博客
关于我
强烈建议你试试无所不能的chatGPT,快点击我
x265-1.7版本-common/slice.h注释
阅读量:2190 次
发布时间:2019-05-02

本文共 10862 字,大约阅读时间需要 36 分钟。

注:问号以及未注释部分 会在x265-1.8版本内更新

 

/***************************************************************************** * Copyright (C) 2015 x265 project * * Authors: Steve Borho 
* * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; either version 2 of the License, or * (at your option) any later version. * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this program; if not, write to the Free Software * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA. * * This program is also available under a commercial proprietary license. * For more information, contact us at license @ x265.com. *****************************************************************************/#ifndef X265_SLICE_H#define X265_SLICE_H#include "common.h"namespace x265 {// private namespaceclass Frame;class PicList;class MotionReference;enum SliceType{ B_SLICE, P_SLICE, I_SLICE};struct RPS//RPS(reference picture set) 在slice中被引用{ int numberOfPictures; //参考帧列表中的参考数目 int numberOfNegativePictures;//前向参考个数 int numberOfPositivePictures;//后向参考帧个数 int poc[MAX_NUM_REF_PICS];//存储当前帧前面已经编码可参考的参考帧 初始化为0 (按照编码顺序排序 倒序) int deltaPOC[MAX_NUM_REF_PICS];//参考帧poc减去存储当前帧poc的值 按照 远 近 (当前帧) 近 远的次序排序 bool bUsed[MAX_NUM_REF_PICS];//记录参考帧列表对应位置帧是否可用 (如果当前帧是关键帧,则全为不可用) //如编码次序:0 4 2 1 3 8 6 5 7 12 10 9 11 //则编码11时, poc存储为: 10 12 6 8 //deltapoc 存储为: -5 -3 -1 12 (排序为 6 8 10 12) RPS()//初始化 : numberOfPictures(0) , numberOfNegativePictures(0) , numberOfPositivePictures(0) { memset(deltaPOC, 0, sizeof(deltaPOC)); memset(poc, 0, sizeof(poc)); memset(bUsed, 0, sizeof(bUsed)); } /** 函数功能 : 将RPS列表中的deltaPOC和bused 按照 远 近 (当前帧) 近 远的次序排序 /* 调用范围 : 只在DPB::computeRPS函数中被调用 * 返回值 : null **/ void sortDeltaPOC();};namespace Profile { enum Name { NONE = 0, MAIN = 1, MAIN10 = 2, MAINSTILLPICTURE = 3, MAINREXT = 4, HIGHTHROUGHPUTREXT = 5 };}namespace Level { enum Tier { MAIN = 0, HIGH = 1, }; enum Name { NONE = 0, LEVEL1 = 30, LEVEL2 = 60, LEVEL2_1 = 63, LEVEL3 = 90, LEVEL3_1 = 93, LEVEL4 = 120, LEVEL4_1 = 123, LEVEL5 = 150, LEVEL5_1 = 153, LEVEL5_2 = 156, LEVEL6 = 180, LEVEL6_1 = 183, LEVEL6_2 = 186, LEVEL8_5 = 255, };}struct ProfileTierLevel{ bool tierFlag; bool progressiveSourceFlag; bool interlacedSourceFlag; bool nonPackedConstraintFlag; bool frameOnlyConstraintFlag; bool profileCompatibilityFlag[32]; bool intraConstraintFlag; bool lowerBitRateConstraintFlag; int profileIdc; int levelIdc; uint32_t minCrForLevel; uint32_t maxLumaSrForLevel; uint32_t bitDepthConstraint; int chromaFormatConstraint;};struct HRDInfo{ uint32_t bitRateScale; uint32_t cpbSizeScale; uint32_t initialCpbRemovalDelayLength; uint32_t cpbRemovalDelayLength; uint32_t dpbOutputDelayLength; uint32_t bitRateValue; uint32_t cpbSizeValue; bool cbrFlag; HRDInfo() : bitRateScale(0) , cpbSizeScale(0) , initialCpbRemovalDelayLength(1) , cpbRemovalDelayLength(1) , dpbOutputDelayLength(1) , cbrFlag(false) { }};struct TimingInfo{ uint32_t numUnitsInTick; uint32_t timeScale;};struct VPS{ uint32_t maxTempSubLayers; uint32_t numReorderPics;//需要重排序的帧数 值:(param.bBPyramid && param.bframes > 1) ? 