Inter Prediction Motion Compensation With Adaptive Uni/Bi Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image compression technologies face challenges in reducing calculation complexity and improving encoding efficiency, particularly in high-definition and ultra-high-definition image processing.
Innovation Solution
The proposed method involves deriving motion information for inter prediction that includes both L0 and L1 motion information, allowing for optional uni-prediction or bi-prediction, and applying weighting averages or motion compensation based on the similarity of L0 and L1 motion information to reduce calculation complexity and improve encoding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If both L0 and L1 motion information are derived and bi-prediction is performed, then prediction accuracy is improved, but calculation complexity increases
Solution Approach 1:
The patent dynamically selects between uni-prediction and bi-prediction based on whether L0 and L1 motion information are identical. When motion information matches, the system switches to simpler uni-prediction; when they differ, it employs bi-prediction. This dynamic adaptation resolves the contradiction by adjusting prediction complexity according to actual motion characteristics.
Solution Approach 2:
The patent changes the prediction mode parameter based on motion information comparison. By monitoring whether L0 and L1 motion vectors are identical, the system adjusts the prediction approach (uni vs bi-prediction), thereby optimizing the balance between accuracy and computational load through parameter-based adaptation.
2Manufacturing precision
If motion compensation is performed for both L0 and L1 reference pictures, then encoding precision is improved, but processing time increases
Solution Approach 1:
The patent dynamically adjusts the number of reference pictures processed by comparing L0 and L1 motion information. When motion information is identical, only one reference picture undergoes motion compensation, reducing processing time. When they differ, both reference pictures are processed for higher precision, thus dynamically balancing speed and accuracy.
Solution Approach 2:
The patent extracts and compares motion information from L0 and L1 references, then selectively applies motion compensation only when necessary. By separating the motion information comparison step from the compensation step, the system avoids redundant processing and reduces unnecessary computational overhead while maintaining encoding precision when needed.
3Reliability
If multiple reference pictures are used for prediction, then prediction reliability is improved, but device complexity increases
Solution Approach 1:
The patent dynamically adjusts the number of active reference pictures based on motion information comparison. When L0 and L1 motion information are identical, the system uses only one reference picture, simplifying device operations. When they differ, it activates both references for improved reliability, thus dynamically adapting device complexity to actual prediction needs.
Solution Approach 2:
The patent changes the reference picture selection parameter based on motion vector comparison results. By using motion information as a control parameter, the system adjusts how many reference pictures are utilized, optimizing the balance between prediction reliability and device operational complexity through parameter-based control.
Data Source
AI summary
An inter-prediction method according to the present invention comprises the steps of: deriving motion information of a current block; and generating a prediction block for the current block on the basis of the derived motion information. According to the present invention, computational complexity can be reduced and encoding efficiency can be improved.