Visual local ization aid for blind

Abstract

This is a computer vision based project that aims to help a blind person find her/his location in indoor environment. The blind person is equipped with a mobile phone that consists of a camera, a speaker as well as the stand-alone localization application. The map of the indoor environments is described by a set of images. The localization is performed by extracting a feature from the image of the unknown location and comparing it the features of the environment. The feature which has the nearest distance to that of the novel image is considered and a voice massage is sent the blind person with the location information. Color histogram is proposed as the image feature. The project is developed using C++ under Symbian OS. The simulation was performed in a supermarket and the results show a success rate of 98% and a cycle duration of 2 seconds