Coefficient Scan Context Coding With Unified Pattern Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Video coding technologies face inefficiencies in assigning contexts for syntax elements indicating significant coefficients, particularly when dealing with non-square coefficient groups and varying scan types, which increases hardware and software complexity and reduces coding efficiency.
Innovation Solution
The use of the same context patterns for multiple scan types, such as horizontal, vertical, and diagonal scans, associated with conditions of neighboring sub-blocks containing non-zero coefficients, allows for computational efficiency and improved coding efficiency by reducing the need for determining context patterns based on transform unit sizes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If different context patterns are assigned for different scan types (horizontal, vertical, diagonal), then coding precision for significance syntax elements is improved, but device complexity increases due to multiple context pattern determination logic
Solution Approach 1:
The patent applies universality by using a single context pattern for multiple scan types (horizontal, vertical, and diagonal scans). Instead of maintaining separate context patterns for each scan direction, the invention uses one unified context pattern that serves all scan types, thereby reducing device complexity while maintaining adequate coding precision through the shared context model.
2Productivity
If context patterns are determined based on transform unit sizes and scan types, then coding efficiency is improved through adaptive context assignment, but productivity decreases due to increased computational resources required
Solution Approach 1:
The patent extracts the dependency on scan type and transform unit size from the context pattern determination process. By removing these variables from the context selection logic, the invention simplifies the computational requirements while maintaining the essential context adaptation needed for efficient coding of significance syntax elements.
Solution Approach 2:
The invention changes the parameter of context pattern selection from being dependent on multiple variables (scan type, transform unit size) to a fixed or simplified parameter structure. This parameter change reduces the computational burden of determining contexts while preserving the adaptive coding benefits through the use of a unified context pattern that works across different scan types and block sizes.
Data Source
AI summary
Techniques are described for a video coder (e.g., video encoder or video decoder) that is configured to select a context pattern from a plurality of context patterns that are the same for a plurality of scan types. Techniques are also described for a video coder that is configured to select a context pattern that is stored as a one-dimensional context pattern and identifies contexts for two or more scan types.