Adaptive Block Splitting for Image Encoding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image encoding/decoding methods are limited by their fixed square block splitting structures, which hinder adaptability to varying local content characteristics, leading to inefficiencies in encoding and decoding high-resolution and high-quality images.
Innovation Solution
The method involves splitting a coding tree unit into coding units using various block splitting structures, including binary and ternary tree splitting in addition to quadtree splitting, with flags determining the splitting modes and sizes, allowing for more flexible and efficient encoding and decoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a fixed square block splitting structure (quadtree only) is used, then the device complexity is reduced and ease of operation is improved, but the adaptability to local content characteristics deteriorates and encoding efficiency is reduced
Solution Approach 1:
The patent divides the image into multiple block types (square blocks, horizontal blocks, vertical blocks) with different splitting structures. The quadtree splitting handles square blocks, while additional binary tree splitting handles rectangular blocks. This segmentation of block types allows the system to adapt to different local content characteristics without overwhelming complexity.
Solution Approach 2:
The patent introduces dynamic block splitting structures that can adapt to content characteristics. By allowing both quadtree splitting (for square blocks) and binary tree splitting (for rectangular blocks) based on content analysis, the system dynamically selects the appropriate splitting structure for each region, improving adaptability while managing complexity through conditional application.
2Adaptability or versatility
If various block splitting structures (binary tree, ternary tree) are introduced, then the adaptability to local content characteristics is improved, but the device complexity increases
Solution Approach 1:
The patent segments the block splitting process into distinct phases: quadtree splitting for initial square block division, followed by optional binary tree splitting for rectangular block refinement. This segmentation allows each splitting type to specialize in specific block shapes, improving adaptability while keeping the overall system manageable through clear separation of functions.
Solution Approach 2:
The patent applies different splitting structures (quadtree, binary tree) to different local regions of the image based on content characteristics. Square regions use quadtree splitting while rectangular regions use binary tree splitting. This local quality approach allows the system to optimize for specific content types in specific regions without requiring all regions to handle all splitting types, thus improving adaptability while controlling complexity.
Data Source
AI summary
The present invention relates to an image encoding/decoding method. The image decoding method includes splitting a coding tree unit (CTU) into at least one coding unit (CU) according to a block splitting structure and performing CU-based decoding, in which the block partition structure is configured such that at least one of binary tree splitting and ternary tree splitting is performed after quadtree splitting is performed.