ENVISION: Assisted Navigation of Visually Impaired Smartphone Users

Abstract

In this work, we propose ENVISION an assistance system for safe navigation of visually impaired smartphone users. The proposed system generates an intelligent decision to manage the navigation of visually impaired people based on the fusion of GPS technology directions and a new obstacle detection method. It copes with many challenges related to such application processing steps and inherent to constraints of the smartphone platform. These include illumination changes, diversity of background and road textures, low quality of the video streams, and low processing capacity. ENVISION uses a new method to detect static and dynamic obstacles robustly and accurately in real-time video streaming recorded by a smartphone with an average hardware capacity.