Design and Implementation of a Real-Time Color Recognition System for the Visually Impaired

There is a growing interest in developing Computer-Based Assistive Technology (CAT) systems able to help the Visually Impaired (VI) in their daily needs and integrate well within society. One aspect that has not been well addressed is helping the visually impaired is the identification of colors for daily activities. Color recognition and perception is very important in interacting with the society and the surrounding environment. This paper presents a proof-of-concept design of a real-time embedded system that can help the visually impaired recognize colors, interact, and take decisions based on their perception of colors. Our approach is based on conveying color information, from the full color space, using a unique set of synthesized sound signals. The hardware part of the system is a pen-like device, which can detect color and generate a language-independent auditory signal representing the HSV values of the identified color. Numerous experiments have been performed using the new system with both the visually impaired and some blindfolded (BLD) participants. The system was proven to be very efficient in relation to training time and leads to high accuracy in color detection, classification, and matching tests. These experiments confirmed that the developed sonification scheme is effective yet simple in achieving color perception for the visually impaired. The proof-of-concept achieves about 93 % recognition accuracy using off-the-shelf components, it is cheap to implement, robust, and requires a much shorter time for training when compared to existing systems.