---|
1607 | UInt uiBlkPos = scan[iScanPos]; |
---|
1608 | // set coeff |
---|
1609 | Int uiQ = piQCoef[uiBlkPos]; |
---|
1610 | Double dTemp = pdErrScale[uiBlkPos]; |
---|
1611 | Int lLevelDouble = plSrcCoeff[ uiBlkPos ]; |
---|
1612 | lLevelDouble = (Int)min<Int64>((Int64)abs((Int)lLevelDouble) * uiQ , MAX_INT - (1 << (iQBits - 1))); |
---|
1613 | #if ADAPTIVE_QP_SELECTION |
---|
1614 | if( m_bUseAdaptQpSelect ) |
---|
1615 | { |
---|
1616 | piArlDstCoeff[uiBlkPos] = (Int)(( lLevelDouble + iAddC) >> iQBitsC ); |
---|
1617 | } |
---|
1618 | #endif |
---|
1619 | UInt uiMaxAbsLevel = (lLevelDouble + (1 << (iQBits - 1))) >> iQBits; |
---|
1620 | |
---|
1621 | Double dErr = Double( lLevelDouble ); |
---|
1622 | pdCostCoeff0[ iScanPos ] = dErr * dErr * dTemp; |
---|
1623 | d64BlockUncodedCost += pdCostCoeff0[ iScanPos ]; |
---|
1624 | piDstCoeff[ uiBlkPos ] = uiMaxAbsLevel; |
---|
1625 | |
---|
1626 | if ( uiMaxAbsLevel > 0 && iLastScanPos < 0 ) |
---|
1627 | { |
---|
1628 | iLastScanPos = iScanPos; |
---|
1629 | uiCtxSet = (iScanPos < SCAN_SET_SIZE || eTType!=TEXT_LUMA) ? 0 : 2; |
---|
1630 | iCGLastScanPos = iCGScanPos; |
---|
1631 | } |
---|
1632 | |
---|
1633 | if ( iLastScanPos >= 0 ) |
---|
1634 | { |
---|
1635 | //===== coefficient level estimation ===== |
---|
1636 | UInt uiLevel; |
---|
1637 | UInt uiOneCtx = 4 * uiCtxSet + c1; |
---|
1638 | UInt uiAbsCtx = uiCtxSet + c2; |
---|
1639 | |
---|
1640 | if( iScanPos == iLastScanPos ) |
---|
1641 | { |
---|
1642 | uiLevel = xGetCodedLevel( pdCostCoeff[ iScanPos ], pdCostCoeff0[ iScanPos ], pdCostSig[ iScanPos ], |
---|
1643 | lLevelDouble, uiMaxAbsLevel, 0, uiOneCtx, uiAbsCtx, uiGoRiceParam, |
---|
1644 | c1Idx, c2Idx, iQBits, dTemp, 1 ); |
---|
1645 | } |
---|
1646 | else |
---|
1647 | { |
---|
1648 | UInt uiPosY = uiBlkPos >> uiLog2BlkSize; |
---|
1649 | UInt uiPosX = uiBlkPos - ( uiPosY << uiLog2BlkSize ); |
---|
1650 | UShort uiCtxSig = getSigCtxInc( patternSigCtx, uiScanIdx, uiPosX, uiPosY, uiLog2BlkSize, eTType ); |
---|
1651 | uiLevel = xGetCodedLevel( pdCostCoeff[ iScanPos ], pdCostCoeff0[ iScanPos ], pdCostSig[ iScanPos ], |
---|
1652 | lLevelDouble, uiMaxAbsLevel, uiCtxSig, uiOneCtx, uiAbsCtx, uiGoRiceParam, |
---|
1653 | c1Idx, c2Idx, iQBits, dTemp, 0 ); |
---|
1654 | sigRateDelta[ uiBlkPos ] = m_pcEstBitsSbac->significantBits[ uiCtxSig ][ 1 ] - m_pcEstBitsSbac->significantBits[ uiCtxSig ][ 0 ]; |
---|
1655 | } |
---|
1656 | deltaU[ uiBlkPos ] = (lLevelDouble - ((Int)uiLevel << iQBits)) >> (iQBits-8); |
---|
1657 | if( uiLevel > 0 ) |
---|
1658 | { |
---|
1659 | Int rateNow = xGetICRate( uiLevel, uiOneCtx, uiAbsCtx, uiGoRiceParam, c1Idx, c2Idx ); |
---|
1660 | rateIncUp [ uiBlkPos ] = xGetICRate( uiLevel+1, uiOneCtx, uiAbsCtx, uiGoRiceParam, c1Idx, c2Idx ) - rateNow; |
---|
1661 | rateIncDown [ uiBlkPos ] = xGetICRate( uiLevel-1, uiOneCtx, uiAbsCtx, uiGoRiceParam, c1Idx, c2Idx ) - rateNow; |
---|
1662 | } |
---|
1663 | else // uiLevel == 0 |
---|
1664 | { |
---|
1665 | rateIncUp [ uiBlkPos ] = m_pcEstBitsSbac->m_greaterOneBits[ uiOneCtx ][ 0 ]; |
---|
1666 | } |
---|
1667 | piDstCoeff[ uiBlkPos ] = uiLevel; |
---|
1668 | d64BaseCost += pdCostCoeff [ iScanPos ]; |
---|
1669 | |
---|
1670 | |
---|
1671 | baseLevel = (c1Idx < C1FLAG_NUMBER) ? (2 + (c2Idx < C2FLAG_NUMBER)) : 1; |
---|
1672 | if( uiLevel >= baseLevel ) |
---|
1673 | { |
---|
1674 | if(uiLevel > 3*(1<<uiGoRiceParam)) |
---|
1675 | { |
---|
1676 | uiGoRiceParam = min<UInt>(uiGoRiceParam+ 1, 4); |
---|
1677 | } |
---|
1678 | } |
---|
1679 | if ( uiLevel >= 1) |
---|
1680 | { |
---|
1681 | c1Idx ++; |
---|
1682 | } |
---|
1683 | |
---|
1684 | //===== update bin model ===== |
---|
1685 | if( uiLevel > 1 ) |
---|
1686 | { |
---|
1687 | c1 = 0; |
---|
1688 | c2 += (c2 < 2); |
---|
1689 | c2Idx ++; |
---|
1690 | } |
---|
1691 | else if( (c1 < 3) && (c1 > 0) && uiLevel) |
---|
1692 | { |
---|
1693 | c1++; |
---|
1694 | } |
---|
1695 | |
---|
1696 | //===== context set update ===== |
---|
1697 | if( ( iScanPos % SCAN_SET_SIZE == 0 ) && ( iScanPos > 0 ) ) |
---|
1698 | { |
---|
1699 | c2 = 0; |
---|
1700 | uiGoRiceParam = 0; |
---|
1701 | |
---|
1702 | c1Idx = 0; |
---|
1703 | c2Idx = 0; |
---|
1704 | uiCtxSet = (iScanPos == SCAN_SET_SIZE || eTType!=TEXT_LUMA) ? 0 : 2; |
---|
1705 | if( c1 == 0 ) |
---|
1706 | { |
---|
1707 | uiCtxSet++; |
---|
1708 | } |
---|
1709 | c1 = 1; |
---|
1710 | } |
---|
1711 | } |
---|
1712 | else |
---|
1713 | { |
---|
1714 | d64BaseCost += pdCostCoeff0[ iScanPos ]; |
---|
1715 | } |
---|
1716 | rdStats.d64SigCost += pdCostSig[ iScanPos ]; |
---|
1717 | if (iScanPosinCG == 0 ) |
---|
1718 | { |
---|
1719 | rdStats.d64SigCost_0 = pdCostSig[ iScanPos ]; |
---|
1720 | } |
---|
1721 | if (piDstCoeff[ uiBlkPos ] ) |
---|
1722 | { |
---|
1723 | uiSigCoeffGroupFlag[ uiCGBlkPos ] = 1; |
---|
1724 | rdStats.d64CodedLevelandDist += pdCostCoeff[ iScanPos ] - pdCostSig[ iScanPos ]; |
---|
1725 | rdStats.d64UncodedDist += pdCostCoeff0[ iScanPos ]; |
---|
1726 | if ( iScanPosinCG != 0 ) |
---|
1727 | { |
---|
1728 | rdStats.iNNZbeforePos0++; |
---|
1729 | } |
---|
1730 | } |
---|
1731 | } //end for (iScanPosinCG) |
---|
1732 | |
---|
1733 | if (iCGLastScanPos >= 0) |
---|
1734 | { |
---|
1735 | if( iCGScanPos ) |
---|
1736 | { |
---|
1737 | if (uiSigCoeffGroupFlag[ uiCGBlkPos ] == 0) |
