Natural contrast enhancement for dichromats using similarity maps

Color vision deficiency impacts the lives of millions of people around the world. It is essential to understand their condition and provide solutions to improve their quality of life and accessibility to different forms of technology. We use mapping of dichromatically ambiguous areas to obtain an efficient image enhancement technique for dichromats. Previously proposed image enhancement algorithms approach the problem by potentially changing all colors in an image, causing the result to look unnatural to both dichromats and trichromats. In this work, we present a novel technique which preserves the colors and subjective ambience of the original image, as well as localizes enhancements to areas of confusion. We evaluated our algorithm on 207 subjects, out of which 53 were color deficient, showing promising results.