Split Flag Context Modeling Using Node Dimensions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining the context model of a split flag in video encoding and decoding are inaccurate, particularly when using binary or ternary tree split methods, leading to low encoding and decoding efficiency due to reliance on depth information rather than node dimensions.
Innovation Solution
A context modeling method that compares the height and width of a current node with adjacent leaf nodes to determine whether it should be further split, using preset conditions to accurately establish a context model for the split flag, thereby improving entropy coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the context model of a split flag is determined based on depth information (quadtree depth comparison), then the method is simple to implement, but the accuracy of context model determination is low, especially when using binary or ternary tree split methods
Solution Approach 1:
The patent changes the parameters used for context model determination from depth-based (quadtree depth) to dimension-based (height and width comparisons). By comparing the height of the current node with adjacent leaf nodes and the width of the current node with adjacent leaf nodes, the method achieves more accurate context model determination that works correctly for both quadtree, binary tree, and ternary tree split methods
Solution Approach 2:
The patent transitions from a one-dimensional depth-based approach to a two-dimensional approach by considering both height and width dimensions of nodes. This dimensional expansion allows accurate determination of whether nodes should be further split regardless of the tree split method used, resolving the inaccuracy problem in binary and ternary tree splits
2Productivity
If the context model is determined using depth information comparison, then the encoding and decoding process is straightforward, but the encoding and decoding efficiency is low due to inaccurate probability models
Solution Approach 1:
By changing from depth-based parameters to dimension-based parameters (height and width), the patent creates more accurate probability models for split flag encoding. The context model index is determined by comparing height differences and width differences between current nodes and adjacent leaf nodes, which accurately reflects the actual splitting patterns and improves encoding efficiency
Data Source
AI summary
A context modeling method and apparatus of a split flag are provided. The method includes: obtaining a height and a width of a current node, a height of a first leaf node in a first direction, and a width of a second leaf node in a second direction; determining whether the current node meets a first preset condition and a second preset condition, where the first preset condition includes the height of the current node is greater than the height of the first leaf node, and the second preset condition includes the width of the current node is greater than the width of the second leaf node; and determining a context model of a split flag of the current node based on whether the first and second preset conditions are met.