Image Residual Coding with Last-Coefficient Limits for Faster CABAC
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high-resolution, high-quality images and videos, such as 4K and 8K Ultra High Definition, along with the growth of virtual reality and immersive media, has led to a need for more efficient image compression techniques to reduce transmission and storage costs, while existing methods like CABAC face throughput issues due to high data dependency.
Innovation Solution
An image decoding method that derives the maximum number of context coded bins based on the position of the last significant coefficient of a block, optimizing the encoding and decoding processes to enhance throughput and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If CABAC is used for residual coding, then compression efficiency is improved, but throughput is reduced due to high data dependency
Solution Approach 1:
The patent segments the residual coding process by dividing the transform coefficients into multiple groups and applying different context modeling strategies to each group. This segmentation reduces the data dependency within each group while maintaining overall compression efficiency, thereby improving throughput without sacrificing compression performance.
Solution Approach 2:
The patent introduces dynamic context adaptation where the context models are updated and adjusted based on the statistical properties of the residual data. This dynamic approach allows the system to adapt to varying data characteristics, reducing data dependency and improving processing throughput while maintaining high compression efficiency.
2Measurement precision
If the number of context coded bins is increased, then coding precision is improved, but processing complexity increases
Solution Approach 1:
The patent applies local quality by using different context modeling approaches for different regions or groups of transform coefficients. Instead of uniformly increasing the number of context coded bins across all coefficients, the patent selectively applies enhanced context modeling only where necessary, thereby maintaining coding precision while reducing overall processing complexity.
Solution Approach 2:
The patent changes the parameters of context modeling by adjusting the number of context coded bins dynamically based on the characteristics of the residual data. This parameter adaptation allows the system to achieve high coding precision when needed while reducing complexity when the data characteristics allow for simpler modeling.
Data Source
AI summary
According to the present disclosure, an image decoding method performed by a decoding device comprises the steps of: receiving a bitstream including residual information on a current block; deriving the maximum number of context coded bins related to the residual information; decoding syntax elements included in the residual information on the basis of the maximum number of context coded bins; deriving transform coefficients for the current block on the basis of the decoded syntax elements; deriving residual samples for the current block on the basis of the transform coefficients; and generating reconstruction samples for the current block on the basis of the residual samples for the current block, wherein the maximum number of context coded bins is derived on the basis of a position of a last significant coefficient of the current block.