Chapter 1 Introduction Video Code: Objective: To remove redundant ingredients in video data, reduce the amount of data characterized.
1.1 overview
1.1.1 video
1. Frame: Video in Video 2. Video Sequence: Video 3. Video 3. Video 3. Video Space Domain and Time Domain: Each image in the video sequence consists of n-multiply M pixels Therefore, the video sequence may represent a three-dimensional matrix, n multiply the M two dimensions to form a spatial domain, and the other dimension represents the time domain of the video. 4. Frame rate (unit: fps): Number of frames played per second Notes: 25 ~ 30FPS
1.1.2 Video Compression & Coding
1. Video compression is a special data compression. 2. Data compression classification (no distortion): 2. Lossless compression compression ratio 5: 1 or less, image 3: 1 More common 2.2 damage compression (video more this way) Essential: Distortion to change compression ratio: Throw it out of certain Information people are difficult to detect or receive measurement of compression algorithm performance parameters: code rate and distortion target: 1. Spatial domain: Distortion must obtain the highest compression ratio (the lowest code rate) or the code rate must be minimal. Distortion. 2. Time domain: Changes in frame rates? .
1.2 Video Coding Standard Definition: Specify the syntax semantics and interpreter of the encoded stream: ie, as long as the encoder encodes the code stream meets the standard rules. Syntax elements: The basic unit of the encoder outputs the code stream data, consists of several bits, characterizes specific physical meaning. Syntax: The organizational structure semantics of each syntax element: describes the specific meaning of grammatical elements
Basic architecture and characteristics of 1.3H.265 / HEVC
1.3.2 encoding framework
1. Inframe prediction: Objective: To remove the spatial relevance of the image. Method: The current pixel block 2. is predicted by encoding the current pixel block 2. Inter-frame prediction purposes: Remove the time-dependent method of the image: through the encoded image as the reference image of the current frame, the motion information of each block is obtained. . 3. Transform Quantization Target: Removal of the frequency domain correlation: Transform the residual data: Divided into two steps: (1) Transform code: Time domain to frequency domain; (2) Quantization module: reduce image encoding scope. 4. Go to block filter elimination square effect 5. Sample adaptive compensation weakening ringing effect 6. Entropy coding purpose: encoding data into binary flow method: context-based adaptive binary arithmetic coding (CABAC)
1.3.3 H.265 / HEVC Characteristic Coding Technology
1. Coding Unit (Cu): Coding Tree Unit (CTU) Coding Tree Unit (CTU): The size of the macroblock converting unit TU set by the encoder supports large-size discrete cosine transformation, and processing a flat region in the high resolution screen. Better improvement compression ratio. 2. Improved intra prediction technology (35 species) 33 angular predictions, DC prediction modes, PLANAR prediction mode 3. Inter Bloom prediction technology (MERGE), advanced motion vector prediction technology (AVMP), based on MERGE SKIP mode. Principle: It is to reduce the redundancy of adjacent blocks by vacancy correlation and time domain relevance. 4. RQT mode based on the adaptive technology of the four-tree structure, the adaptive transform block size, can reserve the optimal compromise in energy concentration and details. 5. Sao technical pixel adaptive compensation technology, a filtering method, reducing ringing effect. 6. IBDI technology increases uncompressed image pixels in the input of the encoder, and recovering the pixel in the output of the decoder. Reduce intra-frame prediction error, but to be larger memory space, solution: introduce the reference frame compression algorithm, reduce the amount of data.
Our other product:
