Classification-Based Post-Filter Scaling for Video Compression Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding systems face challenges in efficiently adapting post-filter corrections to improve video quality, particularly in classification-based systems, leading to suboptimal compression and transmission efficiency.
Innovation Solution
Implementing classification-based post-filter re-scaling with adaptive scaling parameters, utilizing existing classifiers to determine corrections for reconstructed picture samples, and applying neural network-based filtering to enhance video coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If classification-based post-filter correction is applied to improve video quality, then manufacturing precision is improved, but device complexity increases
Solution Approach 1:
The patent applies parameter changes by introducing scaling parameters (scale and offset) that modify the post-filter correction values based on classification. Different scaling parameters are applied to different picture regions or blocks, allowing adaptive adjustment of correction strength without fundamentally changing the filter structure. This resolves the contradiction by improving video quality through parameter adaptation while maintaining the existing filter architecture.
Solution Approach 2:
The patent segments the picture into multiple regions or blocks, each potentially receiving different scaling parameters based on local characteristics. This segmentation allows the system to apply simplified correction strategies to specific regions rather than uniformly complex processing across the entire picture, thereby improving overall quality while managing computational complexity through localized processing.
2Productivity
If adaptive scaling parameters are introduced to optimize post-filter correction, then productivity is improved, but device complexity increases
Solution Approach 1:
The patent implements preliminary action by pre-calculating and storing scaling parameters in lookup tables before the actual video coding process. These parameters are determined based on pre-established criteria and classification schemes, allowing the filtering system to quickly retrieve and apply appropriate scaling factors during encoding/decoding without performing complex real-time calculations. This improves coding efficiency while limiting the increase in system complexity.
Solution Approach 2:
The patent replaces complex mechanical computation with simplified table-based retrieval and basic arithmetic operations. Instead of performing elaborate real-time optimization calculations, the system substitutes pre-computed scaling parameters from tables, reducing computational burden and improving productivity while maintaining acceptable filtering performance.
3Loss of information
If classification-based re-scaling is applied to improve compression efficiency, then loss of information is reduced, but device complexity increases
Solution Approach 1:
The patent uses parameter changes by applying scaling factors to correction values based on classification results. This allows the system to preserve more information in regions where it matters most (e.g., high-frequency areas or important picture regions) while applying simpler correction in less critical areas, thereby reducing overall information loss without requiring uniformly complex processing across the entire video stream.
Data Source
Figure 1A
Figure 1B
Figure 1C
AI summary
A codec may perform classification-based post-filter re-scaling. Scaling parameters may be adapted to associated classifications. Scaling parameters support modification of the post-filter correction. A classifier may re-use existing classifiers used by the post-filter or another classifier. An adaptive classifier may support: classification-based scaling correction; neural network (NN)-based post-filtering with classification-based scaling correction; and/or signaling of the scaling parameters. A codec may determine a classification associated with a reconstructed picture sample. Scaling parameters associated with the classification may be determined. A scaled correction may be determined based on the scaling parameters. The scaled correction may be applied to the reconstructed picture sample. A codec may perform neural network processing to determine a classification associated with a reconstructed picture sample. Scaling parameters associated with the classification may be determined. A scaled correction may be determined based on the scaling parameters. The scaled correction may be applied to the reconstructed picture sample.