Arithmetic Image Coding with Hierarchical Context Probability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional arithmetic coding methods face difficulties in accurately classifying image types, leading to inefficient coding efficiency due to either overly fine or broad classification, which affects the prediction of symbol occurrence probabilities.
Innovation Solution
The proposed method employs a hierarchical context classification structure that uses both high-order and low-order contexts to calculate coding probability information, updating both to improve prediction accuracy and coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If fine classification of image types is performed, then prediction accuracy of symbol occurrence probability is improved, but device complexity and difficulty of classification increase
Solution Approach 1:
The patent segments the context classification into multiple hierarchical levels (first level, second level, third level, etc.), where each level divides the signal types further. This segmentation allows fine-grained classification without requiring a single complex classification structure, as each level handles a specific aspect of signal type differentiation.
Solution Approach 2:
The patent introduces a hierarchical dimension to the classification structure, organizing signal types across multiple levels rather than a single flat level. This dimensional approach allows systematic progression from broad to fine-grained classification, managing complexity through structured hierarchy.
2Device complexity
If broad classification of image types is performed, then device complexity is reduced, but prediction accuracy of symbol occurrence probability deteriorates
Solution Approach 1:
By segmenting the classification into hierarchical levels, the patent achieves fine prediction accuracy without requiring all classification logic to exist at a single level. Each level handles specific classification tasks, maintaining overall system simplicity while achieving fine-grained accuracy through cumulative segmentation.
Solution Approach 2:
The hierarchical classification structure allows dynamic selection of classification depth based on signal type characteristics. The system can adaptively progress through classification levels only as needed, avoiding unnecessary complexity for simple cases while providing fine-grained classification when required.
3Ease of operation
If single context classification is used, then ease of operation is improved, but coding efficiency deteriorates due to inaccurate probability prediction
Solution Approach 1:
The patent segments the probability prediction process into multiple hierarchical contexts, where each context level contributes to the overall probability calculation. This segmentation maintains operational simplicity by keeping each individual context management straightforward, while the combination of multiple contexts achieves accurate probability prediction for improved coding efficiency.
Solution Approach 2:
The patent merges multiple hierarchical context classifications to form a comprehensive probability prediction model. By combining the results from first-level, second-level, third-level (and potentially deeper) classifications, the system achieves accurate symbol occurrence probability prediction while maintaining the operational simplicity of individual context managers.
Data Source
AI summary
An image coding method for performing compression-coding on image data includes: binarizing a signal to be coded of the image data, to generate a binary signal; determining a low-order context which is a context associated with a type of the signal to be coded, and a high-order context that is common to the low-order context and a context associated with another type; calculating coding probability information to be used in arithmetic coding of the binary signal, using (i) high-order probability information corresponding to the determined high-order context and (ii) low-order probability information corresponding to the determined low-order context; performing the arithmetic coding on the binary signal using the coding probability information; and updating the high-order probability information and the low-order probability information based on the binary signal.