Blind Color Defringing via Edge Alignment and Luminance Scaling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques fail to effectively remove color fringes caused by chromatic aberration, particularly at high-contrast edges, as they often soften or eliminate desirable color features, and are not efficient in handling axial chromatic aberration.
Innovation Solution
A 'blind' image defringing process that aligns color edges without knowing the cause of the fringe, using green channel interpolation and luminance-scaling to preserve edge crispness, and separately addresses axial chromatic aberration, avoiding extreme measures unless necessary.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If chroma blur and chroma median filtering are applied to the whole image, then color fringes are reduced, but edge sharpness and color accuracy deteriorate
Solution Approach 1:
The patent applies different processing strengths to different regions of the image based on edge detection. High-contrast edges receive stronger defringing treatment while low-contrast regions maintain their original sharpness. This local adaptation allows the system to reduce color fringes where present without unnecessarily softening edges that are already sharp or are not affected by chromatic aberration.
Solution Approach 2:
The patent segments the image processing into multiple stages: edge detection, fringe identification, and selective defringing. By first identifying which pixels and regions are affected by color fringing, the system can apply processing only where needed, preserving edge sharpness in regions where fringing is not present.
2Object-affected harmful factors
If chroma blur and chroma median filtering are applied to the whole image, then color fringes are reduced, but color accuracy and saturation deteriorate
Solution Approach 1:
The patent applies different processing strengths to different regions of the image based on edge detection. High-contrast edges receive stronger defringing treatment while low-contradiction regions maintain their original sharpness. This local adaptation allows the system to reduce color fringes where present without unnecessarily softening edges that are already sharp or are not affected by chromatic aberration.
Solution Approach 2:
The patent segments the image processing into multiple stages: edge detection, fringe identification, and selective defringing. By first identifying which pixels and regions are affected by color fringing, the system can apply processing only where needed, preserving edge sharpness in regions where fringing is not present.
3Object-affected harmful factors
If axial chromatic aberration is corrected using extreme measures, then color fringes are removed, but image quality and natural appearance deteriorate
Solution Approach 1:
The patent implements a dynamic defringing process that adapts to the specific characteristics of each image. Rather than applying a fixed extreme correction, the system continuously adjusts processing parameters based on detected edge contrast and fringe characteristics, allowing aggressive correction where needed while maintaining natural appearance in typical cases.
Solution Approach 2:
The patent changes processing parameters dynamically based on image content. The degree of defringing applied varies depending on the detected edge contrast, color channel separability, and local image characteristics, enabling optimal correction without introducing artifacts or degrading overall image quality.
Data Source
AI summary
This disclosure pertains to novel devices, methods, and computer readable media for performing “blind” color defringing on images. In one embodiment, the blind defringing process begins with blind color edge alignment. This process largely cancels every kind of fringe, except for axial chromatic aberration. Next, the process looks at the edges and computes natural high and low colors to either side of the edge, attempting to get new pixel colors that aren't contaminated by the fringe color. Next, the process resolves the pixel's estimated new color by interpolating between the low and high colors, based on the green variation across the edge and the amount of green in the pixel that is being repaired. Care is taken to prevent artifacts in areas that generally do not fringe, like red-black boundaries and skin tone. Finally, the process computes the final repaired color by using luminance-scaling of the new pixel color estimate.