2 : !!param.bframes; uint32_t maxDecPicBuffering;//解码需要的buffer大小: 值 X265_MIN(MAX_NUM_REF, X265_MAX(vps.numReorderPics + 2, (uint32_t)param.maxNumReferences) + vps.numReorderPics); uint32_t maxLatencyIncrease; HRDInfo hrdParameters; ProfileTierLevel ptl;};struct Window{ bool bEnabled; int leftOffset; int rightOffset; int topOffset; int bottomOffset; Window() { bEnabled = false; }};struct VUI{ bool aspectRatioInfoPresentFlag; int aspectRatioIdc; int sarWidth; int sarHeight; bool overscanInfoPresentFlag; bool overscanAppropriateFlag; bool videoSignalTypePresentFlag; int videoFormat; bool videoFullRangeFlag; bool colourDescriptionPresentFlag; int colourPrimaries; int transferCharacteristics; int matrixCoefficients; bool chromaLocInfoPresentFlag; int chromaSampleLocTypeTopField; int chromaSampleLocTypeBottomField; Window defaultDisplayWindow; bool frameFieldInfoPresentFlag; bool fieldSeqFlag; bool hrdParametersPresentFlag; HRDInfo hrdParameters; TimingInfo timingInfo;};struct SPS{ int chromaFormatIdc; // use param uint32_t picWidthInLumaSamples; // use param uint32_t picHeightInLumaSamples; // use param uint32_t numCuInWidth; //有多少CTU列 uint32_t numCuInHeight; //有多少CTU行 uint32_t numCUsInFrame; //一帧CTU个数 uint32_t numPartitions; uint32_t numPartInCUSize; int log2MinCodingBlockSize; int log2DiffMaxMinCodingBlockSize; uint32_t quadtreeTULog2MaxSize; uint32_t quadtreeTULog2MinSize;//最小TU的值 (一般默认为4x4,值为2) uint32_t quadtreeTUMaxDepthInter; // use param uint32_t quadtreeTUMaxDepthIntra; // use param bool bUseSAO; // use param bool bUseAMP; // use param uint32_t maxAMPDepth; uint32_t maxTempSubLayers; // max number of Temporal Sub layers uint32_t maxDecPicBuffering; // 继承VPS值 解码需要的buffer大小: 值 X265_MIN(MAX_NUM_REF, X265_MAX(vps.numReorderPics + 2, (uint32_t)param.maxNumReferences) + vps.numReorderPics); these are dups of VPS values uint32_t maxLatencyIncrease; int numReorderPics; bool bUseStrongIntraSmoothing; // use param bool bTemporalMVPEnabled; Window conformanceWindow; VUI vuiParameters;};struct PPS{ uint32_t maxCuDQPDepth; int chromaQpOffset[2]; // use param bool bUseWeightPred; // 当前是否应用P帧加权预测use param bool bUseWeightedBiPred; // 当前是否应用B帧加权预测use param bool bUseDQP; bool bConstrainedIntraPred; // use param bool bTransquantBypassEnabled; // Indicates presence of cu_transquant_bypass_flag in CUs. bool bTransformSkipEnabled; // 是否应用transformSkip 在4x4才应用 use param bool bEntropyCodingSyncEnabled; // use param bool bSignHideEnabled; // use param bool bDeblockingFilterControlPresent; bool bPicDisableDeblockingFilter; int deblockingFilterBetaOffsetDiv2; int deblockingFilterTcOffsetDiv2;};struct WeightParam //存储加权帧的加权信息 用于slice类{ //加权为 w*ref + offset ref为参考帧所有像素值 //其中w为: inputWeight/(1<