---|
1738 | { |
---|
1739 | UInt uiCtxSig = getSigCoeffGroupCtxInc( uiSigCoeffGroupFlag, uiCGPosX, uiCGPosY, uiWidth, uiHeight); |
---|
1740 | d64BaseCost += xGetRateSigCoeffGroup(0, uiCtxSig) - rdStats.d64SigCost;; |
---|
1741 | pdCostCoeffGroupSig[ iCGScanPos ] = xGetRateSigCoeffGroup(0, uiCtxSig); |
---|
1742 | } |
---|
1743 | else |
---|
1744 | { |
---|
1745 | if (iCGScanPos < iCGLastScanPos) //skip the last coefficient group, which will be handled together with last position below. |
---|
1746 | { |
---|
1747 | if ( rdStats.iNNZbeforePos0 == 0 ) |
---|
1748 | { |
---|
1749 | d64BaseCost -= rdStats.d64SigCost_0; |
---|
1750 | rdStats.d64SigCost -= rdStats.d64SigCost_0; |
---|
1751 | } |
---|
1752 | // rd-cost if SigCoeffGroupFlag = 0, initialization |
---|
1753 | Double d64CostZeroCG = d64BaseCost; |
---|
1754 | |
---|
1755 | // add SigCoeffGroupFlag cost to total cost |
---|
1756 | UInt uiCtxSig = getSigCoeffGroupCtxInc( uiSigCoeffGroupFlag, uiCGPosX, uiCGPosY, uiWidth, uiHeight); |
---|
1757 | if (iCGScanPos < iCGLastScanPos) |
---|
1758 | { |
---|
1759 | d64BaseCost += xGetRateSigCoeffGroup(1, uiCtxSig); |
---|
1760 | d64CostZeroCG += xGetRateSigCoeffGroup(0, uiCtxSig); |
---|
1761 | pdCostCoeffGroupSig[ iCGScanPos ] = xGetRateSigCoeffGroup(1, uiCtxSig); |
---|
1762 | } |
---|
1763 | |
---|
1764 | // try to convert the current coeff group from non-zero to all-zero |
---|
1765 | d64CostZeroCG += rdStats.d64UncodedDist; // distortion for resetting non-zero levels to zero levels |
---|
1766 | d64CostZeroCG -= rdStats.d64CodedLevelandDist; // distortion and level cost for keeping all non-zero levels |
---|
1767 | d64CostZeroCG -= rdStats.d64SigCost; // sig cost for all coeffs, including zero levels and non-zerl levels |
---|
1768 | |
---|
1769 | // if we can save cost, change this block to all-zero block |
---|
1770 | if ( d64CostZeroCG < d64BaseCost ) |
---|
1771 | { |
---|
1772 | uiSigCoeffGroupFlag[ uiCGBlkPos ] = 0; |
---|
1773 | d64BaseCost = d64CostZeroCG; |
---|
1774 | if (iCGScanPos < iCGLastScanPos) |
---|
1775 | { |
---|
1776 | pdCostCoeffGroupSig[ iCGScanPos ] = xGetRateSigCoeffGroup(0, uiCtxSig); |
---|
1777 | } |
---|
1778 | // reset coeffs to 0 in this block |
---|
1779 | for (Int iScanPosinCG = uiCGSize-1; iScanPosinCG >= 0; iScanPosinCG--) |
---|
1780 | { |
---|
1781 | iScanPos = iCGScanPos*uiCGSize + iScanPosinCG; |
---|
1782 | UInt uiBlkPos = scan[ iScanPos ]; |
---|
1783 | |
---|
1784 | if (piDstCoeff[ uiBlkPos ]) |
---|
1785 | { |
---|
1786 | piDstCoeff [ uiBlkPos ] = 0; |
---|
1787 | pdCostCoeff[ iScanPos ] = pdCostCoeff0[ iScanPos ]; |
---|
1788 | pdCostSig [ iScanPos ] = 0; |
---|
1789 | } |
---|
1790 | } |
---|
1791 | } // end if ( d64CostAllZeros < d64BaseCost ) |
---|
1792 | } |
---|
1793 | } // end if if (uiSigCoeffGroupFlag[ uiCGBlkPos ] == 0) |
---|
1794 | } |
---|
1795 | else |
---|
1796 | { |
---|
1797 | uiSigCoeffGroupFlag[ uiCGBlkPos ] = 1; |
---|
1798 | } |
---|
1799 | } |
---|
1800 | } //end for (iCGScanPos) |
---|
1801 | |
---|
1802 | //===== estimate last position ===== |
---|
1803 | if ( iLastScanPos < 0 ) |
---|
1804 | { |
---|
1805 | return; |
---|
1806 | } |
---|
1807 | |
---|
1808 | Double d64BestCost = 0; |
---|
1809 | Int ui16CtxCbf = 0; |
---|
1810 | Int iBestLastIdxP1 = 0; |
---|
1811 | if( !pcCU->isIntra( uiAbsPartIdx ) && eTType == TEXT_LUMA && pcCU->getTransformIdx( uiAbsPartIdx ) == 0 ) |
---|
1812 | { |
---|
1813 | ui16CtxCbf = 0; |
---|
1814 | d64BestCost = d64BlockUncodedCost + xGetICost( m_pcEstBitsSbac->blockRootCbpBits[ ui16CtxCbf ][ 0 ] ); |
---|
1815 | d64BaseCost += xGetICost( m_pcEstBitsSbac->blockRootCbpBits[ ui16CtxCbf ][ 1 ] ); |
---|
1816 | } |
---|
1817 | else |
---|
1818 | { |
---|
1819 | ui16CtxCbf = pcCU->getCtxQtCbf( eTType, pcCU->getTransformIdx( uiAbsPartIdx ) ); |
---|
1820 | ui16CtxCbf = ( eTType ? TEXT_CHROMA : eTType ) * NUM_QT_CBF_CTX + ui16CtxCbf; |
---|
1821 | d64BestCost = d64BlockUncodedCost + xGetICost( m_pcEstBitsSbac->blockCbpBits[ ui16CtxCbf ][ 0 ] ); |
---|
1822 | d64BaseCost += xGetICost( m_pcEstBitsSbac->blockCbpBits[ ui16CtxCbf ][ 1 ] ); |
---|
1823 | } |
---|
1824 | |
---|
1825 | Bool bFoundLast = false; |
---|
1826 | for (Int iCGScanPos = iCGLastScanPos; iCGScanPos >= 0; iCGScanPos--) |
---|
1827 | { |
---|
1828 | UInt uiCGBlkPos = scanCG[ iCGScanPos ]; |
---|
1829 | |
---|
1830 | d64BaseCost -= pdCostCoeffGroupSig [ iCGScanPos ]; |
---|
1831 | if (uiSigCoeffGroupFlag[ uiCGBlkPos ]) |
---|
1832 | { |
---|
1833 | for (Int iScanPosinCG = uiCGSize-1; iScanPosinCG >= 0; iScanPosinCG--) |
---|
1834 | { |
---|
1835 | iScanPos = iCGScanPos*uiCGSize + iScanPosinCG; |
---|
1836 | if (iScanPos > iLastScanPos) continue; |
---|
1837 | UInt uiBlkPos = scan[iScanPos]; |
---|
1838 | |
---|
1839 | if( piDstCoeff[ uiBlkPos ] ) |
---|
1840 | { |
---|
1841 | UInt uiPosY = uiBlkPos >> uiLog2BlkSize; |
---|
1842 | UInt uiPosX = uiBlkPos - ( uiPosY << uiLog2BlkSize ); |
---|
1843 | |
---|
1844 | Double d64CostLast= uiScanIdx == SCAN_VER ? xGetRateLast( uiPosY, uiPosX ) : xGetRateLast( uiPosX, uiPosY ); |
---|
1845 | Double totalCost = d64BaseCost + d64CostLast - pdCostSig[ iScanPos ]; |
---|
1846 | |
---|
1847 | if( totalCost < d64BestCost ) |
---|
1848 | { |
---|
1849 | iBestLastIdxP1 = iScanPos + 1; |
