Image Block Partitioning Using Boundary-Aware Tree Structures
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image encoding/decoding methods are limited in enhancing encoding efficiency due to the use of only a quad-tree block structure, which restricts the ability to efficiently handle high-resolution and high-quality video data, leading to increased data transfer and storage costs.
Innovation Solution
A method and apparatus that utilize various block shapes and aspect ratios, and efficiently perform block partitioning on the boundaries of images, including pictures, sub-pictures, slices, tiles, and bricks, by determining the appropriate partitioning method based on the presence of boundaries and block sizes relative to predefined thresholds, allowing for quad-tree, binary-tree, and ternary-tree partitioning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If only quad-tree block structure is used for image encoding/decoding, then the encoding method is simple, but the encoding efficiency cannot be enhanced for high-resolution and high-quality video data
Solution Approach 1:
The patent applies segmentation by dividing the image into multiple block types (square blocks, rectangular blocks, and triangular blocks) instead of using only quad-tree square blocks. This allows different block structures to be selected based on the characteristics of different regions in the image, thereby improving encoding efficiency for high-resolution and high-quality video data while maintaining manageable complexity through systematic classification of block types and partitioning methods.
2Adaptability or versatility
If various block shapes and aspect ratios are introduced to improve encoding efficiency, then adaptability increases, but device complexity increases
Solution Approach 1:
The patent applies local quality by allowing different block shapes and partitioning methods to be used in different regions of the image based on local characteristics. For example, rectangular blocks may be used in regions with strong directional features, while square blocks are used in isotropic regions. This localized adaptation improves encoding efficiency without requiring the entire system to handle all possible block types simultaneously, thus managing complexity.
Solution Approach 2:
The patent applies dynamics by enabling the encoder to dynamically select among multiple partitioning methods (quad-tree, binary-tree, ternary-tree) and block shapes based on the content characteristics of each region. This dynamic selection allows the system to adapt to different image features locally, improving overall encoding efficiency while the selection logic maintains manageable complexity through hierarchical decision-making.
3Productivity
If efficient block partitioning on boundaries is implemented, then processing efficiency improves, but the complexity of determining partitioning methods increases
Solution Approach 1:
The patent applies preliminary action by pre-defining multiple partitioning methods (quad-tree, binary-tree, ternary-tree) and block shapes that can be efficiently applied to boundary regions. Instead of deriving partitioning methods on-the-fly during encoding, the system has these methods prepared in advance, allowing rapid selection and application based on detected boundary characteristics, thus improving processing efficiency while keeping the determination logic manageable through predefined options.
Data Source
AI summary
There is provided a method and apparatus of encoding/decoding an image. The method comprises decoding information on block partitioning of a current block included in a current picture from a bitstream; determining a partitioning method of the current block on the basis of the information; and partitioning the current block by using the determined partitioning method, wherein the partitioning method is determined on the basis of whether or not the current block includes a predetermined boundary.