MIP技術原理介紹:https://blog.csdn.net/BigDream123/article/details/104939645
MIP的相關函數代碼在estIntraPredLumaQT函數中首次被調用,如下代碼所示,其中
- initIntraMip函數用於對參考像素進行下采樣,爲MIP矩陣乘法準備輸入數據
- getNumModesMip函數根據當前塊的尺寸獲取不同的MIP模式數目
- predIntraMip函數進行矩陣運算,並進行上採樣插值,獲得整個塊的預測值
注意:MIP使用存在塊尺寸限制:寬高比不能超過1:8/8:1
//MIP允許使用標誌
const bool mipAllowed = sps.getUseMIP() && isLuma(partitioner.chType) && ((cu.lfnstIdx == 0) || allowLfnstWithMip(cu.firstPU->lumaSize()));
//MIP模式測試標誌:寬高比不能超過1:8/8:1
const bool testMip = mipAllowed && !(cu.lwidth() > (8 * cu.lheight()) || cu.lheight() > (8 * cu.lwidth()));
//MIP支持的最大塊尺寸爲64
const bool supportedMipBlkSize = pu.lwidth() <= MIP_MAX_WIDTH && pu.lheight() <= MIP_MAX_HEIGHT;
...
...
...
// 初始化Mip模式
initIntraPatternChType(cu, pu.Y());//獲取參考像素並濾波
initIntraMip(pu, pu.Y());
// getNumModesMip函數根據當前CU的塊尺寸獲取不同的Mip模式數量16 8 6
const int transpOff = getNumModesMip(pu.Y());
const int numModesFull = (transpOff << 1);//MIP模式總數分別是32 16 12
for (uint32_t uiModeFull = 0; uiModeFull < numModesFull; uiModeFull++)
{
//對於另一半MIP模式,其需要進行轉置
const bool isTransposed = (uiModeFull >= transpOff ? true : false);
const uint32_t uiMode = (isTransposed ? uiModeFull - transpOff : uiModeFull);
pu.mipTransposedFlag = isTransposed;
pu.intraDir[CHANNEL_TYPE_LUMA] = uiMode;
// Mip預測模式的函數入口,進行MIP預測
predIntraMip(COMPONENT_Y, piPred, pu);
// Use the min between SAD and HAD as the cost criterion
// SAD is scaled by 2 to align with the scaling of HAD
Distortion minSadHad =
std::min(distParamSad.distFunc(distParamSad) * 2, distParamHad.distFunc(distParamHad));
m_CABACEstimator->getCtx() = SubCtx(Ctx::MipFlag, ctxStartMipFlag);
uint64_t fracModeBits = xFracModeBitsIntra(pu, uiMode, CHANNEL_TYPE_LUMA);
double cost = double(minSadHad) + double(fracModeBits) * sqrtLambdaForFirstPass;
mipHadCost[uiModeFull] = cost;
DTRACE(g_trace_ctx, D_INTRA_COST, "IntraMIP: %u, %llu, %f (%d)\n", minSadHad, fracModeBits, cost,
uiModeFull);
// 更新候選列表,將最優的Mip模式加入到RDCost的候選列表中
updateCandList(ModeInfo(true, isTransposed, 0, NOT_INTRA_SUBPARTITIONS, uiMode), cost, uiRdModeList,
CandCostList, numModesForFullRD + 1);
updateCandList(ModeInfo(true, isTransposed, 0, NOT_INTRA_SUBPARTITIONS, uiMode),
0.8 * double(minSadHad), uiHadModeList, CandHadList, numHadCand);
}
MIP根據塊尺寸可以分爲以下三種情況:
塊尺寸 | MIP模式 | |
mipSizeId = 0 | 4x4 | 32 |
mipSizeId = 1 | 4xN、Nx4、8x8 | 16 |
mipSizeId = 2 | 其餘塊 | 12 |
其中,MIP的轉置標誌和MIP的模式號有關:
- mipSizdId = 0時,前16種模式的mipTransposedFlag= 0,後16種模式的mipTransposedFlag = 1
- mipSizeId = 1時,前8種模式的mipTransposedFlag= 0,後8種模式的mipTransposedFlag = 1
- mipSizeId = 2時,前6種模式的mipTransposedFlag= 0,後6種模式的mipTransposedFlag = 1
getNumModesMip函數代碼如下:
int getNumModesMip(const Size& block)
{
switch( getMipSizeId(block) )
{
case 0: return 16;
case 1: return 8;
case 2: return 6;
default: THROW( "Invalid mipSizeId" );
}
}
initIntraMip函數相關代碼參考:https://blog.csdn.net/BigDream123/article/details/107018446
predIntraMip函數相關代碼參考:https://blog.csdn.net/BigDream123/article/details/107019458