Adaptive Video Leveling Filtering with 7×7 Cross-Square Kernels
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing loop post-processing techniques in video coding suffer from poor filtering performance, leading to video quality distortion due to coding noises.
Innovation Solution
Implement a filtering method that adaptively selects between a 7*7 cross shape plus 5*5 square centrosymmetric filter and a 7*7 cross shape plus 3*3 square centrosymmetric filter based on the enablement flag, using pixel values from within or closest to the reference pixel for adaptive leveling filtering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing loop post-processing filtering techniques are used, then video coding is performed, but filtering performance is poor and video quality distortion occurs
Solution Approach 1:
The patent applies parameter changes by modifying the filter kernel size and shape dynamically. It switches between a 7×7 filter and a 5×5 filter based on the complexity of the current block, changing the filtering parameters to match different video content characteristics. This resolves the contradiction by adapting the filtering strength to the actual noise level and detail content, improving filtering performance while preventing quality distortion.
Solution Approach 2:
The patent implements dynamics by making the filtering process adaptive rather than static. The filter selection (7×7 or 5×5) changes dynamically based on the calculated complexity metric of each block. This dynamic adaptation allows the system to optimize filtering performance for each specific region, resolving the contradiction between effective noise reduction and preservation of video quality.
2Reliability
If a fixed filter size is used for adaptive leveling filtering, then the filtering process is simple, but filtering effectiveness is limited
Solution Approach 1:
The patent employs dynamics by transitioning from a fixed filter size to a dynamic filter selection mechanism. The system calculates a complexity metric for each block and automatically selects between 7×7 and 5×5 filters based on this metric. This dynamic approach improves filtering effectiveness while adding only moderate complexity through a straightforward complexity calculation and conditional selection process.
Solution Approach 2:
The patent applies parameter changes by varying the filter size parameter (7×7 or 5×5) based on the content characteristics. This parameter adaptation allows the system to optimize filtering effectiveness for different block types without requiring a complex array of different filter types, thus balancing effectiveness with acceptable complexity.
3Reliability
If enhanced adaptive leveling filtering with larger filter kernel is used, then filtering performance improves, but computational complexity increases
Solution Approach 1:
The patent uses dynamics to adapt the computational resources allocated to filtering based on actual need. By calculating block complexity first, the system dynamically selects the appropriate filter size - using the more computationally intensive 7×7 filter only when necessary (low complexity blocks), and the lighter 5×5 filter for blocks that don't require strong filtering. This dynamic resource allocation improves overall filtering performance while controlling computational complexity.
Solution Approach 2:
The patent applies parameter changes by adjusting the filter kernel size parameter based on block characteristics. This parameter adaptation allows the system to optimize the balance between filtering performance and computational complexity - using larger kernels (7×7) only when the content warrants it, and smaller kernels (5×5) otherwise, thus improving filtering performance without consistently incurring high computational costs.
Data Source
AI summary
Filtering methods, apparatus, and devices are provided. In one aspect, a filtering method includes: determining whether enhanced adaptive leveling filtering is enabled for a current adaptive leveling filtering unit; if it is determined that the enhanced adaptive leveling filtering is enabled for the current adaptive leveling filtering unit, performing adaptive leveling filtering on the current adaptive leveling filtering unit by using a first filter; if it is determined that the enhanced adaptive leveling filtering is disabled for the current adaptive leveling filtering unit, performing adaptive leveling filtering on the current adaptive leveling filtering unit by using a second filter. The first filter includes a centrosymmetric filter with a 7*7 cross shape plus a 5*5 square shape, and the second filter includes a centrosymmetric filter with a 7*7 cross shape plus a 3*3 square shape.