Hierarchical Motion Estimation via Adaptive Sampling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current motion estimation methods in video image compression are inefficient and inaccurate, particularly in reducing the complexity of motion prediction while maintaining prediction accuracy.
Innovation Solution
A hierarchical motion estimation method is employed by down-sampling images to improve correlations between frames, allowing for more efficient motion prediction using low-resolution image frames, which reduces complexity and enhances prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If motion estimation is performed on full-resolution frames, then motion prediction accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent divides the motion estimation process into multiple hierarchical levels. First, motion estimation is performed on down-sampled low-resolution frames to obtain coarse motion vectors. Then, refinement is performed on specific blocks at full resolution using these coarse vectors as initial estimates. This segmentation of the estimation process across different resolution levels reduces overall computational complexity while maintaining accuracy.
Solution Approach 2:
The patent performs preliminary motion estimation on down-sampled frames before processing full-resolution frames. The motion vectors obtained from low-resolution frames serve as preliminary estimates that guide subsequent refinement processes. This preliminary action at reduced resolution significantly reduces the search space and computational burden for the final high-precision estimation.
2Device complexity
If down-sampling is applied to reduce resolution, then computational complexity is reduced, but frame correlation deteriorates
Solution Approach 1:
The patent applies different processing qualities to different parts of the frame. Down-sampling is applied globally to reduce overall computational complexity, but local refinement is performed on specific blocks that require higher accuracy. This local quality enhancement maintains frame correlation for critical regions while still achieving overall complexity reduction through selective processing.
Data Source
Figure 1~3
Figure 4
Figure 5
AI summary
A hierarchical motion prediction apparatus and method. The hierarchical motion prediction method splits a current frame and a reference frame into pixel groups, changes a pixel location of each pixel group, and selects one pixel, and thus resolutions of the current frame and reference frame are reduced. A motion vector of a down-sampled current block is obtained based on a down-sampled current frame and reference frame, and is expanded to a motion vector of an original resolution based on a down sampling rate.