0 && (inputWeight > 127)) { log2WeightDenom--; inputWeight >>= 1; } inputWeight = X265_MIN(inputWeight, 127); }};//设置WeightParam类数据#define SET_WEIGHT(w, b, s, d, o) \{ \ (w).inputWeight = (s); \ (w).log2WeightDenom = (d); \ (w).inputOffset = (o); \ (w).bPresentFlag = (b); \}class Slice{public: const SPS* m_sps; //指向encoder的SPS const PPS* m_pps; //指向encoder的PPS WeightParam m_weightPredTable[2][MAX_NUM_REF][3]; // 参考帧加权状态信息 (每个list的第一帧分析加权与否,其它不加权) [list][refIdx][0:Y, 1:U, 2:V] MotionReference (*m_mref)[MAX_NUM_REF + 1]; //指向FrameEncoder中对应的参考帧信息 RPS m_rps; NalUnitType m_nalUnitType; SliceType m_sliceType; //在DPB::prepareEncode函数确定slice类型:B_SLICE,P_SLICE,I_SLICE int m_sliceQp; int m_poc; //当前POC int m_lastIDR; bool m_bCheckLDC; // 目前x265: B帧 false 其它 ture 指的是low-delay mode??? TODO: is this necessary? bool m_sLFaseFlag; // 应用????????表示环路滤波是否可以越过当前slice的上边界和左边界 初始化为ture (只在prepareEncode计算过一次然后就写码流) loop filter boundary flag bool m_colFromL0Flag; // ?????用于时域MV预测的collocated图像来自参考图像列表List0还是List1 目前x265: B帧 false 其它 ture collocated picture from List0 or List1 flag uint32_t m_colRefIdx; // ?????用于时域MV预测的collocated图像的参考索引号 ,一直为0,没被修改过never modified //x265中没有应用长期参考帧 x265中List0 全为前向参考帧 List1全为后向参考帧 //List0 存储方式:按照距离当前帧由近到远排序 //List1 存储方式:按照距离当前帧由近到远排序 int m_numRefIdx[2];//分别为L0的最大帧数(配置的配置的参考帧最大数目与实际前向帧数取最小值) L1 的最大帧数(2或者1 与 实际后向帧数取最小值) 分别为前向帧和后向帧 Frame* m_refPicList[2][MAX_NUM_REF + 1];//存储List帧指针(数据存放在dpb中的m_picList中)setRefPicList函数设定 int m_refPOCList[2][MAX_NUM_REF + 1];//存储List帧poc setRefPicList函数设定 uint32_t m_maxNumMergeCand; //Merge选择的候选个数,默认值为3 use param uint32_t m_endCUAddr;//一帧中最后实际像素在帧中的4x4块标号+1 Slice() { m_lastIDR = 0; m_sLFaseFlag = true; m_numRefIdx[0] = m_numRefIdx[1] = 0; for (int i = 0; i < MAX_NUM_REF; i++) { m_refPicList[0][i] = NULL; m_refPicList[1][i] = NULL; m_refPOCList[0][i] = 0; m_refPOCList[1][i] = 0; } disableWeights(); } /** 函数功能 : 关闭当前帧的加权预测 /* 调用范围 : 只在Slice()、FrameEncoder::compressFrame()、weightAnalyse函数中被调用 * 返回值 : null **/ void disableWeights(); /** 函数功能 : 获取参考帧List 以及 POC /* 调用范围 : 只在DPB::prepareEncode函数中被调用 * \参数 picList : DPB m_picList * 返回值 : null **/ void setRefPicList(PicList& picList); const Frame* getRefPic(int list, int refIdx) const { return refIdx >= 0 ? m_refPicList[list][refIdx] : NULL; } bool getRapPicFlag() const { return m_nalUnitType == NAL_UNIT_CODED_SLICE_IDR_W_RADL || m_nalUnitType == NAL_UNIT_CODED_SLICE_CRA; } bool getIdrPicFlag() const { return m_nalUnitType == NAL_UNIT_CODED_SLICE_IDR_W_RADL; } bool isIRAP() const { return m_nalUnitType >= 16 && m_nalUnitType <= 23; }//判断是否关键帧:IRAP(Intra Random Access Point)三种IRAP IDR(Instantaneous Decoding Refresh) CRA(Clean Random Access) BLA(Broken Link Access) bool isIntra() const { return m_sliceType == I_SLICE; } bool isInterB() const { return m_sliceType == B_SLICE; } bool isInterP() const { return m_sliceType == P_SLICE; } /** 函数功能 : 返回一帧中最后实际像素在帧中的4x4块标号+1 /* 调用范围 : 只在Encoder::encode函数中被调用 * \参数 endCUAddr : (一帧CTU个数)*(在CTU中4x4的个数) * 返回值 : 返回一帧中最后实际像素在帧中的4x4块标号+1 **/ uint32_t realEndAddress(uint32_t endCUAddr) const;};}#endif // ifndef X265_SLICE_H

转载地址:http://fouub.baihongyu.com/

你可能感兴趣的文章
用 Recursive Neural Networks 得到分析树
查看>>
RNN的高级应用
查看>>
TensorFlow-7-TensorBoard Embedding可视化
查看>>
轻松看懂机器学习十大常用算法
查看>>
一个框架解决几乎所有机器学习问题
查看>>
特征工程怎么做
查看>>
机器学习算法应用中常用技巧-1
查看>>
机器学习算法应用中常用技巧-2
查看>>
通过一个kaggle实例学习解决机器学习问题
查看>>
决策树的python实现
查看>>
Sklearn 快速入门
查看>>
了解 Sklearn 的数据集
查看>>
用ARIMA模型做需求预测
查看>>
推荐系统
查看>>
TensorFlow-11-策略网络
查看>>
浅谈 GBDT
查看>>
如何选择优化器 optimizer
查看>>
一文了解强化学习
查看>>
CART 分类与回归树
查看>>
seq2seq 的 keras 实现
查看>>