Abstract
This disclosure describes techniques that enable a viewer with color vision deficiency to better understand color images by automatically mapping problematic colors to distinguishable textures - distinctive black-and-white patterns corresponding to particular colors. A deep learning model selects appropriate textures for each color such that textures are consistent across images. Textures inform the viewer of similar and dissimilar colors and are only applied where needed. The original color image is augmented with textures such that both color-deficient users and users with regular vision can understand the image. The techniques can be used to display images on any device and can also be implemented as a printer plug-in to enable the printing of texture-augmented color images. The techniques can be implemented in web browsers, online image/video hosting services, and by any provider of color images. The techniques can also be used in smart glasses and augmented reality or virtual reality applications. By appropriately selecting the deep-learning network that generates the textures, the smart glasses can be used by multiple users with different types of color deficiencies.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Baluja, Shumeet and Marwood, David, "Augmenting Images with Contextual Texture Cues to Assist Viewers with Color Vision Deficiency", Technical Disclosure Commons, (August 13, 2024)
https://www.tdcommons.org/dpubs_series/7278