Arithmetic Context Sharing for Image Encoding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional arithmetic coding methods do not provide sufficient coding efficiency due to the large number of contexts required for different processing unit sizes and coefficient positions, leading to low update frequencies and accuracy of coded probability information.
Innovation Solution
The method selects shared contexts for signals with the same statistical properties, reducing the number of contexts and increasing their updates, thereby enhancing prediction accuracy and coding efficiency, especially for larger processing units and lower frequency components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If different contexts are used for different processing unit sizes and coefficient positions, then the coding can be adapted to specific conditions, but the number of contexts increases leading to reduced update frequency and lower accuracy of coded probability information
Solution Approach 1:
The patent merges contexts by determining whether neighboring blocks can share the same context model. Instead of creating separate contexts for each block, the system evaluates statistical properties and updates contexts only when necessary, combining context usage across multiple blocks to maintain accuracy while reducing the total number of contexts.
Solution Approach 2:
The patent makes contexts universal by allowing the same context to be used for different processing unit sizes and coefficient positions when statistical properties are similar. The context model is designed to handle multiple block sizes (4x4, 8x8, 16x16, 32x32) and various coefficient positions within a unified framework, increasing update frequency for each context.
2Adaptability or versatility
If the number of contexts is increased to cover all processing unit sizes and coefficient positions, then more specific coding conditions can be addressed, but the memory size for storing contexts increases and update frequency decreases
Solution Approach 1:
The patent combines multiple context models into a unified context that can serve different block sizes and coefficient positions. By evaluating statistical properties and using context sharing between neighboring blocks, the system reduces the total number of contexts from what would be required for complete coverage to a smaller set of shared contexts.
Solution Approach 2:
The context model is designed to be universal, handling multiple processing unit sizes (4x4, 8x8, 16x16, 32x32) and various coefficient positions within a single context framework. This multi-functional context reduces memory requirements while maintaining adaptability through conditional updating based on statistical property changes.
3Measurement precision
If contexts are updated frequently for each block, then the accuracy of coded probability information improves, but the processing complexity and computational load increase
Solution Approach 1:
The patent applies local quality by updating context models selectively based on changes in statistical properties of specific blocks. Instead of uniform updating across all blocks, the system evaluates whether neighboring blocks have similar statistical characteristics and updates contexts only when necessary, reducing processing complexity while maintaining accuracy where needed.
Solution Approach 2:
The patent uses partial updating of context models rather than complete re-evaluation for every block. By determining that certain neighboring blocks can share contexts without significant loss of accuracy, the system performs partial updates that reduce computational load while maintaining sufficient coding accuracy.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Provided are an image coding method, an image coding apparatus, an image coding apparatus, an image decoding method, an image decoding apparatus, and image coding and decoding apparatus. The image coding method comprising: obtaining current signals to be coded of each of the processing units of the image data (S401); generating a binary signal by performing binarization on each of the current signals to be coded (S402); selecting a context for each of the current signals to be coded from among a plurality of contexts (S403); performing arithmetic coding of the binary signal by using coded probability information associated with the context selected in the selecting (S404); and updating the coded probability information based on the binary signal (S405), wherein, in the selecting, the context for the current signal to be coded is selected, as a shared context, for a signal which is included in one of a plurality of processing units and has a size different from a size of the processing unit including the current signal to be coded.