Indoor localization system for assisting visually impaired people

In this paper we developed an image processing based indoor localization system with color detection technique of connected objects. By using color detection technique our system can pinpoint a user's location with very high accuracy for real-time applications. We used our system to filter out an image for a specific color and extracted pixel co-ordinates for that image. User's location is then determined by comparing the matrix for those values against a pre-created matrix of training images. We successfully conducted experiments in indoor environments as well and they yielded very good results. After analyzing these results, we propose to integrate our localization system with indoor navigation systems which can be used for blind people where accuracy is the most important element. To further facilitate the navigation process we also developed an android based application.

I. Introduction

According to the WHO (world health organization), there are nearly 39 million visually impaired people worldwide [1] and 12.05 live in the south east Asian region [2]. There are approximately 675,000 visually impaired adults in Bangladesh [3]. A visually impaired person traditionally uses a white cane, a trained dog or other people's help to navigate around. So they need a very reliable way to navigate around.