---|
1850 | d64BestCost = totalCost; |
---|
1851 | } |
---|
1852 | if( piDstCoeff[ uiBlkPos ] > 1 ) |
---|
1853 | { |
---|
1854 | bFoundLast = true; |
---|
1855 | break; |
---|
1856 | } |
---|
1857 | d64BaseCost -= pdCostCoeff[ iScanPos ]; |
---|
1858 | d64BaseCost += pdCostCoeff0[ iScanPos ]; |
---|
1859 | } |
---|
1860 | else |
---|
1861 | { |
---|
1862 | d64BaseCost -= pdCostSig[ iScanPos ]; |
---|
1863 | } |
---|
1864 | } //end for |
---|
1865 | if (bFoundLast) |
---|
1866 | { |
---|
1867 | break; |
---|
1868 | } |
---|
1869 | } // end if (uiSigCoeffGroupFlag[ uiCGBlkPos ]) |
---|
1870 | } // end for |
---|
1871 | |
---|
1872 | for ( Int scanPos = 0; scanPos < iBestLastIdxP1; scanPos++ ) |
---|
1873 | { |
---|
1874 | Int blkPos = scan[ scanPos ]; |
---|
1875 | Int level = piDstCoeff[ blkPos ]; |
---|
1876 | uiAbsSum += level; |
---|
1877 | piDstCoeff[ blkPos ] = ( plSrcCoeff[ blkPos ] < 0 ) ? -level : level; |
---|
1878 | } |
---|
1879 | |
---|
1880 | //===== clean uncoded coefficients ===== |
---|
1881 | for ( Int scanPos = iBestLastIdxP1; scanPos <= iLastScanPos; scanPos++ ) |
---|
1882 | { |
---|
1883 | piDstCoeff[ scan[ scanPos ] ] = 0; |
---|
1884 | } |
---|
1885 | |
---|
1886 | if( pcCU->getSlice()->getPPS()->getSignHideFlag() && uiAbsSum>=2) |
---|
1887 | { |
---|
1888 | Int64 rdFactor = (Int64) ( |
---|
1889 | g_invQuantScales[m_cQP.rem()] * g_invQuantScales[m_cQP.rem()] * (1<<(2*m_cQP.m_iPer)) |
---|
1890 | / m_dLambda / 16 / (1<<DISTORTION_PRECISION_ADJUSTMENT(2*(uiBitDepth-8))) |
---|
1891 | + 0.5); |
---|
1892 | Int lastCG = -1; |
---|
1893 | Int absSum = 0 ; |
---|
1894 | Int n ; |
---|
1895 | |
---|
1896 | for( Int subSet = (uiWidth*uiHeight-1) >> LOG2_SCAN_SET_SIZE; subSet >= 0; subSet-- ) |
---|
1897 | { |
---|
1898 | Int subPos = subSet << LOG2_SCAN_SET_SIZE; |
---|
1899 | Int firstNZPosInCG=SCAN_SET_SIZE , lastNZPosInCG=-1 ; |
---|
1900 | absSum = 0 ; |
---|
1901 | |
---|
1902 | for(n = SCAN_SET_SIZE-1; n >= 0; --n ) |
---|
1903 | { |
---|
1904 | if( piDstCoeff[ scan[ n + subPos ]] ) |
---|
1905 | { |
---|
1906 | lastNZPosInCG = n; |
---|
1907 | break; |
---|
1908 | } |
---|
1909 | } |
---|
1910 | |
---|
1911 | for(n = 0; n <SCAN_SET_SIZE; n++ ) |
---|
1912 | { |
---|
1913 | if( piDstCoeff[ scan[ n + subPos ]] ) |
---|
1914 | { |
---|
1915 | firstNZPosInCG = n; |
---|
1916 | break; |
---|
1917 | } |
---|
1918 | } |
---|
1919 | |
---|
1920 | for(n = firstNZPosInCG; n <=lastNZPosInCG; n++ ) |
---|
1921 | { |
---|
1922 | absSum += piDstCoeff[ scan[ n + subPos ]]; |
---|
1923 | } |
---|
1924 | |
---|
1925 | if(lastNZPosInCG>=0 && lastCG==-1) |
---|
1926 | { |
---|
1927 | lastCG = 1; |
---|
1928 | } |
---|
1929 | |
---|
1930 | if( lastNZPosInCG-firstNZPosInCG>=SBH_THRESHOLD ) |
---|
1931 | { |
---|
1932 | UInt signbit = (piDstCoeff[scan[subPos+firstNZPosInCG]]>0?0:1); |
---|
1933 | if( signbit!=(absSum&0x1) ) // hide but need tune |
---|
1934 | { |
---|
1935 | // calculate the cost |
---|
1936 | Int64 minCostInc = MAX_INT64, curCost=MAX_INT64; |
---|
1937 | Int minPos =-1, finalChange=0, curChange=0; |
---|
1938 | |
---|
1939 | for( n = (lastCG==1?lastNZPosInCG:SCAN_SET_SIZE-1) ; n >= 0; --n ) |
---|
1940 | { |
---|
1941 | UInt uiBlkPos = scan[ n + subPos ]; |
---|
1942 | if(piDstCoeff[ uiBlkPos ] != 0 ) |
---|
1943 | { |
---|
1944 | Int64 costUp = rdFactor * ( - deltaU[uiBlkPos] ) + rateIncUp[uiBlkPos] ; |
---|
1945 | Int64 costDown = rdFactor * ( deltaU[uiBlkPos] ) + rateIncDown[uiBlkPos] |
---|
1946 | - ( abs(piDstCoeff[uiBlkPos])==1?((1<<15)+sigRateDelta[uiBlkPos]):0 ); |
---|
1947 | |
---|
1948 | if(lastCG==1 && lastNZPosInCG==n && abs(piDstCoeff[uiBlkPos])==1) |
---|
1949 | { |
---|
1950 | costDown -= (4<<15) ; |
---|
1951 | } |
---|
1952 | |
---|
1953 | if(costUp<costDown) |
---|
1954 | { |
---|
1955 | curCost = costUp; |
---|
1956 | curChange = 1 ; |
---|
1957 | } |
---|
1958 | else |
---|
1959 | { |
---|
1960 | curChange = -1 ; |
---|
1961 | if(n==firstNZPosInCG && abs(piDstCoeff[uiBlkPos])==1) |
---|
1962 | { |
---|
1963 | curCost = MAX_INT64 ; |
---|
1964 | } |
---|
1965 | else |
---|
1966 | { |
---|
1967 | curCost = costDown ; |
---|
1968 | } |
---|
1969 | } |
---|
1970 | } |
---|
1971 | else |
---|
1972 | { |
---|
1973 | curCost = rdFactor * ( - (abs(deltaU[uiBlkPos])) ) + (1<<15) + rateIncUp[uiBlkPos] + sigRateDelta[uiBlkPos] ; |
---|
1974 | curChange = 1 ; |
---|
1975 | |
---|
1976 | if(n<firstNZPosInCG) |
---|
1977 | { |
---|
1978 | UInt thissignbit = (plSrcCoeff[uiBlkPos]>=0?0:1); |
---|
1979 | if(thissignbit != signbit ) |
---|
1980 | { |
---|
1981 | curCost = MAX_INT64; |
---|
1982 | } |
---|
1983 | } |
---|
1984 | } |
---|
1985 | |
---|
1986 | if( curCost<minCostInc) |
---|
1987 | { |
---|
1988 | minCostInc = curCost ; |
---|
1989 | finalChange = curChange ; |
---|
1990 | minPos = uiBlkPos ; |
---|
1991 | } |
---|
1992 | } |
---|
1993 | |
---|
1994 | if(piDstCoeff[minPos] == 32767 || piDstCoeff[minPos] == -32768) |
---|
1995 | { |
---|
1996 | finalChange = -1; |
---|
1997 | } |
---|
1998 | |
---|
1999 | if(plSrcCoeff[minPos]>=0) |
---|
2000 | { |
---|
2001 | piDstCoeff[minPos] += finalChange ; |
---|
2002 | } |
---|
2003 | else |
---|
2004 | { |
---|
2005 | piDstCoeff[minPos] -= finalChange ; |
---|
2006 | } |
---|
2007 | } |
---|
2008 | } |
---|
2009 | |
