LIC Parameter Derivation in Mapped Domain Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing video decoding technologies face increased decoding latency due to the need to load an inverse mapping table at the coding unit level for local illumination compensation (LIC) when both LIC and luma mapping with chroma scaling (LMCS) are enabled, which adds hardware implementation burden and prolongs the motion compensation stage.
Innovation Solution
Deriving LIC parameters based on a mapped domain reference template block and a mapped domain neighboring reconstruction template block, eliminating the need to read the inverse mapping table from memory, thereby reducing processing power and latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If inverse mapping table is loaded from memory for LIC parameter derivation in LMCS, then LIC parameter derivation can be performed, but decoding latency increases and hardware implementation burden increases
Solution Approach 1:
The patent extracts and removes the inverse mapping table loading step from the LIC parameter derivation process. By deriving LIC parameters directly in the mapped domain without converting back to the original pixel domain, the inverse mapping table is no longer needed, eliminating the memory access bottleneck and reducing decoding latency.
Solution Approach 2:
The patent segments the processing into distinct domain operations: LMCS transformation is performed first to map pixels to the mapped domain, then LIC parameter derivation is performed independently in the mapped domain. This segmentation allows each operation to work with data in its optimal domain without requiring intermediate conversions.
2Reliability
If inverse mapping table is loaded from memory for LIC parameter derivation in LMCS, then LIC parameter derivation can be performed, but hardware implementation complexity increases
Solution Approach 1:
The patent removes the inverse mapping table and its associated memory access logic from the hardware implementation. By performing LIC parameter derivation directly in the mapped domain, the hardware no longer requires large lookup tables or complex memory management, significantly reducing implementation complexity.
3Reliability
If inverse mapping table is loaded from memory during motion compensation stage, then LIC can be applied, but processing power is consumed and stage duration is prolonged
Solution Approach 1:
The patent extracts the inverse mapping operation from the motion compensation stage entirely. By deriving LIC parameters in the mapped domain using mapped domain reference template blocks and neighboring reconstruction blocks, the motion compensation stage no longer requires memory-intensive table lookups, improving processing efficiency.
Data Source
AI summary
An example device for decoding video data includes memory configured to store the video data and one or more processors implemented in circuitry and communicatively coupled to the memory. The one or more processors are configured to reshape a pixel domain reference template block using a forward mapping function into a mapped domain reference template block and derive local illumination compensation (LIC) model parameters from the mapped domain reference template block and a mapped domain neighboring reconstruction template block. The one or more processors are configured to apply the LIC model parameters to motion-compensated prediction signals and decode the video data based on the application of the LIC model parameters.