Spatial-Temporal Merge Candidates for Accurate Inter-Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression technologies face challenges in efficiently predicting pixel values across high-resolution and high-quality videos, particularly in constructing accurate merge candidate lists and performing inter-prediction, which affects encoding/decoding performance and accuracy.
Innovation Solution
The method and apparatus enhance inter-prediction by constructing a merge candidate list that includes spatial, temporal, and combined merge candidates, deriving motion information through weighted averages, and using affine models to improve prediction accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing inter-prediction technology is used, then video compression can be achieved, but prediction accuracy for high-resolution and high-quality videos is insufficient
Solution Approach 1:
The patent segments the prediction process by dividing merge candidates into distinct categories (spatial merge candidates from neighboring blocks, temporal merge candidates from temporal neighbors, and combined merge candidates). This segmentation allows each candidate type to be processed and evaluated independently, improving prediction accuracy for high-resolution videos while maintaining encoding efficiency through structured evaluation procedures.
Solution Approach 2:
The patent merges multiple prediction sources by combining spatial merge candidates (from spatial neighboring blocks), temporal merge candidates (from temporal neighboring blocks), and combined merge candidates into a unified merge candidate list. This merging approach leverages both spatial and temporal information to achieve more accurate predictions for high-quality video encoding while maintaining efficient processing through integrated evaluation.
2Measurement precision
If merge candidate list construction is simplified, then encoding speed increases, but prediction accuracy decreases
Solution Approach 1:
The patent applies local quality by constructing different types of merge candidates based on local characteristics - spatial merge candidates use information from spatially neighboring blocks, temporal merge candidates use information from temporally neighboring blocks, and combined merge candidates integrate both. This localized approach to quality construction improves prediction accuracy without requiring uniform complex processing across all candidates, thus managing encoding time effectively.
Solution Approach 2:
The patent performs preliminary action by pre-constructing merge candidates from neighboring blocks before the actual prediction process. Spatial merge candidates are derived from spatial neighbors, temporal merge candidates from temporal neighbors, and combined candidates are prepared in advance. This preliminary construction reduces the computational burden during the main prediction process, improving accuracy while controlling encoding time.
3Measurement precision
If motion information derivation uses complex methods, then prediction accuracy improves, but computational complexity increases
Solution Approach 1:
The patent uses copying by deriving motion information through copying and adapting existing motion data from neighboring blocks. Spatial merge candidates copy motion information from spatial neighbors, temporal merge candidates copy from temporal neighbors, and combined candidates integrate both sources. This copying approach achieves accurate motion information derivation without requiring complex computation, as it leverages existing data rather than calculating from scratch.
Solution Approach 2:
The patent applies parameter changes by adjusting how motion information is derived based on the type of merge candidate. Different weighting factors and combination methods are applied when deriving motion information from spatial versus temporal candidates, or when combining both. These parameter adjustments optimize prediction accuracy for different video scenarios while keeping the overall computational complexity manageable through systematic parameter application.
Data Source
AI summary
A video encoding/decoding method is provided, which includes constructing a merge candidate list of a current block, deriving motion information of the current block from the merge candidate list, and performing inter-prediction of the current block using the motion information, where the merge candidate list includes at least one of a spatial merge candidate, a temporal merge candidate, or a combined merge candidate, and the combined merge candidate is derived by combining n merge candidates belonging to the merge candidate list. A video encoding/decoding apparatus is also provided.