Decoder Motion Refinement Using LIC and Template Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face inefficiencies in determining refined motion vectors due to illuminance differences and high-frequency noise, leading to less efficient coding and poorer quality decoded video.
Innovation Solution
Implement decoder-side motion vector refinement techniques, including local illumination compensation (LIC) and template matching, to improve the prediction signal of reference search areas in reference pictures L0 and L1, enhancing the accuracy of motion vector derivation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional motion vector derivation methods are used, then the coding process is simpler, but the decoded video quality deteriorates due to illuminance differences and noise
Solution Approach 1:
The patent applies local illumination compensation (LIC) to the search area before performing template matching. This preliminary action corrects illuminance differences in advance, allowing the subsequent template matching to operate on corrected reference data, thereby improving motion vector accuracy without requiring complex real-time adjustments during the matching process itself.
Solution Approach 2:
The patent introduces an intermediary processing step (LIC model application) between the reference picture generation and template matching processes. This intermediary corrects the reference search area data to compensate for illuminance variations and noise, enabling more accurate motion vector derivation without directly complicating the core template matching algorithm.
2Measurement precision
If template matching is applied without illumination compensation, then the coding process is faster, but the motion vector determination accuracy deteriorates
Solution Approach 1:
The LIC model is applied to the search area in advance before template matching begins. This preliminary illumination compensation prepares corrected reference data that accounts for lighting variations, enabling the template matching to achieve higher accuracy without requiring multiple iterative adjustments that would slow down the encoding process.
3Reliability
If search area refinement with LIC is performed, then the prediction signal quality improves, but the processing time increases
Solution Approach 1:
The patent applies illumination compensation specifically to the search area rather than the entire picture. This local processing approach improves the prediction signal quality where it matters most (in the motion estimation region) while minimizing the additional processing time by limiting the LIC application to only the necessary regional area.
Data Source
AI summary
An example device includes one or more processors configured to determine a first search area in a first reference picture for a current block of the video data. The one or more processors are configured to determine a first initial reference block in the first search area. The one or more processors are configured to apply a local illumination compensation model to the first search area to generate a refined first search area. The one or more processors are configured to apply template matching to the refined first search area to determine a first candidate motion vector having a lowest template matching cost for the refined first search area. The one or more processors are configured to decode the current block based on the first candidate motion vector.