Context Modeling for Entropy Coding Transform Skip
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing video coding standards, such as HEVC, face challenges in reducing the number of context coded bins in residual coding of transform coefficients while maintaining coding efficiency, particularly in the transform skip mode, which affects the throughput of the CABAC coding algorithm.
Innovation Solution
The proposed method involves entropy coding of residual symbol levels in a versatile video coding (VVC) system by context modeling based on at least three neighboring pixels, including a first pixel on the left, a second pixel above, and a third pixel on the left-above diagonal, to achieve a reduced number of context coded bins, specifically targeting 1.75 bins per transform coefficient with or without transform skip.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the number of context coded bins is reduced to improve throughput, then coding efficiency may be sacrificed
Solution Approach 1:
The patent segments the transform coefficient block into multiple sub-blocks and processes them independently with separate context models. This allows the use of fewer context models per sub-block while maintaining overall coding efficiency through the collective contribution of all sub-blocks. The segmentation enables parallel processing and reduces the per-coefficient overhead.
Solution Approach 2:
The patent applies different context modeling strategies to different regions of the transform coefficient block. By using local statistics and adaptive context selection within sub-blocks, the system achieves efficient coding with reduced context models. Each sub-block can adapt its context usage based on local characteristics, maintaining efficiency while reducing overall complexity.
2Loss of substance
If transform skip mode is used to improve compression, then the number of context coded bins increases to 2 bins per coefficient
Solution Approach 1:
The patent merges the context modeling for transform skip mode with the regular transform mode by using a unified set of context models that work for both cases. This consolidation allows the system to handle transform skip coefficients efficiently without requiring a separate, larger set of context models, thereby maintaining throughput while supporting the compression benefits of transform skip.
Solution Approach 2:
The patent dynamically adjusts context modeling parameters based on the transform mode being used. By changing context selection criteria and model initialization parameters according to whether transform skip is active, the system optimizes the balance between compression efficiency and coding throughput for each specific mode without requiring fixed, oversized context model sets.
Data Source
AI summary
A versatile video coning method is provided for an electronic device. The method includes obtaining a video signal; partitioning the video signal into a plurality of coding blocks; generating a residual coding block from each coding block with or without transformation and quantization; splitting the residual coding block into a plurality of sub-blocks and each sub-block into a plurality of pixels; entropy coding a residual symbol level at each pixel in the residual coding block; and outputting a bitstream including entropy coded residual symbol levels. Entropy coding the residual symbol level at each pixel in the residual coding block includes context modeling of the residual symbol level of the pixel in a transform skip mode based on context information of at least three neighboring pixels of the pixel including a first pixel on the left, a second pixel on the above, and a third pixel on the left-above diagonal.