Adaptive Image Filtering via Integrated Spatial-Photometric Weighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Bilateral filters in existing image processing methods fail to effectively remove noise along edges and color boundaries due to separate handling of spatial distance and pixel value difference weighting coefficients, leading to uncorrected noise and artifacts.
Innovation Solution
An image processing method that determines a single integrated weighting coefficient based on the product of spatial distance and signal intensity differences, using this coefficient to filter image data and ensure accurate noise removal without damaging the image structure.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If separate weighting coefficients are used for spatial distance and pixel value difference in bilateral filters, then the filtering process is simpler and faster, but noise and color artifacts remain uncorrected along edges and color boundaries
Solution Approach 1:
The patent merges the separate spatial weighting coefficient and photometric weighting coefficient into a single integrated weighting coefficient. This coefficient simultaneously considers both spatial distance and signal intensity differences, allowing the filter to effectively remove noise while preserving edges and color boundaries. The integration resolves the contradiction by combining the simplicity of separate coefficients with the accuracy of joint consideration.
2Manufacturing precision
If a single integrated weighting coefficient is used based on product of spatial distance and signal intensity difference, then noise removal accuracy improves, but the computational complexity increases
Solution Approach 1:
The patent introduces a standardized form for the integrated weighting coefficient that incorporates both spatial and photometric parameters. By defining the coefficient as a product of normalized spatial distance and signal intensity difference, the patent achieves accurate noise removal while maintaining computational efficiency through parameter standardization and pre-computation of distance transforms.
3Device complexity
If separate weighting coefficients are handled independently, then the filter structure is simpler, but edges and color boundaries are not properly preserved
Solution Approach 1:
The patent combines spatial and photometric weighting into a single coefficient that is applied uniformly across the filtering operation. This merging ensures that both spatial location and intensity differences are considered together when determining pixel weights, thereby preserving edges and color boundaries effectively while maintaining a relatively simple filter structure.
Data Source
AI summary
An image processing method for filtering image data, which is constituted with a plurality of pixels, at each pixel position by using pixel values indicated by pixels present within a predetermined range containing a target pixel, includes: determining two arguments that are a first argument defined by a spatial distance from the target pixel and a second argument defined by a difference in signal intensity relative to the target pixel, in correspondence to each of the pixels present within the predetermined range; obtaining a weighting coefficient to be used when filtering each of the pixels present within the predetermined range, based upon a single integrated argument represented by a product of the first argument and the second argument; and filtering the image data by using a filter coefficient corresponding to the weighting coefficient having been obtained.