Context Modeling for Video Coding Syntax Elements
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing context modeling in video and image coding, such as in the HEVC standard, suffers from data dependency issues, particularly in the coding of syntax elements like transform coefficients, which limits coding efficiency and throughput.
Innovation Solution
A new context modeling method that determines a selected context for a syntax element based on the statistics of other syntax elements within a predefined region, allowing for context selection independent of previously decoded bins, thereby reducing data dependency and enabling parallel processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional context modeling based on coded bins is used, then coding efficiency is maintained, but data dependency exists during the coding process
Solution Approach 1:
The patent extracts the data dependency issue from the conventional context modeling process by separating context selection from previous bin decoding. Instead of relying on previously decoded bins to determine context, the invention uses a predefined context set where the context is determined by syntax element position and value ranges, effectively removing the harmful data dependency while preserving coding efficiency.
Solution Approach 2:
The patent segments the context modeling process into distinct components: a predefined context set, position-based context selection, and value-range-based context selection. This segmentation allows parallel processing of different syntax elements since each can independently select its context without waiting for previous elements to be decoded, thus resolving the data dependency issue.
2Reliability
If sequential processing of syntax elements is used, then data dependency is maintained for accurate context modeling, but throughput is limited
Solution Approach 1:
The patent divides the context modeling task into independent segments where each syntax element can select its context independently based on its position and value characteristics. This segmentation enables parallel processing of multiple syntax elements simultaneously, increasing throughput while maintaining context modeling accuracy through the use of a comprehensive predefined context set.
Solution Approach 2:
The patent performs preliminary action by pre-defining a context set before the actual coding process begins. This context set contains all possible contexts organized by position and value range, allowing syntax elements to immediately select appropriate contexts without sequential dependency, thereby enabling parallel processing while maintaining modeling accuracy.
3Measurement precision
If context selection depends on previously decoded bins, then context accuracy is improved, but parallel processing is prevented
Solution Approach 1:
The patent extracts the dependency on previously decoded bins from the context selection process. Instead of using previous bins to determine context, the invention uses a predefined context set where context is selected based on the current syntax element's position and value characteristics, maintaining accuracy while enabling parallel processing.
Solution Approach 2:
The patent introduces an intermediary mechanism - the predefined context set - that mediates between the need for accurate context selection and the desire for parallel processing. This context set acts as a lookup table that provides accurate context selection based on position and value range without requiring sequential decoding of previous elements, thus enabling both accuracy and parallelism.
Data Source
AI summary
A method and apparatus of video and image coding using context modeling for video and image are disclosed. A selected context is determined from a context set for a first syntax element at a first location based on statistics of one or more second syntax elements at one or more second locations within a predefined region. The first syntax element is different from the second syntax element. The first syntax element is then encoded or decoded according to the selected context. In one embodiment, the context modelling is applied to transform coefficients. The first syntax element corresponds to coeff_abs_level_greater1_flag and the second syntax element corresponds to significant_coeff_flag.