Motion Estimation Using Combined Reference Bi-Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video encoding methods for motion estimation in bi-predictive slices are inefficient, especially in the presence of cross-fades, as they consider each candidate reference separately, leading to potential loss in coding efficiency and computational complexity.
Innovation Solution
A method and apparatus for motion estimation using combined reference bi-prediction, where motion vectors for one reference picture are predicted while initializing others to a predefined value, allowing for parallel iterations to optimize remaining hypotheses and refine motion vectors based on a best reference, thereby improving estimation accuracy and reducing complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If motion estimation considers each candidate reference separately, then device complexity is reduced, but coding efficiency deteriorates
Solution Approach 1:
The patent segments the motion estimation process into two distinct phases: a first pass where each reference picture is processed independently to obtain initial motion vectors, and a second pass where these vectors are refined by considering the combined effect of multiple references. This segmentation allows the system to balance computational load while achieving better coding efficiency through the refinement stage.
Solution Approach 2:
The patent applies preliminary action by performing a first pass of motion estimation that obtains initial motion vectors for each reference picture before proceeding to the refinement stage. This preliminary estimation provides a foundation that reduces the search space in the second pass, thereby improving overall efficiency without requiring complete joint optimization from the start.
2Measurement precision
If motion estimation jointly optimizes multiple reference pictures, then motion estimation accuracy is improved, but device complexity increases
Solution Approach 1:
The patent divides the complex joint optimization problem into two manageable passes: an initial estimation phase that processes references separately, and a refinement phase that jointly optimizes the combined prediction. This segmentation makes the overall process computationally tractable while still achieving high accuracy through the collaborative refinement of motion vectors.
Solution Approach 2:
The patent performs preliminary motion estimation for each reference picture independently before conducting the joint refinement. This preliminary action establishes initial motion vectors that serve as starting points for the subsequent optimization, reducing the computational burden of the joint optimization while maintaining accuracy.
3Productivity
If bi-predictive coding uses linear combination of two predictions, then coding efficiency is improved, but loss of information increases due to fade and cross-fade artifacts
Solution Approach 1:
The patent employs feedback by using the results from the first pass of motion estimation to inform and guide the second pass of refinement. The initial motion vectors obtained from separate reference processing are fed into the refinement stage, where they are adjusted based on the actual prediction error and the combined reference information, thereby reducing loss of information while maintaining coding efficiency.
Solution Approach 2:
The patent performs preliminary motion estimation using separate reference processing before applying the linear combination for bi-prediction. This preliminary action provides initial motion vectors that are then refined to account for fade and cross-fade artifacts, reducing prediction error while maintaining the benefits of linear combination.
Data Source
AI summary
A method and apparatus are provided for motion estimation using combined reference bi-prediction. The apparatus includes an encoder (200) for encoding a multi-prediction picture from a combination of two or more reference pictures by respectively predicting a motion vector for a particular one of the two or more reference pictures in a motion estimation process while initializing motion vectors for remaining ones of the two or more reference pictures to a predefined value for use by the motion estimation process.