---|
2010 | if(lastCG==1) |
---|
2011 | { |
---|
2012 | lastCG=0 ; |
---|
2013 | } |
---|
2014 | } |
---|
2015 | } |
---|
2016 | } |
---|
2017 | |
---|
2018 | /** Pattern decision for context derivation process of significant_coeff_flag |
---|
2019 | * \param sigCoeffGroupFlag pointer to prior coded significant coeff group |
---|
2020 | * \param posXCG column of current coefficient group |
---|
2021 | * \param posYCG row of current coefficient group |
---|
2022 | * \param width width of the block |
---|
2023 | * \param height height of the block |
---|
2024 | * \returns pattern for current coefficient group |
---|
2025 | */ |
---|
2026 | Int TComTrQuant::calcPatternSigCtx( const UInt* sigCoeffGroupFlag, UInt posXCG, UInt posYCG, Int width, Int height ) |
---|
2027 | { |
---|
2028 | if( width == 4 && height == 4 ) return -1; |
---|
2029 | |
---|
2030 | UInt sigRight = 0; |
---|
2031 | UInt sigLower = 0; |
---|
2032 | |
---|
2033 | width >>= 2; |
---|
2034 | height >>= 2; |
---|
2035 | if( posXCG < width - 1 ) |
---|
2036 | { |
---|
2037 | sigRight = (sigCoeffGroupFlag[ posYCG * width + posXCG + 1 ] != 0); |
---|
2038 | } |
---|
2039 | if (posYCG < height - 1 ) |
---|
2040 | { |
---|
2041 | sigLower = (sigCoeffGroupFlag[ (posYCG + 1 ) * width + posXCG ] != 0); |
---|
2042 | } |
---|
2043 | return sigRight + (sigLower<<1); |
---|
2044 | } |
---|
2045 | |
---|
2046 | /** Context derivation process of coeff_abs_significant_flag |
---|
2047 | * \param patternSigCtx pattern for current coefficient group |
---|
2048 | * \param posX column of current scan position |
---|
2049 | * \param posY row of current scan position |
---|
2050 | * \param log2BlockSize log2 value of block size (square block) |
---|
2051 | * \param width width of the block |
---|
2052 | * \param height height of the block |
---|
2053 | * \param textureType texture type (TEXT_LUMA...) |
---|
2054 | * \returns ctxInc for current scan position |
---|
2055 | */ |
---|
2056 | Int TComTrQuant::getSigCtxInc ( |
---|
2057 | Int patternSigCtx, |
---|
2058 | UInt scanIdx, |
---|
2059 | Int posX, |
---|
2060 | Int posY, |
---|
2061 | Int log2BlockSize, |
---|
2062 | TextType textureType |
---|
2063 | ) |
---|
2064 | { |
---|
2065 | const Int ctxIndMap[16] = |
---|
2066 | { |
---|
2067 | 0, 1, 4, 5, |
---|
2068 | 2, 3, 4, 5, |
---|
2069 | 6, 6, 8, 8, |
---|
2070 | 7, 7, 8, 8 |
---|
2071 | }; |
---|
2072 | |
---|
2073 | if( posX + posY == 0 ) |
---|
2074 | { |
---|
2075 | return 0; |
---|
2076 | } |
---|
2077 | |
---|
2078 | if ( log2BlockSize == 2 ) |
---|
2079 | { |
---|
2080 | return ctxIndMap[ 4 * posY + posX ]; |
---|
2081 | } |
---|
2082 | |
---|
2083 | Int offset = log2BlockSize == 3 ? (scanIdx==SCAN_DIAG ? 9 : 15) : (textureType == TEXT_LUMA ? 21 : 12); |
---|
2084 | |
---|
2085 | Int posXinSubset = posX-((posX>>2)<<2); |
---|
2086 | Int posYinSubset = posY-((posY>>2)<<2); |
---|
2087 | Int cnt = 0; |
---|
2088 | if(patternSigCtx==0) |
---|
2089 | { |
---|
2090 | cnt = posXinSubset+posYinSubset<=2 ? (posXinSubset+posYinSubset==0 ? 2 : 1) : 0; |
---|
2091 | } |
---|
2092 | else if(patternSigCtx==1) |
---|
2093 | { |
---|
2094 | cnt = posYinSubset<=1 ? (posYinSubset==0 ? 2 : 1) : 0; |
---|
2095 | } |
---|
2096 | else if(patternSigCtx==2) |
---|
2097 | { |
---|
2098 | cnt = posXinSubset<=1 ? (posXinSubset==0 ? 2 : 1) : 0; |
---|
2099 | } |
---|
2100 | else |
---|
2101 | { |
---|
2102 | cnt = 2; |
---|
2103 | } |
---|
2104 | |
---|
2105 | return (( textureType == TEXT_LUMA && ((posX>>2) + (posY>>2)) > 0 ) ? 3 : 0) + offset + cnt; |
---|
2106 | } |
---|
2107 | |
---|
2108 | /** Get the best level in RD sense |
---|
2109 | * \param rd64CodedCost reference to coded cost |
---|
2110 | * \param rd64CodedCost0 reference to cost when coefficient is 0 |
---|
2111 | * \param rd64CodedCostSig reference to cost of significant coefficient |
---|
2112 | * \param lLevelDouble reference to unscaled quantized level |
---|
2113 | * \param uiMaxAbsLevel scaled quantized level |
---|
2114 | * \param ui16CtxNumSig current ctxInc for coeff_abs_significant_flag |
---|
2115 | * \param ui16CtxNumOne current ctxInc for coeff_abs_level_greater1 (1st bin of coeff_abs_level_minus1 in AVC) |
---|
2116 | * \param ui16CtxNumAbs current ctxInc for coeff_abs_level_greater2 (remaining bins of coeff_abs_level_minus1 in AVC) |
---|
2117 | * \param ui16AbsGoRice current Rice parameter for coeff_abs_level_minus3 |
---|
2118 | * \param iQBits quantization step size |
---|
2119 | * \param dTemp correction factor |
---|
2120 | * \param bLast indicates if the coefficient is the last significant |
---|
2121 | * \returns best quantized transform level for given scan position |
---|
2122 | * This method calculates the best quantized transform level for a given scan position. |
---|
2123 | */ |
---|
2124 | __inline UInt TComTrQuant::xGetCodedLevel ( Double& rd64CodedCost, |
---|
2125 | Double& rd64CodedCost0, |
---|
2126 | Double& rd64CodedCostSig, |
---|
2127 | Int lLevelDouble, |
---|
2128 | UInt uiMaxAbsLevel, |
---|
2129 | UShort ui16CtxNumSig, |
---|
2130 | UShort ui16CtxNumOne, |
---|
2131 | UShort ui16CtxNumAbs, |
---|
2132 | UShort ui16AbsGoRice, |
---|
2133 | UInt c1Idx, |
---|
2134 | UInt c2Idx, |
---|
2135 | Int iQBits, |
---|
2136 | Double dTemp, |
---|
2137 | Bool bLast ) const |
