这篇博客讲HM代码中xPredIntraPlanar、xPredIntraAng、xDCPredFiltering三个函数,有关帧内预测的理论知识看 https://blog.csdn.net/shayashi/article/details/82877875 。
xPredIntraPlanar函数是对Planar模式预测的函数,Planar模式对应0,该模式的预测采用同双线性差值方式,代码如下:
Void TComPrediction::xPredIntraPlanar( const Pel* pSrc, Int srcStride, Pel* rpDst, Int dstStride, UInt width, UInt height )
{
assert(width <= height);
Int leftColumn[MAX_CU_SIZE+1], topRow[MAX_CU_SIZE+1], bottomRow[MAX_CU_SIZE], rightColumn[MAX_CU_SIZE];//参考像素
UInt shift1Dhor = g_aucConvertToBit[ width ] + 2;//水平移动位数
UInt shift1Dver = g_aucConvertToBit[ height ] + 2;//竖直移动位数
//g_aucConvertToBit是对应的bit数 [4]==0 [8]==1 [16]==2 [32]==3 [64]==4 other -1
// Get left and above reference column and row获取参考像素
for(Int k=0;k<width+1;k++)
{
topRow[k] = pSrc[k-srcStride];//srcStride == 2 * width + 1
}
for (Int k=0; k < height+1; k++)
{
leftColumn[k] = pSrc[k*srcStride-1];
}
// Prepare intermediate variables used in interpolation
Int bottomLeft = leftColumn[height];
Int topRight = topRow[width];
//bottomRow[k]= leftColumn[height]-topRow[k],且将topRow的数值乘height
for(Int k=0;k<width;k++)
{
bottomRow[k] = bottomLeft - topRow[k];
topRow[k] <<= shift1Dver;
}
//rightColumn[k]=topRow[width]-leftColumn[k],将leftColumn的数值乘width
for(Int k=0;k<height;k++)
{
rightColumn[k] = topRight - leftColumn[k];
leftColumn[k] <<= shift1Dhor;
}
const UInt topRowShift = 0;
// 生成预测信号
//双线性插值就是像素对应的上、下和左、右参考像素成比例的和。
//horPred(x,y) = (width - x) * leftColumn[y] + x * topRight
//vertPred(x,y) = (width - y) * topRight[y] + y * bottomLeft
//rpDst(x,y) = (horPred(x,y) + vertPred(x,y)) >> (shift1Dhor+1)
for (Int y=0;y<height;y++)
{
Int horPred = leftColumn[y] + width;
for (Int x=0;x<width;x++)
{
horPred += rightColumn[y];
topRow[x] += bottomRow[x];
Int vertPred = ((topRow[x] + topRowShift)>>topRowShift);
rpDst[y*dstStride+x] = ( horPred + vertPred ) >> (shift1Dhor+1);//dstStride == width
}
}
}
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12
- 13
- 14
- 15
- 16
- 17
- 18
- 19
- 20
- 21
- 22
- 23
- 24
- 25
- 26
- 27
- 28
- 29
- 30
- 31
- 32
- 33
- 34
- 35
- 36
- 37
- 38
- 39
- 40
- 41
- 42
- 43
- 44
- 45
- 46
- 47
- 48
- 49
- 50
- 51
- 52
- 53
- 54
- 55
xPredIntraAng函数是对DC模式和角度模式进行预测的函数:
Void TComPrediction::xPredIntraAng( Int bitDepth,
const Pel* pSrc, Int srcStride,
Pel* pTrueDst, Int dstStrideTrue,
UInt uiWidth, UInt uiHeight, ChannelType channelType,
UInt dirMode, const Bool bEnableEdgeFilters
)
{
Int width=Int(uiWidth);
Int height=Int(uiHeight);
// Map the mode index to main prediction direction and angle
assert( dirMode != PLANAR_IDX ); //no planar
const Bool modeDC = dirMode==DC_IDX;
// Do the DC prediction DC模式就是对块中的每一个像素设置为平均值
if (modeDC)
{
const Pel dcval = predIntraGetPredValDC(pSrc, srcStride, width, height);
for (Int y=height;y>0;y--, pTrueDst+=dstStrideTrue)
{
for (Int x=0; x<width;) // width is always a multiple of 4.
