Hybrid Coding Split Tree Structure for Video Image
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The H.265 video coding standard's quadtree-based method is inadequate for adapting to various textures due to its limitation to square coding units, and the cascaded quadtree and binary tree split mode results in excessive split levels and inability to split leaf nodes of different types.
Innovation Solution
A video image coding and decoding method that uses a hybrid coding split tree structure with multiple split modes, allowing for rectangular coding units and determining split modes based on aspect ratios and widths to optimize split levels and CU shapes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a quadtree-based CTU split method is used with square CUs, then the coding structure is simple, but the CU cannot adapt well to textures of various shapes
Solution Approach 1:
The patent divides the CTU into multiple CUs using both quadtree and binary tree segmentation methods. The quadtree splits a node into four equal quadrants, while the binary tree splits a node into two unequal parts. This multi-segmentation approach allows the coding structure to adapt to various texture shapes while maintaining reasonable complexity through systematic division rules.
Solution Approach 2:
The patent introduces asymmetric binary tree splitting where a node can be divided into two unequal parts (e.g., 1:3 or 3:1 ratio) in addition to the symmetric quadtree splitting. This asymmetry enables better adaptation to non-square textures and various shape requirements, resolving the contradiction between adaptability and structural simplicity.
2Adaptability or versatility
If a cascaded QT and BT split mode is used, then rectangular CUs can be generated, but there are excessive split levels
Solution Approach 1:
The patent merges quadtree and binary tree splitting modes into a unified coding structure. Instead of using them separately or in a complex cascaded manner, the patent integrates both methods at the same hierarchical level, allowing the encoder to select between QT and BT splits based on content requirements. This reduces the total number of split levels while maintaining the ability to generate rectangular CUs.
3Adaptability or versatility
If a cascaded QT and BT split mode is used, then rectangular CUs can be generated, but the leaf node of QT cannot be split in QT manner
Solution Approach 1:
The patent creates a universal splitting system where any non-leaf node, regardless of whether it was created by QT or BT splitting, can undergo either QT or BT splitting. This multi-functional approach allows leaf nodes of either type to be further divided using either method, providing maximum flexibility and ease of operation while generating rectangular CUs as needed.
Data Source
AI summary
A video image decoding method for parsing split mode information of a coding tree node in a bitstream, obtaining a split mode of the coding tree node from a candidate split mode set of the coding tree node based on the split mode information, parsing the bitstream to obtain coding information of the coding tree node, and reconstructing a pixel value of the coding tree node based on the split mode information and the coding information of the coding tree node.