---|
2138 | { |
---|
2139 | Double dCurrCostSig = 0; |
---|
2140 | UInt uiBestAbsLevel = 0; |
---|
2141 | |
---|
2142 | if( !bLast && uiMaxAbsLevel < 3 ) |
---|
2143 | { |
---|
2144 | rd64CodedCostSig = xGetRateSigCoef( 0, ui16CtxNumSig ); |
---|
2145 | rd64CodedCost = rd64CodedCost0 + rd64CodedCostSig; |
---|
2146 | if( uiMaxAbsLevel == 0 ) |
---|
2147 | { |
---|
2148 | return uiBestAbsLevel; |
---|
2149 | } |
---|
2150 | } |
---|
2151 | else |
---|
2152 | { |
---|
2153 | rd64CodedCost = MAX_DOUBLE; |
---|
2154 | } |
---|
2155 | |
---|
2156 | if( !bLast ) |
---|
2157 | { |
---|
2158 | dCurrCostSig = xGetRateSigCoef( 1, ui16CtxNumSig ); |
---|
2159 | } |
---|
2160 | |
---|
2161 | UInt uiMinAbsLevel = ( uiMaxAbsLevel > 1 ? uiMaxAbsLevel - 1 : 1 ); |
---|
2162 | for( Int uiAbsLevel = uiMaxAbsLevel; uiAbsLevel >= uiMinAbsLevel ; uiAbsLevel-- ) |
---|
2163 | { |
---|
2164 | Double dErr = Double( lLevelDouble - ( uiAbsLevel << iQBits ) ); |
---|
2165 | Double dCurrCost = dErr * dErr * dTemp + xGetICRateCost( uiAbsLevel, ui16CtxNumOne, ui16CtxNumAbs, ui16AbsGoRice, c1Idx, c2Idx ); |
---|
2166 | dCurrCost += dCurrCostSig; |
---|
2167 | |
---|
2168 | if( dCurrCost < rd64CodedCost ) |
---|
2169 | { |
---|
2170 | uiBestAbsLevel = uiAbsLevel; |
---|
2171 | rd64CodedCost = dCurrCost; |
---|
2172 | rd64CodedCostSig = dCurrCostSig; |
---|
2173 | } |
---|
2174 | } |
---|
2175 | |
---|
2176 | return uiBestAbsLevel; |
---|
2177 | } |
---|
2178 | |
---|
2179 | /** Calculates the cost for specific absolute transform level |
---|
2180 | * \param uiAbsLevel scaled quantized level |
---|
2181 | * \param ui16CtxNumOne current ctxInc for coeff_abs_level_greater1 (1st bin of coeff_abs_level_minus1 in AVC) |
---|
2182 | * \param ui16CtxNumAbs current ctxInc for coeff_abs_level_greater2 (remaining bins of coeff_abs_level_minus1 in AVC) |
---|
2183 | * \param ui16AbsGoRice Rice parameter for coeff_abs_level_minus3 |
---|
2184 | * \returns cost of given absolute transform level |
---|
2185 | */ |
---|
2186 | __inline Double TComTrQuant::xGetICRateCost ( UInt uiAbsLevel, |
---|
2187 | UShort ui16CtxNumOne, |
---|
2188 | UShort ui16CtxNumAbs, |
---|
2189 | UShort ui16AbsGoRice |
---|
2190 | , UInt c1Idx, |
---|
2191 | UInt c2Idx |
---|
2192 | ) const |
---|
2193 | { |
---|
2194 | Double iRate = xGetIEPRate(); |
---|
2195 | UInt baseLevel = (c1Idx < C1FLAG_NUMBER)? (2 + (c2Idx < C2FLAG_NUMBER)) : 1; |
---|
2196 | |
---|
2197 | if ( uiAbsLevel >= baseLevel ) |
---|
2198 | { |
---|
2199 | UInt symbol = uiAbsLevel - baseLevel; |
---|
2200 | UInt length; |
---|
2201 | if (symbol < (COEF_REMAIN_BIN_REDUCTION << ui16AbsGoRice)) |
---|
2202 | { |
---|
2203 | length = symbol>>ui16AbsGoRice; |
---|
2204 | iRate += (length+1+ui16AbsGoRice)<< 15; |
---|
2205 | } |
---|
2206 | else |
---|
2207 | { |
---|
2208 | length = ui16AbsGoRice; |
---|
2209 | symbol = symbol - ( COEF_REMAIN_BIN_REDUCTION << ui16AbsGoRice); |
---|
2210 | while (symbol >= (1<<length)) |
---|
2211 | { |
---|
2212 | symbol -= (1<<(length++)); |
---|
2213 | } |
---|
2214 | iRate += (COEF_REMAIN_BIN_REDUCTION+length+1-ui16AbsGoRice+length)<< 15; |
---|
2215 | } |
---|
2216 | if (c1Idx < C1FLAG_NUMBER) |
---|
2217 | { |
---|
2218 | iRate += m_pcEstBitsSbac->m_greaterOneBits[ ui16CtxNumOne ][ 1 ]; |
---|
2219 | |
---|
2220 | if (c2Idx < C2FLAG_NUMBER) |
---|
2221 | { |
---|
2222 | iRate += m_pcEstBitsSbac->m_levelAbsBits[ ui16CtxNumAbs ][ 1 ]; |
---|
2223 | } |
---|
2224 | } |
---|
2225 | } |
---|
2226 | else |
---|
2227 | if( uiAbsLevel == 1 ) |
---|
2228 | { |
---|
2229 | iRate += m_pcEstBitsSbac->m_greaterOneBits[ ui16CtxNumOne ][ 0 ]; |
---|
2230 | } |
---|
2231 | else if( uiAbsLevel == 2 ) |
---|
2232 | { |
---|
2233 | iRate += m_pcEstBitsSbac->m_greaterOneBits[ ui16CtxNumOne ][ 1 ]; |
---|
2234 | iRate += m_pcEstBitsSbac->m_levelAbsBits[ ui16CtxNumAbs ][ 0 ]; |
---|
2235 | } |
---|
2236 | else |
---|
2237 | { |
---|
2238 | assert (0); |
---|
2239 | } |
---|
2240 | return xGetICost( iRate ); |
---|
2241 | } |
---|
2242 | |
---|
2243 | __inline Int TComTrQuant::xGetICRate ( UInt uiAbsLevel, |
---|
2244 | UShort ui16CtxNumOne, |
---|
2245 | UShort ui16CtxNumAbs, |
---|
2246 | UShort ui16AbsGoRice |
---|
2247 | , UInt c1Idx, |
---|
2248 | UInt c2Idx |
---|
2249 | ) const |
---|
2250 | { |
---|
2251 | Int iRate = 0; |
---|
2252 | UInt baseLevel = (c1Idx < C1FLAG_NUMBER)? (2 + (c2Idx < C2FLAG_NUMBER)) : 1; |
---|
2253 | |
---|
2254 | if ( uiAbsLevel >= baseLevel ) |
---|
2255 | { |
---|
2256 | UInt uiSymbol = uiAbsLevel - baseLevel; |
---|
2257 | UInt uiMaxVlc = g_auiGoRiceRange[ ui16AbsGoRice ]; |
---|
2258 | Bool bExpGolomb = ( uiSymbol > uiMaxVlc ); |
---|
2259 | |
---|
2260 | if( bExpGolomb ) |
---|
2261 | { |
---|
2262 | uiAbsLevel = uiSymbol - uiMaxVlc; |
---|
2263 | Int iEGS = 1; for( UInt uiMax = 2; uiAbsLevel >= uiMax; uiMax <<= 1, iEGS += 2 ); |
---|
2264 | iRate += iEGS << 15; |
---|
2265 | uiSymbol = min<UInt>( uiSymbol, ( uiMaxVlc + 1 ) ); |
---|
2266 | } |
---|
2267 | |
---|
2268 | UShort ui16PrefLen = UShort( uiSymbol >> ui16AbsGoRice ) + 1; |
---|
2269 | UShort ui16NumBins = min<UInt>( ui16PrefLen, g_auiGoRicePrefixLen[ ui16AbsGoRice ] ) + ui16AbsGoRice; |
---|
2270 | |
---|
2271 | iRate += ui16NumBins << 15; |
---|
2272 | |
---|
