Adaptive Rice Parameter Derivation for High Bit-Depth Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High bit-depth video coding in existing standards like VVC faces limitations due to insufficient ranges of Rice parameter values, leading to a drop in coding performance.
Innovation Solution
Adaptive derivation of Rice parameter values by calculating a local sum of absolute values, normalizing it, and using the normalized value to determine an extended Rice parameter range, which is then applied for high bit-depth video coding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing Rice parameter ranges are used in high bit-depth video coding, then device complexity is kept simple, but coding performance deteriorates due to insufficient parameter ranges
Solution Approach 1:
The Rice parameter derivation is made dynamic by introducing adaptive normalization based on locSumAbs values. The shift value is derived from the actual data characteristics (locSumAbs) rather than using fixed ranges, allowing the parameter to adapt to high bit-depth scenarios while maintaining a relatively simple derivation process.
Solution Approach 2:
The invention changes the parameter derivation approach by introducing a normalization step with shift values based on locSumAbs. This transforms the fixed Rice parameter range into an adaptive range that scales with the local sum of absolute values, enabling effective high bit-depth coding without significantly increasing complexity.
2Reliability
If Rice parameter range is extended for high bit-depth coding, then coding performance improves, but device complexity increases due to additional normalization steps
Solution Approach 1:
The invention extends the Rice parameter range through parameter transformation rather than creating entirely new parameters. By normalizing locSumAbs with shift values and deriving Rice parameters from the normalized values, the system achieves extended range while reusing existing computational building blocks.
Solution Approach 2:
The normalization process serves dual purposes: it enables extended Rice parameter ranges for high bit-depth coding while simultaneously adapting to local data characteristics through locSumAbs calculation. The system uses its own data statistics (locSumAbs) to drive the normalization, making the process self-adapting rather than requiring external control.
Data Source
AI summary
An example device for binarizing video data includes a memory configured to store video data; and one or more processors implemented in circuitry and configured to: calculate a local sum of absolute values (locSumAbs value) of neighboring coefficients to a current coefficient of a current block of video data; derive a shift value from the locSumAbs value; normalize the locSumAbs value using the shift value; determine a Rice parameter using the normalized locSumAbs value; and binarize or inverse binarize the current coefficient using the Rice parameter. In this manner, these techniques may allow for more appropriate Rice parameter value selection when binarizing high bitdepth data in conjunction with performing context-adaptive binary arithmetic coding (CABAC).