{
pTrueDst[x++] = dcval;
}
}
}
else // Do angular predictions
{
const Bool bIsModeVer = (dirMode >= 18);//是否是竖直模式
const Int intraPredAngleMode = (bIsModeVer) ? (Int)dirMode - VER_IDX : -((Int)dirMode - HOR_IDX);//竖直模式:mode - 26 ;水平模式:10-mode
const Int absAngMode = abs(intraPredAngleMode);
const Int signAng = intraPredAngleMode < 0 ? -1 : 1;//sin正负
const Bool edgeFilter = bEnableEdgeFilters && isLuma(channelType) && (width <= MAXIMUM_INTRA_FILTERED_WIDTH) && (height <= MAXIMUM_INTRA_FILTERED_HEIGHT);//当块大小小于等于16x16且是亮度预测且bEnableEdgeFilters ==true进行边缘滤波
// Set bitshifts and scale the angle parameter to block size
static const Int angTable[9] = {0, 2, 5, 9, 13, 17, 21, 26, 32};//mode偏移量绝对值
static const Int invAngTable[9] = {0, 4096, 1638, 910, 630, 482, 390, 315, 256}; // (256 * 32) / Angle
Int invAngle = invAngTable[absAngMode];
Int absAng = angTable[absAngMode];
Int intraPredAngle = signAng * absAng;
Pel* refMain;
Pel* refSide;
Pel refAbove[2*MAX_CU_SIZE+1];
Pel refLeft[2*MAX_CU_SIZE+1];
// Initialize the Main and Left reference array.映射
if (intraPredAngle < 0)
{
const Int refMainOffsetPreScale = (bIsModeVer ? height : width ) - 1;
const Int refMainOffset = height - 1;
for (Int x=0;x<width+1;x++)
{
refAbove[x+refMainOffset] = pSrc[x-srcStride-1];
}
for (Int y=0;y<height+1;y++)
{
refLeft[y+refMainOffset] = pSrc[(y-1)*srcStride-1];
}
refMain = (bIsModeVer ? refAbove : refLeft) + refMainOffset;
refSide = (bIsModeVer ? refLeft : refAbove) + refMainOffset;
// Extend the Main reference to the left.
Int invAngleSum = 128; // rounding for (shift by 8)
for (Int k=-1; k>(refMainOffsetPreScale+1)*intraPredAngle>>5; k--)
{
invAngleSum += invAngle;
refMain[k] = refSide[invAngleSum>>8];
}
}
else// (2 <= mode <= 10) || (26 <= mode <= 34)
{
for (Int x=0;x<2*width+1;x++)
{
refAbove[x] = pSrc[x-srcStride-1];//srcStride == 2*width+1
}
for (Int y=0;y<2*height+1;y++)
{
refLeft[y] = pSrc[(y-1)*srcStride-1];
}
refMain = bIsModeVer ? refAbove : refLeft ;//主参考像素
refSide = bIsModeVer ? refLeft : refAbove;
}
// swap width/height if we are doing a horizontal mode:
Pel tempArray[MAX_CU_SIZE*MAX_CU_SIZE];
const Int dstStride = bIsModeVer ? dstStrideTrue : MAX_CU_SIZE;
Pel *pDst = bIsModeVer ? pTrueDst : tempArray;
if (!bIsModeVer)
{
std::swap(width, height);
}
if (intraPredAngle == 0) // pure vertical or pure horizontal
{
for (Int y=0;y<height;y++)
{
for (Int x=0;x<width;x++)
{
pDst[y*dstStride+x] = refMain[x+1];
}
}
if (edgeFilter)
{
for (Int y=0;y<height;y++)
{//边缘滤波 当前像素值+(对应参考像素值和左上角参考像素和的1/2)
pDst[y*dstStride] = Clip3 (0, ((1 << bitDepth) - 1), pDst[y*dstStride] + (( refSide[y+1] - refSide[0] ) >> 1) );
}
}
}
else
{
Pel *pDsty=pDst;
for (Int y=0, deltaPos=intraPredAngle; y<height; y++, deltaPos+=intraPredAngle, pDsty+=dstStride)
