H. 264 concept

One 、H.264 Concept

H.264 By ITU-T Video expert group and ISO/IEC A joint group of motion picture experts （JVT,Joint Video Team） Proposed Video codec standard for highly compressed data .

Two 、H.264 Data format

H.264 By ** Video coding layer （VCL） and Network adaptation layer （NAL）** form .

①VCL：H264 code / The core of compression , Mainly responsible for encoding video data / Compress , Re segmentation .

②NAL The basic unit of a layer is NALU（ Network abstraction layer unit ）
blog.csdnimg.cn/25b161a63ce74813a55a1b090a51d6ea.png)
NALU = NALU header + NALU payload
(1) General structure ：

RBSP： Raw data （ It could be encoded video data , It could also be other data ）, Add... For alignment “0”
bits .

(2)NALU The head of the （8bit）

(3)nalu The type of
![ Insert picture description here ](https://img-

3、 ... and 、VCL How to manage H.264 Video data

① Compress ： forecast （ Intra prediction and inter prediction ）–>DCT Change and quantification –> Bitstream coding .
② Segmentation data , Mainly for the third step .“ section （slice）”、“ Macroblock （macroblock）” Is in VCL The concept of , On the one hand, improve coding efficiency and reduce coding rate , On the other hand, improve the flexibility of network transmission .
③ Package as NAL.
VCL Finally, it will be packaged as NAL.

Four 、NAL The data structure of the header

5、 ... and 、H.264 Bitstream structure

H.264 = start_code + NALU
（start_code：00000001 or 000001）

6、 ... and 、H.264 Three frames

Most video compression , All the frames are classified , Mainly divided into I frame 、P frame 、B frame , These are not H264 Unique , We just H264 Let's talk about these categories .

①I frame ：Intra-coded picture, Represents a keyframe , Full intra compression . Compressing this frame does not require reference to other frames .

②P frame ：Predicted picture, Forward prediction frame , To compress this frame, you need to refer to the previous one I/P Frame as reference , Do motion estimation , In order to get a comparison I Larger compression ratio of frames .

③B frame ：Bi-predictive picture, Two way prediction frame . Compressing this frame not only refers to the previous I/P frame , You also need to refer to the following one P frame , Perform bidirectional motion estimation , In some cases, a ratio of P Higher compression ratio of frames .

Compression ratio ：B > P > I
image quality ：I > P > B （ Because the motion vector obtained from motion estimation will be quantized , Loss introduced . And P、B The frame may refer to other P frame （ chain ）, Thus, it is easier to cause error accumulation and error drift .）

7、 ... and 、 Decode timestamp and display timestamp