2273 | if (c1Idx < C1FLAG_NUMBER) |
---|
2274 | { |
---|
2275 | iRate += m_pcEstBitsSbac->m_greaterOneBits[ ui16CtxNumOne ][ 1 ]; |
---|
2276 | |
---|
2277 | if (c2Idx < C2FLAG_NUMBER) |
---|
2278 | { |
---|
2279 | iRate += m_pcEstBitsSbac->m_levelAbsBits[ ui16CtxNumAbs ][ 1 ]; |
---|
2280 | } |
---|
2281 | } |
---|
2282 | } |
---|
2283 | else |
---|
2284 | if( uiAbsLevel == 0 ) |
---|
2285 | { |
---|
2286 | return 0; |
---|
2287 | } |
---|
2288 | else if( uiAbsLevel == 1 ) |
---|
2289 | { |
---|
2290 | iRate += m_pcEstBitsSbac->m_greaterOneBits[ ui16CtxNumOne ][ 0 ]; |
---|
2291 | } |
---|
2292 | else if( uiAbsLevel == 2 ) |
---|
2293 | { |
---|
2294 | iRate += m_pcEstBitsSbac->m_greaterOneBits[ ui16CtxNumOne ][ 1 ]; |
---|
2295 | iRate += m_pcEstBitsSbac->m_levelAbsBits[ ui16CtxNumAbs ][ 0 ]; |
---|
2296 | } |
---|
2297 | else |
---|
2298 | { |
---|
2299 | assert(0); |
---|
2300 | } |
---|
2301 | return iRate; |
---|
2302 | } |
---|
2303 | |
---|
2304 | __inline Double TComTrQuant::xGetRateSigCoeffGroup ( UShort uiSignificanceCoeffGroup, |
---|
2305 | UShort ui16CtxNumSig ) const |
---|
2306 | { |
---|
2307 | return xGetICost( m_pcEstBitsSbac->significantCoeffGroupBits[ ui16CtxNumSig ][ uiSignificanceCoeffGroup ] ); |
---|
2308 | } |
---|
2309 | |
---|
2310 | /** Calculates the cost of signaling the last significant coefficient in the block |
---|
2311 | * \param uiPosX X coordinate of the last significant coefficient |
---|
2312 | * \param uiPosY Y coordinate of the last significant coefficient |
---|
2313 | * \returns cost of last significant coefficient |
---|
2314 | */ |
---|
2315 | /* |
---|
2316 | * \param uiWidth width of the transform unit (TU) |
---|
2317 | */ |
---|
2318 | __inline Double TComTrQuant::xGetRateLast ( const UInt uiPosX, |
---|
2319 | const UInt uiPosY ) const |
---|
2320 | { |
---|
2321 | UInt uiCtxX = g_uiGroupIdx[uiPosX]; |
---|
2322 | UInt uiCtxY = g_uiGroupIdx[uiPosY]; |
---|
2323 | Double uiCost = m_pcEstBitsSbac->lastXBits[ uiCtxX ] + m_pcEstBitsSbac->lastYBits[ uiCtxY ]; |
---|
2324 | if( uiCtxX > 3 ) |
---|
2325 | { |
---|
2326 | uiCost += xGetIEPRate() * ((uiCtxX-2)>>1); |
---|
2327 | } |
---|
2328 | if( uiCtxY > 3 ) |
---|
2329 | { |
---|
2330 | uiCost += xGetIEPRate() * ((uiCtxY-2)>>1); |
---|
2331 | } |
---|
2332 | return xGetICost( uiCost ); |
---|
2333 | } |
---|
2334 | |
---|
2335 | /** Calculates the cost for specific absolute transform level |
---|
2336 | * \param uiAbsLevel scaled quantized level |
---|
2337 | * \param ui16CtxNumOne current ctxInc for coeff_abs_level_greater1 (1st bin of coeff_abs_level_minus1 in AVC) |
---|
2338 | * \param ui16CtxNumAbs current ctxInc for coeff_abs_level_greater2 (remaining bins of coeff_abs_level_minus1 in AVC) |
---|
2339 | * \param ui16CtxBase current global offset for coeff_abs_level_greater1 and coeff_abs_level_greater2 |
---|
2340 | * \returns cost of given absolute transform level |
---|
2341 | */ |
---|
2342 | __inline Double TComTrQuant::xGetRateSigCoef ( UShort uiSignificance, |
---|
2343 | UShort ui16CtxNumSig ) const |
---|
2344 | { |
---|
2345 | return xGetICost( m_pcEstBitsSbac->significantBits[ ui16CtxNumSig ][ uiSignificance ] ); |
---|
2346 | } |
---|
2347 | |
---|
2348 | /** Get the cost for a specific rate |
---|
2349 | * \param dRate rate of a bit |
---|
2350 | * \returns cost at the specific rate |
---|
2351 | */ |
---|
2352 | __inline Double TComTrQuant::xGetICost ( Double dRate ) const |
---|
2353 | { |
---|
2354 | return m_dLambda * dRate; |
---|
2355 | } |
---|
2356 | |
---|
2357 | /** Get the cost of an equal probable bit |
---|
2358 | * \returns cost of equal probable bit |
---|
2359 | */ |
---|
2360 | __inline Double TComTrQuant::xGetIEPRate ( ) const |
---|
2361 | { |
---|
2362 | return 32768; |
---|
2363 | } |
---|
2364 | |
---|
2365 | /** Context derivation process of coeff_abs_significant_flag |
---|
2366 | * \param uiSigCoeffGroupFlag significance map of L1 |
---|
2367 | * \param uiBlkX column of current scan position |
---|
2368 | * \param uiBlkY row of current scan position |
---|
2369 | * \param uiLog2BlkSize log2 value of block size |
---|
2370 | * \returns ctxInc for current scan position |
---|
2371 | */ |
---|
2372 | UInt TComTrQuant::getSigCoeffGroupCtxInc ( const UInt* uiSigCoeffGroupFlag, |
---|
2373 | const UInt uiCGPosX, |
---|
2374 | const UInt uiCGPosY, |
---|
2375 | Int width, Int height) |
---|
2376 | { |
---|
2377 | UInt uiRight = 0; |
---|
2378 | UInt uiLower = 0; |
---|
2379 | |
---|
2380 | width >>= 2; |
---|
2381 | height >>= 2; |
---|
2382 | if( uiCGPosX < width - 1 ) |
---|
2383 | { |
---|
2384 | uiRight = (uiSigCoeffGroupFlag[ uiCGPosY * width + uiCGPosX + 1 ] != 0); |
---|
2385 | } |
---|
2386 | if (uiCGPosY < height - 1 ) |
---|
2387 | { |
---|
2388 | uiLower = (uiSigCoeffGroupFlag[ (uiCGPosY + 1 ) * width + uiCGPosX ] != 0); |
---|
2389 | } |
---|
2390 | return (uiRight || uiLower); |
---|
2391 | |
---|
2392 | } |
---|
2393 | /** set quantized matrix coefficient for encode |
---|
2394 | * \param scalingList quantaized matrix address |
---|
2395 | */ |
---|
2396 | Void TComTrQuant::setScalingList(TComScalingList *scalingList) |
---|
2397 | { |
---|
2398 | UInt size,list; |
---|
2399 | UInt qp; |
---|
2400 | |
---|
2401 | for(size=0;size<SCALING_LIST_SIZE_NUM;size++) |
