Dynamic Video Super-Resolution Network Blending Weight
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional super-resolution technologies using neural network algorithms face challenges in maintaining smooth video playback when switching between different video qualities due to frequent switching of resolution weights.
Innovation Solution
A method and circuit system that perform image quality assessment and calculate a blending weight using a dynamic video-based super-resolution network, incorporating a moving average algorithm to select weight tables and adjust weights between different video qualities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If different resolution weights are frequently switched to adapt to different video qualities, then the super-resolution effect is optimized, but the video playback smoothness deteriorates
Solution Approach 1:
The patent implements dynamic weight adjustment by calculating blending weights based on real-time image quality assessment scores. Instead of using fixed resolution-specific weights, the system continuously adapts the blending weight according to the assessed quality of the current frame, enabling the super-resolution algorithm to dynamically respond to varying video conditions while maintaining playback smoothness.
Solution Approach 2:
The patent changes the parameter from discrete resolution-based weight selection to continuous quality-score-based weight calculation. By introducing a blending weight that varies continuously with image quality assessment results, the system transitions from abrupt weight switching to smooth parameter adjustment, resolving the contradiction between optimization and stability.
2Adaptability or versatility
If multiple fixed weight tables are stored for different resolutions, then the adaptability to different video qualities is improved, but the device complexity increases
Solution Approach 1:
The patent extracts the weight selection logic from pre-stored multiple weight tables and replaces it with a single dynamic weight calculation mechanism. Instead of maintaining separate weight tables for different resolutions and switching between them, the system extracts only the necessary weight information by calculating it on-demand based on image quality assessment, thereby reducing storage requirements and system complexity.
Solution Approach 2:
The patent creates a universal weight calculation mechanism that can adapt to any video quality level without requiring separate weight tables for each resolution. The single blending weight formula serves multiple functions by working with images of any quality, eliminating the need for multiple specialized weight tables and simplifying the overall system architecture.
Data Source
AI summary
A method for processing a video with a dynamic video-based super-resolution network and a circuit system are provided. In the method, quality scores used to assess a quality of an input video are calculated based on image features of the input video. A moving average algorithm is performed on the quality scores of multiple frames of the input video for obtaining a moving average score. Two corresponding weight tables are selected according to the moving average score. The two weight tables are used to calculate a blending weight that is applied to a neural network super-resolution algorithm. The blending weight is applied to the neural network super-resolution algorithm, so as to produce an output video.