A Modular Computer Vision Sonification Model for the Visually Impaired

This paper presents a Modular Computer Vision Sonification Model which is a general framework for acquisition, exploration and sonification of visual information to support visually impaired people. The model exploits techniques from Computer Vision and aims to convey as much information as possible about the image to the user, including color, edges and what we refer to as Orientation maps and Micro-Textures. We deliberatively focus on low level features to provide a very general image analysis tool. Our sonification approach relies on MIDI using “real-world” instead of synthetic instruments. The goal is to provide direct perceptual access to images or environments actively and in real time. Our system is already in use, at an experimental stage, at a local residential school, helping congenital blind children develop various cognitive abilities such as geometric understanding and spatial sense as well as offering an intuitive approach to colors and textures.