White Cane device : a mobile assistant for visually challenged people

Abstract

The goal of this study is to exploit the portability and availability of mobile phones to provide a means of guidance to the visually impaired while travelling. This system assists a visually impaired person while travelling through the city by implementing image recognition through a mobile phone and offering information of the immediate vicinity. The ‘White Cane Device’ being proposed enables the user to capture a picture, and receives back information about the surroundings. The White Cane Device was implemented in a client-server architecture, where the client acts as a peripheral device, and the server implements the SIFT feature extraction algorithm to perform object recognition. The GPS coordinates of the images are used to optimize the latter process. A thorough evaluation of the proposed system gave exciting results and showed that White Cane Device performs reliable matching even when images contain occlusion or are taken from various perspectives. The tested system returned results within more than acceptable time spans and received positive feedback from the usability evaluations held.