---|
2402 | { |
---|
2403 | for(list = 0; list < g_scalingListNum[size]; list++) |
---|
2404 | { |
---|
2405 | for(qp=0;qp<SCALING_LIST_REM_NUM;qp++) |
---|
2406 | { |
---|
2407 | xSetScalingListEnc(scalingList,list,size,qp); |
---|
2408 | xSetScalingListDec(scalingList,list,size,qp); |
---|
2409 | setErrScaleCoeff(list,size,qp); |
---|
2410 | } |
---|
2411 | } |
---|
2412 | } |
---|
2413 | } |
---|
2414 | /** set quantized matrix coefficient for decode |
---|
2415 | * \param scalingList quantaized matrix address |
---|
2416 | */ |
---|
2417 | Void TComTrQuant::setScalingListDec(TComScalingList *scalingList) |
---|
2418 | { |
---|
2419 | UInt size,list; |
---|
2420 | UInt qp; |
---|
2421 | |
---|
2422 | for(size=0;size<SCALING_LIST_SIZE_NUM;size++) |
---|
2423 | { |
---|
2424 | for(list = 0; list < g_scalingListNum[size]; list++) |
---|
2425 | { |
---|
2426 | for(qp=0;qp<SCALING_LIST_REM_NUM;qp++) |
---|
2427 | { |
---|
2428 | xSetScalingListDec(scalingList,list,size,qp); |
---|
2429 | } |
---|
2430 | } |
---|
2431 | } |
---|
2432 | } |
---|
2433 | /** set error scale coefficients |
---|
2434 | * \param list List ID |
---|
2435 | * \param uiSize Size |
---|
2436 | * \param uiQP Quantization parameter |
---|
2437 | */ |
---|
2438 | Void TComTrQuant::setErrScaleCoeff(UInt list,UInt size, UInt qp) |
---|
2439 | { |
---|
2440 | |
---|
2441 | UInt uiLog2TrSize = g_aucConvertToBit[ g_scalingListSizeX[size] ] + 2; |
---|
2442 | Int bitDepth = (size < SCALING_LIST_32x32 && list != 0 && list != 3) ? g_bitDepthC : g_bitDepthY; |
---|
2443 | Int iTransformShift = MAX_TR_DYNAMIC_RANGE - bitDepth - uiLog2TrSize; // Represents scaling through forward transform |
---|
2444 | |
---|
2445 | UInt i,uiMaxNumCoeff = g_scalingListSize[size]; |
---|
2446 | Int *piQuantcoeff; |
---|
2447 | Double *pdErrScale; |
---|
2448 | piQuantcoeff = getQuantCoeff(list, qp,size); |
---|
2449 | pdErrScale = getErrScaleCoeff(list, size, qp); |
---|
2450 | |
---|
2451 | Double dErrScale = (Double)(1<<SCALE_BITS); // Compensate for scaling of bitcount in Lagrange cost function |
---|
2452 | dErrScale = dErrScale*pow(2.0,-2.0*iTransformShift); // Compensate for scaling through forward transform |
---|
2453 | for(i=0;i<uiMaxNumCoeff;i++) |
---|
2454 | { |
---|
2455 | pdErrScale[i] = dErrScale / piQuantcoeff[i] / piQuantcoeff[i] / (1<<DISTORTION_PRECISION_ADJUSTMENT(2*(bitDepth-8))); |
---|
2456 | } |
---|
2457 | } |
---|
2458 | |
---|
2459 | /** set quantized matrix coefficient for encode |
---|
2460 | * \param scalingList quantaized matrix address |
---|
2461 | * \param listId List index |
---|
2462 | * \param sizeId size index |
---|
2463 | * \param uiQP Quantization parameter |
---|
2464 | */ |
---|
2465 | Void TComTrQuant::xSetScalingListEnc(TComScalingList *scalingList, UInt listId, UInt sizeId, UInt qp) |
---|
2466 | { |
---|
2467 | UInt width = g_scalingListSizeX[sizeId]; |
---|
2468 | UInt height = g_scalingListSizeX[sizeId]; |
---|
2469 | UInt ratio = g_scalingListSizeX[sizeId]/min(MAX_MATRIX_SIZE_NUM,(Int)g_scalingListSizeX[sizeId]); |
---|
2470 | Int *quantcoeff; |
---|
2471 | Int *coeff = scalingList->getScalingListAddress(sizeId,listId); |
---|
2472 | quantcoeff = getQuantCoeff(listId, qp, sizeId); |
---|
2473 | |
---|
2474 | processScalingListEnc(coeff,quantcoeff,g_quantScales[qp]<<4,height,width,ratio,min(MAX_MATRIX_SIZE_NUM,(Int)g_scalingListSizeX[sizeId]),scalingList->getScalingListDC(sizeId,listId)); |
---|
2475 | } |
---|
2476 | /** set quantized matrix coefficient for decode |
---|
2477 | * \param scalingList quantaized matrix address |
---|
2478 | * \param list List index |
---|
2479 | * \param size size index |
---|
2480 | * \param uiQP Quantization parameter |
---|
2481 | */ |
---|
2482 | Void TComTrQuant::xSetScalingListDec(TComScalingList *scalingList, UInt listId, UInt sizeId, UInt qp) |
---|
2483 | { |
---|
2484 | UInt width = g_scalingListSizeX[sizeId]; |
---|
2485 | UInt height = g_scalingListSizeX[sizeId]; |
---|
2486 | UInt ratio = g_scalingListSizeX[sizeId]/min(MAX_MATRIX_SIZE_NUM,(Int)g_scalingListSizeX[sizeId]); |
---|
2487 | Int *dequantcoeff; |
---|
2488 | Int *coeff = scalingList->getScalingListAddress(sizeId,listId); |
---|
2489 | |
---|
2490 | dequantcoeff = getDequantCoeff(listId, qp, sizeId); |
---|
2491 | processScalingListDec(coeff,dequantcoeff,g_invQuantScales[qp],height,width,ratio,min(MAX_MATRIX_SIZE_NUM,(Int)g_scalingListSizeX[sizeId]),scalingList->getScalingListDC(sizeId,listId)); |
---|
2492 | } |
---|
2493 | |
---|
2494 | /** set flat matrix value to quantized coefficient |
---|
2495 | */ |
---|
2496 | #if IL_SL_SIGNALLING_N0371 |
---|
2497 | Void TComTrQuant::setFlatScalingList( UInt m_layerId ) |
---|
2498 | #else |
---|
2499 | Void TComTrQuant::setFlatScalingList() |
---|
2500 | #endif |
---|
2501 | { |
---|
2502 | UInt size,list; |
---|
2503 | UInt qp; |
---|
2504 | |
---|
2505 | for(size=0;size<SCALING_LIST_SIZE_NUM;size++) |
---|
2506 | { |
---|
2507 | for(list = 0; list < g_scalingListNum[size]; list++) |
---|
2508 | { |
---|
2509 | #if IL_SL_SIGNALLING_N0371 |
---|
2510 | ref_scalingListDC[m_layerId][size][list] = SCALING_LIST_DC; |
---|
2511 | for(UInt i=0; i<MAX_MATRIX_COEF_NUM; i++) |
---|
2512 | { |
---|
2513 | ref_scalingListCoef[m_layerId][size][list][i] = SCALING_LIST_DC; |