{
const Int deltaInt = deltaPos >> 5;//坐标整数部分
const Int deltaFract = deltaPos & (32 - 1);//坐标分数部分
//pDsty[x] = ((32 - deltaFract) * refMain[x + deltaInt+1] + deltaFract * refMain[x + deltaInt+2] + 16) >> 5
if (deltaFract)
{
// Do linear filtering
const Pel *pRM=refMain+deltaInt+1;
Int lastRefMainPel=*pRM++;
for (Int x=0;x<width;pRM++,x++)
{
Int thisRefMainPel=*pRM;
pDsty[x+0] = (Pel) ( ((32-deltaFract)lastRefMainPel + deltaFractthisRefMainPel +16) >> 5 );
lastRefMainPel=thisRefMainPel;
}
}
else
{
// Just copy the integer samples
for (Int x=0;x<width; x++)
{
pDsty[x] = refMain[x+deltaInt+1];
}
}
}
}
// Flip the block if this is the horizontal mode水平模式翻转
if (!bIsModeVer)
{
for (Int y=0; y<height; y++)
{
for (Int x=0; x<width; x++)
{
pTrueDst[x*dstStrideTrue] = pDst[x];
}
pTrueDst++;
pDst+=dstStride;
}
}
}
}
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12
- 13
- 14
- 15
- 16
- 17
- 18
- 19
- 20
- 21
- 22
- 23
- 24
- 25
- 26
- 27
- 28
- 29
- 30
- 31
- 32
- 33
- 34
- 35
- 36
- 37
- 38
- 39
- 40
- 41
- 42
- 43
- 44
- 45
- 46
- 47
- 48
- 49
- 50
- 51
- 52
- 53
- 54
- 55
- 56
- 57
- 58
- 59
- 60
- 61
- 62
- 63
- 64
- 65
- 66
- 67
- 68
- 69
- 70
- 71
- 72
- 73
- 74
- 75
- 76
- 77
- 78
- 79
- 80
- 81
- 82
- 83
- 84
- 85
- 86
- 87
- 88
- 89
- 90
- 91
- 92
- 93
- 94
- 95
- 96
- 97
- 98
- 99
- 100
- 101
- 102
- 103
- 104
- 105
- 106
- 107
- 108
- 109
- 110
- 111
- 112
- 113
- 114
- 115
- 116
- 117
- 118
- 119
- 120
- 121
- 122
- 123
- 124
- 125
- 126
- 127
- 128
- 129
- 130
- 131
- 132
- 133
- 134
- 135
- 136
- 137
- 138
- 139
- 140
- 141
- 142
- 143
- 144
- 145
- 146
- 147
- 148
- 149
- 150
- 151
- 152
- 153
- 154
- 155
- 156
- 157
- 158
- 159
- 160
xDCPredFiltering函数是对DC模式进行边缘滤波的函数,对第一行和第一列进行滤波:
Void TComPrediction::xDCPredFiltering( const Pel* pSrc, Int iSrcStride, Pel* pDst, Int iDstStride, Int iWidth, Int iHeight, ChannelType channelType ) { Int x, y, iDstStride2, iSrcStride2; //亮度模式且小于等于16x16才进行滤波 if (isLuma(channelType) && (iWidth <= MAXIMUM_INTRA_FILTERED_WIDTH) && (iHeight <= MAXIMUM_INTRA_FILTERED_HEIGHT)) { //top-left pDst[0] = 当前像素值和左边像素上边参考像素进行2 1 1滤波 pDst[0] = (Pel)((pSrc[-iSrcStride] + pSrc[-1] + 2 * pDst[0] + 2) >> 2);
//top row (vertical filter)第一行像素值等于当前像素和上边参考像素进行3 1滤波 for ( x = 1; x < iWidth; x++ ) { pDst[x] = (Pel)((pSrc[x - iSrcStride] + 3 * pDst[x] + 2) >> 2); } //left column (horizontal filter)第一列像素值等于当前像素和左边参考像素进行3 1滤波 for ( y = 1, iDstStride2 = iDstStride, iSrcStride2 = iSrcStride-1; y < iHeight; y++, iDstStride2+=iDstStride, iSrcStride2+=iSrcStride ) { pDst[iDstStride2] = (Pel)((pSrc[iSrcStride2] + 3 * pDst[iDstStride2] + 2) >> 2); }
}
return;
}
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12
- 13
- 14
- 15
- 16
- 17
- 18
- 19
- 20
- 21
- 22
- 23
- 24