Adaptive Loop Filtering in QTBT Image Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high-definition and high-quality images in various applications poses challenges in maintaining consistent image quality across different devices and network environments, as existing image coding technologies struggle to enhance both subjective and objective image quality while improving coding efficiency.
Innovation Solution
An adaptive loop filtering (ALF) method and apparatus that determines the ALF control depth and applies it to specific block units within an image, using a quad-tree binary-tree (QTBT) structure for recursive splitting, to enhance image quality and coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If adaptive loop filtering is applied to all blocks uniformly, then image quality is improved, but coding efficiency deteriorates due to increased processing overhead
Solution Approach 1:
The patent divides the picture into multiple ALF control units at different depths in the QTBT structure. Different blocks can be filtered at different levels of granularity, allowing selective application of filtering only where needed rather than uniformly across all blocks, thus improving coding efficiency while maintaining image quality.
Solution Approach 2:
The patent enables different filtering strategies for different regions by determining ALF control units based on QTBT split depth. Blocks at different positions and depths can have different filtering applied, optimizing the balance between image quality improvement and processing overhead for each local region.
2Manufacturing precision
If image resolution is increased to meet high-definition demands, then image quality is improved, but the amount of information increases leading to higher processing requirements
Solution Approach 1:
The patent extracts and processes only the most important information by applying filtering selectively at the ALF control unit level based on QTBT depth. Instead of processing all high-resolution data uniformly, the system identifies and filters only the blocks that contribute most to perceived image quality, reducing the effective amount of information that needs intensive processing.
Solution Approach 2:
The patent introduces dynamic adaptability by allowing the filtering granularity to vary based on QTBT split depth. The ALF control depth can be adjusted dynamically, enabling the system to adapt the processing intensity to the actual content complexity and quality requirements of different image regions.
3Manufacturing precision
If filtering is applied at finer block levels, then image quality is improved, but device complexity increases
Solution Approach 1:
The patent segments the filtering process into multiple levels corresponding to different QTBT depths. This hierarchical segmentation allows the system to manage complexity by organizing filtering operations at discrete levels rather than attempting uniform fine-grained filtering across the entire picture, reducing overall device complexity while maintaining quality.
Solution Approach 2:
The patent applies filtering partially rather than excessively by determining ALF control units based on QTBT depth thresholds. Instead of applying filtering at the finest possible granularity everywhere, the system applies filtering only at appropriate depths and only to blocks where it provides benefit, reducing device complexity while maintaining adequate image quality.
Data Source
AI summary
A filtering method by a decoding device according to the present invention comprises the steps of: receiving information relating to ALF control depth; deciding an ALF control unit on the basis of information relating to the division depth and the ALF control depth of a current block in a restoration picture with respect to a current picture; determining whether or not ALF is applied to the current block in an ALF control unit level; and, if the ALF is determined to be applied to the current block, performing the ALF for the current block. In the filtering method, the current picture is recursively divided on the basis of a QTBT structure, and the current block is one of blocks divided on the basis of the QTBT structure.