Unified Scan Order for Transform Coefficients in Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding techniques face inefficiencies in scanning and coding transform coefficients, particularly in the High Efficiency Video Coding (HEVC) standard, due to mismatched scan orders for significance maps and coefficient levels, complex context sets, and inefficient data fetching processes.
Innovation Solution
Proposed techniques include harmonizing the scan order for significance maps and coefficient levels, scanning in sub-sets of transform coefficients, and using adaptive context models based on local neighborhoods for entropy coding with Context Adaptive Binary Arithmetic Coding (CABAC), allowing for unified inverse scan directions and reduced context sets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If different scan orders are used for significance maps and coefficient levels, then coding flexibility is improved, but device complexity and processing overhead increase
Solution Approach 1:
The patent merges the scan order for significance maps and coefficient levels into a single unified scan order. This eliminates the need to maintain and process separate scan orders, thereby reducing device complexity and processing overhead while preserving coding flexibility through the ability to select from different predefined scan order options.
Solution Approach 2:
The unified scan order serves multiple functions simultaneously: it is used for both significance map scanning and coefficient level scanning. This multi-functionality reduces the need for separate processing paths and data structures, simplifying the overall device architecture while maintaining adaptability through selectable scan order patterns.
2Measurement precision
If complex context sets are used for entropy coding, then coding precision is improved, but processing time and computational complexity increase
Solution Approach 1:
The patent segments the context sets into smaller, more manageable subsets based on the unified scan order. This segmentation reduces the computational complexity of processing each context set while maintaining coding precision by preserving the essential statistical relationships needed for accurate entropy coding.
Solution Approach 2:
The patent changes the parameters of the context sets by reorganizing them according to the unified scan order, transforming the data structure to improve access efficiency and reduce processing time while maintaining the precision required for accurate entropy coding.
3Measurement precision
If separate scanning processes are used for significance maps and coefficient levels, then coding accuracy is improved, but data fetching efficiency deteriorates
Solution Approach 1:
The patent combines the scanning processes for significance maps and coefficient levels into a single unified scanning operation. This merging eliminates the need for separate data fetching operations, improving efficiency by reducing memory access overhead and simplifying the data flow while maintaining coding accuracy through the preserved statistical relationships.
4Device complexity
If unified scan order is used for significance maps and coefficient levels, then device complexity is reduced, but adaptability to different coding scenarios may be limited
Solution Approach 1:
The patent introduces dynamic selection capability that allows the unified scan order to be adaptively chosen based on the specific coding scenario. This dynamic approach maintains low device complexity by using a single scan order structure while preserving adaptability through the ability to select from multiple predefined scan order patterns depending on the content characteristics.
Data Source
AI summary
In an example, a method includes coding significance information for transform coefficients in a set of transform coefficients associated with the residual video data according to a scan order in one or more first passes. The method also includes coding a first set of one or more bins of information for the transform coefficients according to the scan order in one or more second passes, where the first set of one or more bins are coded in a first mode of a coding process. The method also includes coding a second set of one or more bins of information for the transform coefficients according to the scan order in one or more third passes, where the second set of one or more bins are coded in a second mode of the coding process.