Arithmetic Image Coding with Dynamic Frequency-Range Contexts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional arithmetic coding methods do not achieve sufficient coding efficiency due to inefficient context management and fixed group sizes, which affect the updating frequency of symbol occurrence probabilities and overall coding performance.
Innovation Solution
The proposed method segments image data into blocks based on frequency component types, determining contexts for each group and performing binary arithmetic coding using probability information specific to these contexts, allowing for variable group sizes and improved coding efficiency by updating probability information based on coefficients within each group.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If the block is segmented into a large number of small groups, then the coding precision can be improved by using more specific contexts, but the device complexity increases due to more contexts needed
Solution Approach 1:
The block is segmented into groups based on frequency ranges (low-frequency, mid-frequency, high-frequency groups) rather than individual coefficients. This segmentation allows the use of fewer contexts while still capturing the statistical characteristics of different frequency regions, resolving the contradiction between coding precision and device complexity.
2Productivity
If the group size is fixed, then the device complexity is reduced, but the coding efficiency deteriorates due to insufficient updating frequency of symbol occurrence probabilities
Solution Approach 1:
The group sizes are made dynamic rather than fixed. The low-frequency group, mid-frequency group, and high-frequency group can have different sizes depending on the actual coefficient distribution. This dynamic grouping allows for more frequent updating of symbol occurrence probabilities in regions where coefficients are dense, improving coding efficiency without requiring excessive complexity.
3Productivity
If the symbol occurrence probability is updated frequently for each coefficient, then the coding efficiency is improved, but the loss of time increases due to repeated updating operations
Solution Approach 1:
Multiple coefficients are merged into groups, and a single context is used for each group. The symbol occurrence probability is updated once per group rather than once per coefficient. This merging approach maintains coding efficiency by capturing the statistical characteristics of the group while reducing the time loss associated with frequent updating operations.
Data Source
AI summary
An image coding method for coding image data on a block-by-block basis includes: segmenting a block including coefficients of frequency components, into groups corresponding to frequency ranges based on types of the coefficients, and determining a context corresponding to each of the groups; performing, for each of the coefficients, binary arithmetic coding on the coefficient according to a probability information item corresponding to the context, the context being corresponding to the group having the coefficient; and updating, for each of probability information items, the probability information item corresponding to the context, based on the coefficient included in the group corresponding to the context, wherein in the segmenting, the block is segmented into the groups with various numbers of coefficients.