Incorporating certain human vision properties in vision substitution by stereo acoustic transform

This paper focuses on the implementation of certain human vision properties in the real time image processing for vision substitution by acoustic transform of visual information. An experimental prototype has been developed with vision sensor fixed in a headgear and stereo earphone and a laptop computer interconnected. Captured image from vision sensor is processed and transformed into a stereo sound through the earphone. The image processing is critically designed as a model of human vision. It involves edge detection, background suppression, normalization and stepped weightage. The scheme is tested with a blind volunteer and his suggestions in formatting the sound pattern are considered. The software module suits most of the commercially available vision sensors