H.264 Study Notes-related concepts

Here we record some basic concepts in the AVC process and update them occasionally.

 

Frame: A frame. It is equivalent to an image and contains a brightness matrix and two color matrices.

Field: field. A frame of image is scanned across rows to obtain the odd and even fields, which are called the top field, the bottom field, the odd field, and the even field.

Macroblock/MB: the basic unit for processing (prediction, transformation, and quantification) in H.264, with a size of 16x16 pixels.

Slice Group: A strip group. In each frame/field image, the MB/MB pairs are divided into multiple slice according to the order of raster scanning ).

I/P/B Macro Block: The I macro block can only use the encoded pixels in the slice for intra-frame prediction. The P Macro Block uses the pixels of the encoded image for inter-frame prediction. It is a Forward prediction. That is, the reference image is the image before the frame image in the playing sequence. The B Macro Block uses the encoded image pixels for inter-frame prediction. It is a bidirectional prediction, that is, the reference image can be an image before and after the playing sequence.

I/P/B Strip: The I strip that only contains the I Macro Block and the P macro block, B strip is included in both I strip and B macro block.

I/P/B frame: I frame only contains I band, P frame only contains P band, B frame only contains B band.

List0/list1: list0 is the list of forward reference frames, and list1 is the list of backward reference frames. Therefore, p-Frame Prediction only uses list0, and B-Frame Prediction uses list0 and list1.

Profile: grade. The so-called grade is a set of tools supported by H.264 and specific application fields. AVC specifies several different grades, including baseline profile, main profile, and extended profile ).

There are three types of tools:

Supports intra-and inter-frame encoding between I and P frames.

Using cavlc (context-adaptive variable-length encoding) for entropy Encoding

Use block Removal Filter

"Zigzag scan

"Motion estimation with 1/4 Pixel Precision

Minimum 3-level motion estimation block of 4x4

YUV sampling

Basic features:

Features of major grades:

B-frame, weighted intra-frame prediction, and cabac (context-adaptive binary arithmetic encoding)

Scalability features:

Including all features of basic grades and B-frame and weighted intra-Frame Prediction

Supports SP/Si strip for inter-stream switching, splicing, and random access.

 

Reference: deep understanding of Video Codec Technology-Based on H.264 standard and Reference Model

H.264 Study Notes-related concepts