H265/hevc noun explanation -- CTU, CTB, Cu, CB, Tu, PU, TB, Pb, LCU, slice, tile, chroma, luma, I frame, B frame, P frame

H265/HEVC A term is used to explain

learn H265 When , A pile of nouns CTU,CTB,CU,CB,TU,PU,TB,PB,LCU,Slice,Tile,Chroma,Luma,I frame ,B frame ,P frame , Headache around , Write a blog to summarize , Try not to involve complex concepts

---

---

One 、I frame ,B frame ,P frame

  * I frame （ Frames are encoded within frames ,Intra-Picture） Is a complete compressed image ,I Frames can be decompressed to get a complete image （ The largest amount of data ）;
  *P frame （ Pre and post prediction coding frame , also called Predictive-Picture） When the back image is with I When the frame difference is large , It will be recorded as P frame ,P Frame record with the previous P Frame or I The difference between frames （ Probably ）;
  *B frame （ Bidirectional predictive interpolation coding frames ,Bi-directional interpolated prediction frame）, According to the next frame and the previous frame , Record a motion vector （ Minimum amount of data ）;

 Ф The pictures that make up the video in chronological order are usually encoded into I BB…B P B…BP . … I BBBB P IBB, in other words , In chronological order , I The frame and P The image in the middle of the frame , Two P The images between frames will be compressed into B frame ;
When decoding , You need to decode I frame （ Full image ）, According to I Frame decoding P frame （ Stored with the previous I/P The difference between frames ）, According to I The frame and P Frame decoding B frame （ Sports information ）, To restore this video ;

   therefore ,I Frames are indispensable ,P Frame recovery requires I frame ,B Frame recovery also requires I The frame of P frame , So in the actual video stream The transmission sequence in is I PBB…B P B…BP . … I PBB IPBB, That is, first transmit this segment I,P, Then transmit them B; Empathy , If the current receiver gets B perhaps P, Did not get the front I frame , Then you can't answer the code , Can only wait for the next group I frame ; That's why when we watch videos , Many videos cannot drag the progress bar to a certain point in time , Even some short videos can't drag the progress bar , It's because the image of their progress point is not I frame , Therefore, decoding cannot be started ;

Two 、Slice and Tilie

  * In the process of encoding and decoding each frame of image , In order to improve efficiency ,H265 Divide an image into multiple slice perhaps Tile, Each of a frame of image Slice or Tile Independent codec , The mutual direct dependence is not high , So as to achieve the effect of quickly encoding and decoding a burst of images ; and slice and Tile The difference lies in the way of division ;

  As shown in the figure , Blue and green represent different Slice：

  As shown in the figure , Divide an image into different Tile：

   in other words ,slice The way of division is banded ,Tile Is a rectangle

3、 ... and 、Chroma,Luma

  * Luma（ brightness ）： YUV Format pictures Y component ;
  * Chroma（ chroma ）：YUV Format pictures U,V component ;

Four 、CTU,CU,CTB,CB,TU,PU,TB,PB

  * CTU（coding tree unit）：H265 Divide a frame of image into several rectangular blocks , The size is 16x16、 32x32 and 64x64, These blocks and the corresponding coding information are CTU;
  * CU（coding unit）：H265 Will be part of CTU Make a more detailed division , Divide into smaller matrix blocks , These blocks and the corresponding coding information are CU;

  * CTB（coding tree block）： Because an image is usually made of Y,U,V Three components , So a CTU It contains three CTB(CTB-Y,CTB-U,CTB-V);
  * CB（coding block）： and CTB Empathy ,CU Also by Y,U,V Three components make up , So a CU It contains three CB(CB-Y,CB-U,CB-V);

  * PU（prediction unit）： Storage CU Prediction mode information between （ A kind of encoding and decoding information ）;
  * PB（coding block）： Because an image is usually made of Y,U,V Three components , So a PU It contains three PB(PB-Y,PB-U,PB-V);

  * TU（transform unit）： Storage CU Transform the quantized information （ A kind of encoding and decoding information ）;
  * TB（transform block）： Because an image is usually made of Y,U,V Three components , So a TU It contains three TB(TB-Y,TB-U,TB-V);

5、 ... and 、LCU and CTU

    Baidu and CSDN Nothing about LCU（logic coding unit） and CTU The difference between , From the concept and definition , I don't see any difference , Personally, I think it's the same thing Wanted H265 and HEVC Often written as H265/HEVC equally ,LCU and CTU It is often written as LCU/CTU.

Reference material

【1】H.265/HEVC Analysis of coding principle and processing flow
【2】CSDN： Video image I frame 、P frame 、B Frame details
【3】CSDN：H265/HEVC Codec series （1）： Image segmentation （Slice、Tile、CTU）
【4】HEVC Intra prediction learning （ One ）CTU、CU、PU、TU Understanding of unit division