---|
2514 | } |
---|
2515 | #endif |
---|
2516 | |
---|
2517 | for(qp=0;qp<SCALING_LIST_REM_NUM;qp++) |
---|
2518 | { |
---|
2519 | xsetFlatScalingList(list,size,qp); |
---|
2520 | setErrScaleCoeff(list,size,qp); |
---|
2521 | } |
---|
2522 | } |
---|
2523 | } |
---|
2524 | } |
---|
2525 | |
---|
2526 | /** set flat matrix value to quantized coefficient |
---|
2527 | * \param list List ID |
---|
2528 | * \param uiQP Quantization parameter |
---|
2529 | * \param uiSize Size |
---|
2530 | */ |
---|
2531 | Void TComTrQuant::xsetFlatScalingList(UInt list, UInt size, UInt qp) |
---|
2532 | { |
---|
2533 | UInt i,num = g_scalingListSize[size]; |
---|
2534 | Int *quantcoeff; |
---|
2535 | Int *dequantcoeff; |
---|
2536 | Int quantScales = g_quantScales[qp]; |
---|
2537 | Int invQuantScales = g_invQuantScales[qp]<<4; |
---|
2538 | |
---|
2539 | quantcoeff = getQuantCoeff(list, qp, size); |
---|
2540 | dequantcoeff = getDequantCoeff(list, qp, size); |
---|
2541 | |
---|
2542 | for(i=0;i<num;i++) |
---|
2543 | { |
---|
2544 | *quantcoeff++ = quantScales; |
---|
2545 | *dequantcoeff++ = invQuantScales; |
---|
2546 | } |
---|
2547 | } |
---|
2548 | |
---|
2549 | /** set quantized matrix coefficient for encode |
---|
2550 | * \param coeff quantaized matrix address |
---|
2551 | * \param quantcoeff quantaized matrix address |
---|
2552 | * \param quantScales Q(QP%6) |
---|
2553 | * \param height height |
---|
2554 | * \param width width |
---|
2555 | * \param ratio ratio for upscale |
---|
2556 | * \param sizuNum matrix size |
---|
2557 | * \param dc dc parameter |
---|
2558 | */ |
---|
2559 | Void TComTrQuant::processScalingListEnc( Int *coeff, Int *quantcoeff, Int quantScales, UInt height, UInt width, UInt ratio, Int sizuNum, UInt dc) |
---|
2560 | { |
---|
2561 | Int nsqth = (height < width) ? 4: 1; //height ratio for NSQT |
---|
2562 | Int nsqtw = (width < height) ? 4: 1; //width ratio for NSQT |
---|
2563 | for(UInt j=0;j<height;j++) |
---|
2564 | { |
---|
2565 | for(UInt i=0;i<width;i++) |
---|
2566 | { |
---|
2567 | quantcoeff[j*width + i] = quantScales / coeff[sizuNum * (j * nsqth / ratio) + i * nsqtw /ratio]; |
---|
2568 | } |
---|
2569 | } |
---|
2570 | if(ratio > 1) |
---|
2571 | { |
---|
2572 | quantcoeff[0] = quantScales / dc; |
---|
2573 | } |
---|
2574 | } |
---|
2575 | /** set quantized matrix coefficient for decode |
---|
2576 | * \param coeff quantaized matrix address |
---|
2577 | * \param dequantcoeff quantaized matrix address |
---|
2578 | * \param invQuantScales IQ(QP%6)) |
---|
2579 | * \param height height |
---|
2580 | * \param width width |
---|
2581 | * \param ratio ratio for upscale |
---|
2582 | * \param sizuNum matrix size |
---|
2583 | * \param dc dc parameter |
---|
2584 | */ |
---|
2585 | Void TComTrQuant::processScalingListDec( Int *coeff, Int *dequantcoeff, Int invQuantScales, UInt height, UInt width, UInt ratio, Int sizuNum, UInt dc) |
---|
2586 | { |
---|
2587 | for(UInt j=0;j<height;j++) |
---|
2588 | { |
---|
2589 | for(UInt i=0;i<width;i++) |
---|
2590 | { |
---|
2591 | dequantcoeff[j*width + i] = invQuantScales * coeff[sizuNum * (j / ratio) + i / ratio]; |
---|
2592 | } |
---|
2593 | } |
---|
2594 | if(ratio > 1) |
---|
2595 | { |
---|
2596 | dequantcoeff[0] = invQuantScales * dc; |
---|
2597 | } |
---|
2598 | } |
---|
2599 | |
---|
2600 | /** initialization process of scaling list array |
---|
2601 | */ |
---|
2602 | Void TComTrQuant::initScalingList() |
---|
2603 | { |
---|
2604 | for(UInt sizeId = 0; sizeId < SCALING_LIST_SIZE_NUM; sizeId++) |
---|
2605 | { |
---|
2606 | for(UInt listId = 0; listId < g_scalingListNum[sizeId]; listId++) |
---|
2607 | { |
---|
2608 | for(UInt qp = 0; qp < SCALING_LIST_REM_NUM; qp++) |
---|
2609 | { |
---|
2610 | m_quantCoef [sizeId][listId][qp] = new Int [g_scalingListSize[sizeId]]; |
---|
2611 | m_dequantCoef [sizeId][listId][qp] = new Int [g_scalingListSize[sizeId]]; |
---|
2612 | m_errScale [sizeId][listId][qp] = new Double [g_scalingListSize[sizeId]]; |
---|
2613 | } |
---|
2614 | } |
---|
2615 | } |
---|
2616 | // alias list [1] as [3]. |
---|
2617 | for(UInt qp = 0; qp < SCALING_LIST_REM_NUM; qp++) |
---|
2618 | { |
---|
2619 | m_quantCoef [SCALING_LIST_32x32][3][qp] = m_quantCoef [SCALING_LIST_32x32][1][qp]; |
---|
2620 | m_dequantCoef [SCALING_LIST_32x32][3][qp] = m_dequantCoef [SCALING_LIST_32x32][1][qp]; |
---|
2621 | m_errScale [SCALING_LIST_32x32][3][qp] = m_errScale [SCALING_LIST_32x32][1][qp]; |
---|
2622 | } |
---|
2623 | } |
---|
2624 | /** destroy quantization matrix array |
---|
2625 | */ |
---|
2626 | Void TComTrQuant::destroyScalingList() |
---|
2627 | { |
---|
2628 | for(UInt sizeId = 0; sizeId < SCALING_LIST_SIZE_NUM; sizeId++) |
---|
2629 | { |
---|
2630 | for(UInt listId = 0; listId < g_scalingListNum[sizeId]; listId++) |
---|
2631 | { |
---|
2632 | for(UInt qp = 0; qp < SCALING_LIST_REM_NUM; qp++) |
---|
2633 | { |
---|
2634 | if(m_quantCoef [sizeId][listId][qp]) delete [] m_quantCoef [sizeId][listId][qp]; |
---|
2635 | if(m_dequantCoef [sizeId][listId][qp]) delete [] m_dequantCoef [sizeId][listId][qp]; |
---|
2636 | if(m_errScale [sizeId][listId][qp]) delete [] m_errScale [sizeId][listId][qp]; |
---|
2637 | } |
---|
2638 | } |
---|
2639 | } |
---|
2640 | } |
---|
2641 | |
---|
2642 | //! \} |
---|