Motion-Adaptive Video Noise Reduction Blending
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video noise reduction techniques are inadequate in low light and noisy conditions, particularly in mobile and automotive applications, due to high computational complexity and ineffective noise suppression.
Innovation Solution
The implementation of motion-adaptive temporal and spatio-temporal noise reduction methods that determine a blending factor for video frames based on a difference metric, allowing for joint processing across multiple color channels, non-linear association, noise variance restriction, and dynamic coefficient adjustment to enhance noise suppression while minimizing ghosting artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If motion-compensated TNR or motion-compensated STNR is used to achieve superior noise suppression, then noise reduction quality is improved, but computational complexity increases too high for efficient implementation
Solution Approach 1:
The patent segments the video processing into separate spatial and temporal filtering stages, with spatial filtering applied first to individual frames, followed by temporal filtering on the spatially filtered results. This segmentation allows each filter to operate independently with reduced complexity, avoiding the need for complex motion compensation while maintaining effective noise suppression through the combination of both filtering approaches.
2Productivity
If low complexity motion-adaptive TNR/STNR techniques are used to reduce computational complexity, then processing efficiency is improved, but noise suppression effectiveness deteriorates in low light conditions
Solution Approach 1:
The patent merges spatial filtering and temporal filtering into a unified spatio-temporal noise reduction pipeline. By combining the strengths of both filtering approaches—spatial filtering for local noise patterns and temporal filtering for temporal consistency—the system achieves superior noise suppression effectiveness in low light conditions while maintaining low computational complexity through the simplified non-motion-compensated architecture.
3Reliability
If conventional TNR techniques are used to suppress noise by blending previous frames, then noise reduction is achieved, but ghosting artifacts increase due to motion between frames
Solution Approach 1:
The patent applies spatial filtering as a preliminary action before temporal filtering. By first processing each frame through spatial filtering to reduce local noise patterns, the subsequent temporal filtering operates on already-cleaned frames, which reduces the blending of motion artifacts between frames. This preliminary spatial processing step prevents ghosting artifacts while maintaining effective noise reduction through the subsequent temporal blending.
Data Source
AI summary
Systems and techniques for noise reduction in video are described. Example implementations provide improved motion-adaptive temporal or spatio-temporal noise reduction that use an improved blending of the current frame with previous frames. The improved blending may be particularly effective for processing video captured in noisy environments such as low-light and/or mobile environments. In some example implementations, the improved blending is based on more accurately distinguishing between pixel difference in adjacent images that are caused by motion rather than noise.