NN-Based Loop Filtering for Higher Video Coding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video coding techniques suffer from low coding efficiency, which is undesirable for modern digital video applications.
Innovation Solution
Implementing a neural network (NN)-based loop filter for video processing, which includes applying a unified NN-based filter for rate-distortion optimization across different types of slices and color components, and using an NN-based in-loop filter with residual samples to enhance coding effectiveness and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional video coding techniques are used, then device complexity is reduced, but coding efficiency deteriorates
Solution Approach 1:
The patent replaces conventional mechanical video coding techniques with a neural network-based system. The neural network model performs video coding tasks that traditionally required complex mechanical processing, achieving higher coding efficiency through learned patterns and representations rather than fixed algorithmic operations.
Solution Approach 2:
The patent changes the fundamental parameters of video coding by using neural network weights and activation functions instead of traditional coding parameters. The neural network adapts its internal parameters through training to optimize coding efficiency, whereas conventional methods rely on manually tuned parameters and fixed transformation operations.
2Productivity
If conventional video coding techniques are used, then ease of operation is maintained, but coding efficiency deteriorates
Solution Approach 1:
The neural network-based video coding system performs self-optimization through automatic feature extraction and parameter adaptation. The model automatically learns the optimal coding strategies from training data without requiring manual intervention, thereby improving coding efficiency while maintaining ease of operation through automated decision-making.
3Productivity
If neural network-based loop filter is applied, then coding efficiency is improved, but device complexity increases
Solution Approach 1:
The patent segments the video coding process into distinct neural network modules, including separate loops for feature extraction, filtering, and reconstruction. This segmentation allows each module to be optimized independently and deployed efficiently, managing device complexity through modular architecture while maintaining overall coding efficiency.
Solution Approach 2:
The neural network-based loop filter is designed to perform multiple functions including deblocking, deblocking, and overall video quality enhancement. This multi-functionality reduces the need for separate filtering stages, thereby improving coding efficiency without proportionally increasing device complexity.
4Productivity
If unified neural network-based filter is applied for different slices and color components, then coding efficiency is improved, but device complexity increases
Solution Approach 1:
The unified neural network filter is designed to process multiple slice types and color components (Y, U, V) simultaneously through a single model architecture. This universal filter handles diverse input conditions without requiring separate specialized filters for each case, improving coding efficiency while managing device complexity through parameter sharing and unified processing.
Solution Approach 2:
The unified neural network filter dynamically adapts its processing based on the input characteristics, adjusting its behavior for different slice types and color components. This dynamic adaptation allows the single filter to efficiently handle varying conditions without requiring static specialized filters for each scenario, thereby improving coding efficiency with controlled complexity.
Data Source
AI summary
Embodiments of the present disclosure provide a solution for video processing. A method for video processing is proposed. In the method, a conversion between a current video unit of a video and a bitstream of the video is performed. A neural network (NN)-based loop filter is applied for the conversion. During a parameter selection process for the NN-based loop filter, the number of parameter candidates in a parameter candidate list for the NN-based loop filter is less than a threshold number.