Video Noise Reduction via Motion-Adaptive Temporal and Spatial Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Digital images suffer from noise during processing, transmission, and compression, leading to quality deterioration and efficiency drops, necessitating effective noise reduction methods that preserve signal characteristics.
Innovation Solution
A method and apparatus for reducing temporal and spatial noise in video images by calculating motion information and using weighted sums of frames, along with adaptive filtering and directional averaging to minimize noise while avoiding motion blur.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If temporal noise reduction is applied to video frames, then noise is reduced, but motion blur may occur in moving regions
Solution Approach 1:
The patent divides the video frame into multiple local windows and further segments each window into motion and non-motion regions based on motion detection. This segmentation allows different noise reduction strategies to be applied to different regions, reducing temporal noise in stationary areas while preserving motion details in moving areas, thus avoiding motion blur.
Solution Approach 2:
The patent applies different noise reduction intensities to different regions of the frame based on local motion characteristics. In non-motion regions, stronger temporal noise reduction is applied, while in motion regions, reduced noise reduction is applied to avoid motion blur. This local quality approach ensures optimal noise reduction without compromising motion regions.
2Manufacturing precision
If noise reduction processing is applied to video images, then image quality is improved, but processing complexity increases
Solution Approach 1:
The patent performs preliminary motion detection and classification of regions before applying noise reduction. By pre-identifying motion and non-motion regions through motion compensation and motion detection coefficients, the system prepares the framework for selective noise reduction, avoiding the need for complex iterative optimization during the actual noise reduction process.
Solution Approach 2:
The patent uses dynamic motion detection coefficients that are calculated adaptively for each local window based on motion activity. The noise reduction strength is dynamically adjusted according to the detected motion characteristics, allowing the system to respond to varying motion conditions without requiring complex manual parameter tuning or post-processing adjustments.
3Object-affected harmful factors
If spatial noise reduction is applied, then spatial noise is reduced, but edge details may be lost
Solution Approach 1:
The patent applies spatial noise reduction selectively based on local edge detection. In regions identified as containing edges or significant spatial variations, the spatial noise reduction is reduced or disabled to preserve edge details. In homogeneous regions without edges, stronger spatial noise reduction is applied, effectively removing spatial noise while protecting important structural information.
Data Source
AI summary
A method and apparatus for reducing noise of a video image are provided. The method includes: reducing noise in a difference between a current frame and a previous frame in which temporal noise is reduced; detecting motion information about the current frame based on the difference in which the noise is reduced; reducing temporal noise in the current frame via a weighted sum of the current frame and the previous frame in which the temporal noise is reduced, according to the motion information; and reducing spatial noise in the current frame based on an edge direction of the current frame in which the temporal noise is reduced.