Motion Compensated Residual Prediction for Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding systems face inefficiencies in residual prediction, particularly due to the difficulty in finding suitable predictors for residual blocks with mismatched sign values and the impact of in-loop filtering, which affects prediction quality and transform coding performance.
Innovation Solution
The proposed systems and methods enhance residual prediction by using motion compensated residual prediction (MCRP) that generates adaptive residual reference pictures, applies de-noising filters, and optimizes encoder search criteria, allowing for improved prediction efficiency across both inter and intra coded blocks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If motion compensated prediction is used for residual prediction, then prediction efficiency is improved, but handling residual blocks with mismatched sign values becomes difficult
Solution Approach 1:
The patent segments the residual prediction process into multiple stages: generating residual reference pictures from reference frames, filtering these pictures to handle sign value mismatches, and then using the filtered pictures for motion compensated prediction. This segmentation allows each stage to address specific problems independently, improving overall handling of mismatched sign values while maintaining prediction efficiency.
Solution Approach 2:
The patent introduces residual reference pictures as an intermediary between the reference frames and the current block prediction. These intermediary pictures are specifically processed to handle sign value mismatches through filtering operations, acting as a mediator that resolves the contradiction between maintaining prediction efficiency and handling sign value difficulties.
2Manufacturing precision
If in-loop filtering is applied to reference pictures, then prediction quality is improved, but transform coding performance is affected
Solution Approach 1:
The patent applies filtering selectively to residual reference pictures based on local characteristics. By analyzing the specific regions and applying filtering only where needed (rather than uniformly across all reference pictures), the system improves prediction quality in problematic areas while minimizing the impact on transform coding performance in areas where filtering is not applied.
Solution Approach 2:
The patent employs partial filtering actions on residual reference pictures rather than complete filtering. By applying filtering only to specific residual reference pictures or specific regions within them, the system achieves sufficient prediction quality improvement without the excessive filtering that would degrade transform coding performance.
3Measurement precision
If adaptive residual reference picture generation is used, then prediction accuracy is improved, but storage requirements increase
Solution Approach 1:
The patent performs preliminary generation and filtering of residual reference pictures in advance, before they are needed for prediction. This allows the system to prepare high-accuracy reference data beforehand and then store only the essential filtered results, reducing the storage burden while maintaining the accuracy benefits of adaptive generation.
Solution Approach 2:
The patent creates residual reference pictures as simplified copies or representations of the actual residual data, rather than storing complete high-precision residual information. These copied reference pictures maintain sufficient prediction accuracy while occupying significantly less storage space than the original detailed residual data.
Data Source
AI summary
Systems and methods are disclosed for improving the prediction efficiency for residual prediction using motion compensated residual prediction (MCRP). Exemplary residual prediction techniques employ motion compensated prediction and processed residual reference pictures. Further disclosed herein are systems and methods for generating residual reference pictures. These pictures can be generated adaptively with or without considering in-loop filtering effects. Exemplary de-noising filter designs are also described for enhancing the quality of residual reference pictures, and compression methods are described for reducing the storage size of reference pictures. Further disclosed herein are exemplary syntax designs for communicating residuals' motion information.