Diagonal Scan for Transform Coefficients in Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards face challenges in parallelizing context selection for entropy coding of transform coefficients, particularly at decoders where numerous dependencies complicate efficient processing, impacting coding efficiency and parallel processing capabilities.
Innovation Solution
The method involves diagonal scanning in a fixed direction to transform 2D arrays of transform coefficients into 1D arrays, simplifying context selection by reducing dependencies to fewer than 11 neighbors, and using arithmetic coding for generating coded bits.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If traditional context selection methods are used for entropy coding of transform coefficients, then coding efficiency is improved, but parallel processing capability deteriorates due to numerous dependencies
Solution Approach 1:
The patent segments the transform coefficient array into multiple independent blocks that can be processed in parallel. By dividing the large coefficient array into smaller independent units, the context selection process for each block can proceed independently without dependencies on other blocks, enabling parallel processing while maintaining coding efficiency through localized context modeling.
Solution Approach 2:
The patent introduces a new dimension to the scanning process by using diagonal scanning patterns across multiple coefficient blocks simultaneously. This dimensional approach allows the encoder to process coefficients along diagonal paths that span across block boundaries in a coordinated manner, reducing inter-block dependencies while preserving the statistical relationships needed for efficient coding.
2Loss of information
If context selection depends on many neighboring coefficients, then coding efficiency is improved, but device complexity increases due to more dependency checks
Solution Approach 1:
The patent applies local quality by using different context selection strategies for different regions of the transform coefficient array. Specifically, it uses diagonal scanning that adapts the number and type of neighboring coefficients considered for context selection based on the local position and characteristics of each coefficient block, reducing overall complexity while maintaining efficiency in critical regions.
Solution Approach 2:
The patent implements partial action by selectively applying full context dependency only where necessary for coding efficiency, while using simplified context selection for other regions. The diagonal scanning approach processes only the necessary neighboring coefficients for each position rather than checking all possible neighbors, reducing device complexity while maintaining adequate coding performance.
Data Source
AI summary
A method and apparatus for encoding bit code utilizing context dependency simplification to reduce dependent scans. The method includes retrieving at least one two-dimensional array of transform coefficient, transforming the at least one two-dimensional array of transform coefficient to a one-dimensional coefficient scanning using a diagonal scan in a fixed direction, utilizing the at least one one-dimensional array of transform coefficients for context selection based on fewer than 11 neighbors, potentially selected based on scan direction, slice type, coding unit type and binarization, and performing arithmetic coding to generate coded bit utilizing context selection and binarization.