Template-Matched Block Prediction With Filtered Reference Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding and decoding technologies suffer from low prediction accuracy due to incomplete consideration of deviations in template matching, leading to inefficiencies in encoding and decoding processes.
Innovation Solution
Implement a method that involves determining a first template of a current block, a matching template, and a reference block, filtering the reference block using a model parameter, and determining prediction values based on the filtered reference block to improve prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If template matching prediction is used to find a matching template with minimum cost value, then prediction speed is improved, but prediction accuracy deteriorates due to large deviations in some scenarios
Solution Approach 1:
A filter model is introduced as an intermediary component between the template matching prediction and the final prediction value. The filter model processes the reference block to generate a filtered reference block, which then serves as the basis for determining the prediction value. This intermediary filtering step corrects deviations and improves prediction accuracy while maintaining the efficiency of template matching.
Solution Approach 2:
The patent applies filter coefficients as parameters to modify the reference block. By changing the parameters of the reference block through filtering operations with learnable filter coefficients, the system adapts the prediction to better match the actual content, thereby improving prediction accuracy without sacrificing the speed advantage of template matching.
2Device complexity
If reconstructed samples of the best matching reconstructed block are directly used as predicted samples, then device complexity is reduced, but prediction accuracy deteriorates due to incomplete consideration of deviations
Solution Approach 1:
The filter model serves as an intermediary that processes the reference block before it is used as the prediction value. This adds a filtering stage that improves accuracy without significantly increasing overall system complexity, as the filter operates on the existing reference block data rather than requiring completely new processing paths.
Solution Approach 2:
The filter model performs preliminary processing on the reference block to account for deviations before the final prediction is made. By preparing and filtering the reference block in advance, the system pre-corrects for known deviations, improving accuracy without requiring complex real-time adjustments.
Data Source
AI summary
The embodiments of the present invention provide an encoding method which includes: determining a first template of a current block, and determining a matching template and a reference block; determining a model parameter according to the first template and the matching template; filtering the reference block according to the model parameter, to determine a filtered reference block; determining a prediction value of the current block according to the filtered reference block; and determining a reconstructed value of the current block according to the prediction value of the current block.