Neural Entropy Coding for Layered Video Probability Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression technologies face challenges in improving entropy coding efficiency, particularly in deep learning-based layered coding, which affects the compression ratio with minimal impact on picture quality.
Innovation Solution
An entropy encoding/decoding method utilizing a neural network to process base and enhancement layer information, generating probability values for entropy encoding/decoding, and employing feature extraction techniques to enhance coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If deep learning-based layered coding is used to improve compression ratio, then data compression efficiency is improved, but entropy coding precision deteriorates
Solution Approach 1:
The patent segments the coding process into base layer and enhancement layer, applying different coding strategies to each. The base layer uses traditional entropy coding methods while the enhancement layer utilizes neural network-based probability estimation, allowing optimized compression for each layer's specific characteristics without compromising overall precision
Solution Approach 2:
The patent introduces a neural network as an intermediary component between the enhancement layer data and the entropy coding process. This neural network estimates probability distributions for enhancement layer symbols based on base layer information, serving as a mediator that bridges the gap between layered coding compression and accurate probability estimation for entropy coding
2Productivity
If base layer information is used to estimate enhancement layer probability distributions, then coding efficiency is improved, but system complexity increases
Solution Approach 1:
The patent designs the neural network to perform multiple functions: it processes base layer information, estimates enhancement layer probability distributions, and outputs parameters for entropy coding. This multi-functional approach consolidates what could be separate complex modules into a single unified component, reducing overall system complexity while maintaining coding efficiency
Solution Approach 2:
The patent changes the parameter representation by using neural network outputs (such as mean and variance for Gaussian distributions) to define probability distributions instead of directly coding enhancement layer symbols. This parameter-based approach simplifies the coding process while improving efficiency by focusing on distribution characteristics rather than individual symbol values
Data Source
AI summary
The technology of this application relates to an entropy encoding method that includes obtaining base layer information of a to-be-encoded picture block, where the base layer information corresponds to M samples in the picture block, and M is a positive integer, obtaining K elements corresponding to enhancement layer information of the picture block, where the enhancement layer information corresponds to N samples in the picture block, both K and N are positive integers, and N≥M, inputting the base layer information into a neural network to obtain K groups of probability values, where the K groups of probability values correspond to the K elements, and any group of probability values is for representing probabilities of a plurality of candidate values of a corresponding element, and performing entropy encoding on the K elements based on the K groups of probability values.