Image Coding Apparatus Spatial-Activity Partitioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image coding methods, such as those in the ISO/IEC 14496-2 Part 10 International Standard, fail to adequately suppress the increase in calculation amount and degradation in image quality due to the generation of artifacts like ringing and blocking artifacts, which are not effectively considered in the coding cost calculations, and do not account for the human visual system's impact.
Innovation Solution
An image coding method that determines coding conditions for blocks based on spatial-activity values, comparing these values with thresholds to decide between smaller and larger partitioning sizes, and considers motion vectors to optimize coding efficiency and quality, thereby reducing artifacts and subjective quality degradation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If exhaustive selection processes are used to select the most appropriate sub-macroblock coding tools, then coding quality may be improved, but calculation amount increases significantly
Solution Approach 1:
The patent changes the parameter used for coding tool selection from comprehensive coding cost (rate + distortion) to spatial-activity value, which reflects human visual system characteristics. This parameter change enables efficient selection of appropriate partitioning sizes without exhaustive calculation while maintaining perceptual quality.
Solution Approach 2:
The patent performs preliminary classification of blocks into flat and non-flat regions using spatial-activity values before selecting coding tools. This preliminary action avoids the need for exhaustive coding cost calculations later, significantly reducing calculation amount while maintaining coding quality.
2Loss of information
If conventional coding cost calculations are used, then coding rate and distortion are considered, but ringing artifacts and blocking artifacts are not reflected
Solution Approach 1:
The patent introduces spatial-activity value as a new parameter that directly reflects human visual sensitivity to artifacts. This parameter change allows the coding system to anticipate and prevent ringing and blocking artifacts by identifying flat regions where such artifacts are most noticeable, rather than relying on post-hoc distortion measurements.
Solution Approach 2:
The patent uses spatial-activity values to provide feedback about region characteristics (flat vs. non-flat) that informs coding tool selection. This feedback mechanism enables the encoder to adaptively choose partitioning sizes that prevent artifact generation in the first place, improving image quality without exhaustive search.
3Manufacturing precision
If small partitioning sizes are used to reduce artifacts in flat regions, then image quality improves, but coding efficiency decreases
Solution Approach 1:
The patent applies different partitioning sizes locally based on spatial-activity values. Flat regions (low spatial-activity) use small partitioning sizes to prevent artifacts, while non-flat regions (high spatial-activity) use large partitioning sizes for coding efficiency. This local adaptation resolves the contradiction by applying the right partitioning size in the right location.
Solution Approach 2:
The patent dynamically adjusts partitioning size based on the spatial-activity characteristics of each block. Rather than using a fixed partitioning size for all blocks, the system adapts the partitioning size to local image content, achieving both artifact reduction in flat regions and coding efficiency in complex regions.
Data Source
AI summary
A method of coding an original image, including: partitioning (S110) the original image into blocks; determining (S120) a coding condition for a current one of the blocks; and coding (S130) the current block according to the determined coding condition, wherein the determining (S120) of the coding condition for the current block includes calculating (S210) a first spatial-activity value indicating complexity of at least a part of regions of the current block; comparing (S220) the first spatial-activity value with a first predetermined threshold value; determining (S230), as the coding condition for the current block, a first coding condition for small partitioning size, when the first spatial-activity value is smaller than the first predetermined threshold value; and determining (S240), as the coding condition for the current block, a second coding condition for large partitioning size, when the first spatial-activity value is equal to or larger than the first predetermined threshold value.