Line Graph Transform Cores for AV1 Coding Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding and decoding technologies, such as AV1, lack efficient transform cores for certain transforms like the 8-point line graph transform, limiting coding efficiency and accuracy.
Innovation Solution
Introduce a complete set of 8-bit and 10-bit transform cores, including 4-point, 8-point, 16-point, and 32-point primary transforms, optimized for better coding efficiency and accuracy, using line graph transforms (LGTs) and discrete sine transforms (DSTs).
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If existing video encoding technologies (AV1) are used, then basic encoding functionality is provided, but coding efficiency and accuracy are limited due to lack of efficient transform cores for certain transforms
Solution Approach 1:
The patent segments the transform functionality into distinct transform cores for different transform types (4-point, 8-point, 16-point, 32-point primary transforms) and bit depths (8-bit and 10-bit). This segmentation allows each transform core to be optimized independently for its specific function, improving overall coding efficiency while maintaining manageable complexity through modular design.
2Measurement precision
If a complete set of transform cores is implemented, then coding accuracy is improved, but device complexity increases
Solution Approach 1:
The patent implements universal transform cores that can handle multiple transform types and bit depths through a unified architecture. The transform cores are designed to process 4-point, 8-point, 16-point, and 32-point transforms, as well as both 8-bit and 10-bit data, using the same hardware structure. This multi-functionality improves coding accuracy by providing complete transform support while avoiding the complexity of implementing separate dedicated cores for each transform variant.
Data Source
AI summary
A method, computer program, and computer system is provided for coding video data. Video data is received. One or more transform cores corresponding to a transform associated with the video data are identified. The one or more transform cores include one or more of a line graph transform (LGT) and a discrete sine transform (DST) The video data is decoded based on the identified transform core. The transform cores correspond to one or more from among an 8-bit transform core and a 10-bit transform core. The transform corresponds to one or more from among a 2-point transform, a 4-point transform, an 8-point transform, and a 16-point transform.