Asymmetric Interpolation Filter for Sub-Pel-Unit Pixel Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image encoding and decoding methods face inefficiencies in accurately performing motion compensation due to limitations in interpolation filters used for generating sub-pel-unit pixels, which affect the precision of inter prediction.
Innovation Solution
The implementation of a video decoding apparatus and corresponding encoding method that selectively chooses between symmetric and asymmetric interpolation filters based on sub-pel-unit interpolation locations, allowing for the use of odd- or even-number-tap filters to generate sub-pel-unit pixel values, thereby improving interpolation accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single interpolation filter is used for all sub-pel-unit locations, then the device complexity is reduced, but the interpolation accuracy deteriorates
Solution Approach 1:
The patent applies local quality by selecting different interpolation filters based on the specific sub-pel-unit interpolation location. For locations at 1/4 pixel position, one filter is used; for locations at 3/4 pixel position, a different filter is used. This localized adaptation of filter characteristics to specific spatial positions improves interpolation accuracy without requiring a completely complex system, as the selection is based on simple position classification.
2Measurement precision
If asymmetric interpolation filters are used, then the interpolation accuracy for sub-pel-unit pixels is improved, but the computational complexity increases
Solution Approach 1:
The patent employs asymmetric interpolation filters where the filter coefficients are not symmetric around the center tap. The filter coefficients are specifically designed with asymmetric values to better match the interpolation requirements at different sub-pel-unit locations. This asymmetry improves the accuracy of sub-pel-unit pixel generation by better modeling the actual interpolation needs, while the computational complexity is managed through pre-calculated fixed coefficients.
3Measurement precision
If interpolation is performed at sub-pel-unit precision, then the inter prediction accuracy is improved, but the processing time increases
Solution Approach 1:
The patent applies partial action by performing interpolation only at the necessary sub-pel-unit precision levels required for motion compensation, rather than uniformly across all pixels. The asymmetric interpolation filters are specifically designed to provide adequate accuracy at 1/4 and 3/4 pixel positions without requiring full sub-pixel precision for all cases. This selective approach maintains inter prediction accuracy where needed while reducing overall processing time by avoiding excessive interpolation computations.
Data Source
Figure 1~2
Figure 3~4B
Figure 4C
AI summary
A sub-pel-unit image interpolation method using a transformation-based interpolation filter includes, according to a sub-pel-unit interpolation location in a region supported by interpolation filters for generating at least one sub-pel-unit pixel value located between integer-pel-unit pixels, individually selecting a symmetric or asymmetric interpolation filter with respect to the interpolation location from among the interpolation filters; and generating the at least one sub-pel-unit pixel value by interpolating the integer-pel-unit pixels by using the selected interpolation filter.