Neural Network SAO Video Coding Offset Classification

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing video coding technologies face challenges in efficiently compressing video data while maintaining video quality, particularly due to limited bandwidth and memory resources.

Innovation Solution

The implementation of a neural network-based Sample Adaptive Offset (SAO) system for video coding, which classifies reconstructed samples into categories, determines offsets for these categories, and applies SAO filtering to improve video coding efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of energy

If traditional video coding methods are used, then compression is achieved, but video quality deteriorates

Engineering Contradiction:

Improvecompression efficiencyVSAvoidvideo quality

Core Design Contradiction:

Loss of energyVSManufacturing precision

Solution Approach 1:

The patent replaces traditional mechanical filtering systems with a neural network-based SAO system that learns nonlinear mapping relationships between original and reconstructed images, enabling more precise correction of artifacts while maintaining compression efficiency

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent introduces Sample Adaptive Offset parameters that are dynamically adjusted based on sample classification categories, allowing the system to optimize both compression efficiency and video quality by adapting offset values to different regions and characteristics of the video data

Inventive Principle:

Principle #35Parameter changes

2Manufacturing precision

If neural network based SAO is applied, then video quality improves, but processing complexity increases

Engineering Contradiction:

Improvevideo qualityVSAvoidprocessing complexity

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent segments the video processing task into distinct stages: neural network-based in-loop filtering for quality enhancement, sample classification into categories, and selective SAO offset application, making the complex neural network processing more manageable and efficient

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies different SAO offset values to different categories of reconstructed samples based on their local characteristics, allowing the system to maintain high video quality in critical regions while reducing processing complexity in less important areas

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS12309364B2System and method for applying neural network based sample adaptive offset for video coding

Publication Date: 2025.05.20 BEIJING DAJIA INTERNET INFORMATION TECH CO LTD

AI summary

Embodiments of the disclosure provide systems and methods for applying neural network based sample adaptive offset (SAO) for video coding. The method may include classifying reconstructed samples of a reconstructed block into a set of categories based on neural network based in-loop filtering (NNLF). The reconstructed block includes a reconstructed version of a video block of a video frame from a video. The method may further include determining a set of offsets for the set of categories based on the classification of the reconstructed samples. The method may additionally include, responsive to the NNLF being performed on the reconstructed block, performing SAO filtering on the NNLF filtered samples based on the set of offsets. The NNLF filtered samples are generated from the reconstructed samples using the NNLF.