Arithmetic Decoding Context Update Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional arithmetic decoding methods experience processing delays due to dependencies between context update and context load processes, limiting parallelization and reducing execution time efficiency.
Innovation Solution
The proposed method involves performing context update processes in segments of certain sizes within processing units, allowing for parallelization of arithmetic decoding processes without depending on selected contexts, and executing these updates in an order inverse to the decoding order to optimize symbol occurrence probability usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If conventional arithmetic decoding is performed with sequential context update and context load processes, then coding accuracy is maintained, but processing speed decreases due to processing delays and inability to parallelize
Solution Approach 1:
The decoding process is divided into multiple processing units, with each unit handling a specific segment of decoding tasks. Context update and context load processes are segmented and can be executed in parallel across different processing units, eliminating sequential processing delays while maintaining coding accuracy through consistent application of the inverse decoding order principle
Solution Approach 2:
The context update process is executed in an order inverse to the decoding order of signals. This inversion allows the context update to be performed independently of the decoding sequence, enabling parallel execution without affecting the accuracy of symbol occurrence probability calculations, thus resolving the contradiction between speed and time loss
2Productivity
If parallel processing is implemented in conventional arithmetic decoding, then processing speed increases, but coding efficiency decreases due to dependencies between context update and context load processes
Solution Approach 1:
By executing context update in an order inverse to the decoding order, the patent decouples the dependency between context update and context load processes. This allows parallel processing implementation while maintaining coding efficiency, as each processing unit can independently update contexts without waiting for sequential updates from other units
Solution Approach 2:
Context updates are performed in advance for subsequent decoding operations by executing them in inverse order. This preliminary action ensures that when parallel decoding occurs, the necessary context information is already prepared, maintaining coding efficiency while enabling parallel processing throughput
Data Source
AI summary
An image decoding method includes: arithmetic decoding steps for performing arithmetic decoding processes on decoding target signals according to contexts determined based on the types of the decoding target signals and the symbol occurrence probabilities determined based on the contexts; and a context update step for executing a group of context update processes on the decoding target signals in each of processing units each obtained as a segment having a certain size, according to the decoded signals resulting from the arithmetic decoding processes.