Context Design for Video Prediction Mode Entropy Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding technologies, such as HEVC and VVC, face inefficiencies in entropy coding of prediction modes and coded block flags, particularly when dealing with intra-inter prediction modes, as they often rely on a single context for pred_mode_flag signaling and do not effectively utilize neighboring block information for context design.
Innovation Solution
The method involves setting a prediction mode flag for a current block to indicate whether it is coded in inter, intra, or inter-intra prediction mode, deriving a context value based on this flag, and performing entropy coding using specific context values associated with each prediction mode, allowing for separate contexts for intra, inter, and inter-intra prediction modes to optimize coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single context is used for pred_mode_flag signaling in intra-inter prediction modes, then device complexity is reduced, but coding efficiency deteriorates due to inability to differentiate between intra and inter prediction characteristics
Solution Approach 1:
The patent segments the context design for pred_mode_flag signaling into separate contexts based on prediction mode type. Specifically, it uses different context models for intra prediction modes versus inter prediction modes, allowing the entropy coding process to adapt to the statistical characteristics of each prediction type. This segmentation enables more efficient probability modeling without significantly increasing overall system complexity.
Solution Approach 2:
The patent implements dynamic context selection where the context used for pred_mode_flag coding changes based on the prediction mode being encoded. The context model dynamically adapts to the current prediction mode state, switching between intra-prediction-specific contexts and inter-prediction-specific contexts. This dynamic adaptation improves coding efficiency by matching the context to the actual data characteristics.
2Device complexity
If neighboring block information is not utilized for context design, then device complexity is reduced, but coding efficiency deteriorates due to loss of spatial correlation information
Solution Approach 1:
The patent incorporates feedback from neighboring block prediction modes into the context design for current block coding. The context selection and probability modeling utilize information from already-coded neighboring blocks to predict the likelihood of different prediction modes for the current block. This feedback mechanism captures spatial correlation without requiring complex additional processing.
Solution Approach 2:
The patent performs preliminary analysis of neighboring block prediction modes to establish the context before coding the current block's pred_mode_flag. By pre-computing context probabilities based on neighboring blocks, the system prepares optimized context models in advance, reducing the need for complex real-time calculations during actual coding.
3Productivity
If separate contexts are used for intra and inter prediction modes, then coding efficiency is improved through better probability modeling, but device complexity increases
Solution Approach 1:
The patent applies local quality by using different context models tailored to specific prediction mode categories. Instead of a uniform context model, it implements specialized context designs for intra prediction regions and inter prediction regions, where each local context is optimized for its specific prediction type's statistical characteristics. This localized optimization improves coding efficiency while keeping the overall structure manageable.
Data Source
AI summary
A method of controlling intra-inter prediction for decoding or encoding of a video sequence, is performed by at least one processor. The method may include setting a prediction mode flag associated with a current block to indicate whether the current block is coded in inter prediction mode, intra prediction mode, or inter-intra prediction mode; deriving a context value associated with a coded block flag associated with the current block based on the prediction mode flag associated with the current block. Deriving the context value may include based on the current block being determined to be coded in the intra prediction mode, setting the context value of the coded block flag to a first value, and based on the current block being determined to be coded in the inter prediction mode, setting the context value of the coded block flag to a second value.