Motion Estimation for Frame Rate Conversion Using Reliable Vector Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional motion estimation methods, such as the 3DRS algorithm, often select incorrect motion vectors, leading to inaccurate interpolation in frame rate conversion, and calculate vectors in a predetermined order, which can result in incorrect image data in interpolation frames.
Innovation Solution
A motion estimation method that generates unidirectional motion vectors, classifies them as reliable or mismatch vectors, maps reliable vectors onto interpolation frames, and determines assigned vectors for high confidence blocks based on their reliability and confidence, while using a non-sequential 3DRS method to calculate bidirectional vectors after high confidence blocks are processed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If the 3DRS algorithm is used to calculate motion vectors in a predetermined order, then the processing sequence is simple and systematic, but the accuracy of motion vector selection deteriorates leading to wrong MVs and incorrect image data in interpolation frames
Solution Approach 1:
The patent applies preliminary action by first classifying blocks into high-confidence and low-confidence categories before performing motion estimation. High-confidence blocks (with reliable unidirectional MVs) are processed first to establish accurate reference points, while low-confidence blocks are processed later using information from neighboring blocks. This preliminary classification and selective processing order improves overall MV accuracy without significantly complicating the processing framework.
2Extent of automation
If the cost function is applied to select motion vectors from multiple candidates, then the selection process is automated, but the reliability of MV selection deteriorates due to wrong MVs being selected
Solution Approach 1:
The patent generates unidirectional motion vectors from reference frames before the main motion estimation process. These unidirectional MVs serve as preliminary results that are classified as reliable or unreliable based on matching criteria. This preliminary generation and classification automates the filtering process while improving reliability by excluding obviously wrong MVs before the automated cost function selection is applied to remaining candidates.
Solution Approach 2:
The patent implements feedback by using motion vectors from already-processed high-confidence blocks to improve the estimation for low-confidence blocks. The system feeds back the results from reliable blocks to inform the motion estimation of unreliable blocks, creating a progressive refinement process that enhances overall MV selection reliability through iterative improvement.
Data Source
AI summary
A motion estimation method for a frame rate converter includes generating a plurality of unidirectional motion vectors (MVs) based on a first original frame and a second original frame; classifying each of the plurality of unidirectional MVs as a reliable MV or a mismatch MV; mapping the reliable MV onto one of a plurality of blocks of an interpolation frame; classifying each of the plurality of blocks as a high confidence block or a low confidence block based on whether there is at least one reliable MV mapped onto the block; and determining an assigned MV for each of the high confidence block by selecting one of the at least one reliable MV mapped onto the high confidence block.