Adaptive Loop Filter Classification Using Geometry and HoG
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in accurately selecting adaptive loop filters based on the activity and directionality of video blocks, leading to increased distortion in decoded video data.
Innovation Solution
Implementing classifiers based on geometry partitioning and histogram of gradients to determine filter classes for video blocks, optimizing the selection of filters for different types of video content.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Laplacian-based classifiers are used for adaptive loop filtering, then the filtering process can be implemented, but the accuracy of determining activity and directionality of samples is insufficient, leading to distortion
Solution Approach 1:
The patent changes the classification parameters from Laplacian-based methods to geometry partitioning and histogram of gradients (HoG) features. This parameter transformation enables more accurate characterization of sample activity and directionality, directly improving measurement precision and reducing distortion in the filtered video data
Solution Approach 2:
The patent replaces the traditional Laplacian-based classification mechanism with alternative classification approaches (geometry partitioning and HoG). This substitution introduces new computational mechanisms that better capture the structural characteristics of video blocks, leading to improved filtering accuracy and reduced distortion
2Measurement precision
If multiple classifiers based on geometry partition and histogram of gradients are implemented, then filter selection accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent segments the classification process into multiple specialized classifiers (geometry partition-based and HoG-based), each handling specific aspects of block characterization. This segmentation allows for more accurate filter selection while organizing the computational complexity into manageable, modular components that can be processed efficiently
Solution Approach 2:
The patent designs classifiers that serve multiple functions: geometry partitioning simultaneously provides spatial organization and activity measurement, while HoG features provide both directional information and texture characterization. This multi-functionality improves filter selection accuracy without proportionally increasing computational burden
Data Source
AI summary
A method and apparatus for coding video data includes filtering video data using a loop filter, such as an adaptive loop filter (ALF), that is determined using a classifier from a plurality of classifiers. A video coder may reconstruct a block of video data to generate a reconstructed block, determine a filter class for the reconstructed block using a classifier from a plurality of classifiers, wherein the plurality of classifiers includes one or more of a first classifier based on a geometry partition and a second classifier based on a histogram of gradients, determine a filter based on the filter class, and apply the filter to the reconstructed block.