A Novel Approach to Image Recoloring for Color Vision Deficiency

Area of Science:

• Computer Vision
• Image Processing
• Human-Computer Interaction

Background:

• Color vision deficiencies, such as protanopia and deuteranopia, limit the perception of color information in digital images.
• Existing image modification methods often fail to preserve image naturalness or adequately enhance color contrast for color-blind individuals.

Purpose of the Study:

• To propose a novel method for modifying color images to improve discrimination for individuals with protanopia and deuteranopia.
• To ensure the proposed method preserves image naturalness and enhances color contrast.

Main Methods:

• Utilizes fuzzy clustering for color segmentation to extract key image colors.
• Maps key colors to the CIE 1931 chromaticity diagram and employs confusion lines to shift indistinguishable colors.
• Adapts key colors by optimizing a regularized objective function balancing naturalness and contrast.

Main Results:

• Quantitative evaluation using nonparametric statistical analysis showed the proposed method significantly outperforms three related methods (Bonferroni adjusted p-values <0.015).
• Qualitative evaluation confirmed the aesthetic quality and improved perceptibility of the recolored images.
• Experiments were conducted on 195 natural images and six digitized art paintings.

Conclusions:

• The proposed method effectively modifies color images for protanopia and deuteranopia, enhancing discriminability.
• The approach successfully balances image naturalness with improved color contrast for color-blind users.
• This work offers a valuable tool for creating more accessible visual content.