A hue-preserving unsharp masking with adjustable saturation using logarithmic ratio operations
Year: 2025
Type: article
Abstract: Unsharp masking is a typical image sharpening technique. When this technique is applied to a color image, each RGB component is processed independently and the color image is reconstructed from the outputs obtained for each component. In this case, the hue of the resultant image is significantly different from that of the input image. In addition, when the sharpening parameter is large, clipping processing is required for the output pixel values that exceed the displayable color gamut, which may cause a loss of detailed patterns. To address these problems in image sharpening, this study proposes a hue-preserving unsharp masking method with adjustable saturation. The proposed method first applies an unsharp masking technique with logarithmic ratio operations to each RGB component of the input image to obtain a sharpened image with a reduced loss of detailed patterns. The sharpened image is then approximated by a linear transformation that satisfies a conditional equation for hue preservation in the RGB color space. If the approximated pixel values are out of gamut, they are modified to be in their corresponding equi-hue planes of the RGB color space while preserving the lightness values of the sharpened image. The effectiveness and validity of the proposed method are demonstrated by comparing it with previous methods using various images. Specifically, the proposed method outperformed the comparison methods in ARNIQA, the latest non-reference image quality assessment, with a value of 0.5539. Furthermore, detailed quantitative evaluation indices such as hue preservation (HD), sharpening effect (GRVE), ratio of black and white skips (CR), and saturation index (CN) were comparable to those of the other methods.
Source: Franklin Open
Language: English
Topic: Image Enhancement Techniques
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $1,900
OpenAlex ID: W4410858540