A vision-based localization algorithm for an indoor navigation app

Current indoor navigation solutions face many problems in hospitals. Heavy electro-magnetic interference from hospital equipment affects the compass on a mobile device. Existing infrastructure may also need to be replaced or modified to provide better signal triangulation indoors. Current solutions are challenging in terms of cost (for both hospitals and users) and disruption of normal operation. In this context, this paper introduces the patent-pending SmartIndoor navigation and wayfinding system. With SmartIndoor, the user only has to point his smartphone to any sign inside the hospital. The recognized text is sent to a server which contains a database of sign texts along with their position inside the building. This paper describes the computer vision task of recognizing texts with a smartphone, arguably the most important module of the whole system. Preliminary results show that the proposed approach can be effective for such smartphone-based indoor navigation system.