Image Based Indoor Positioning App Development

Abstract

Positioning technologies are widely used in mobile devices. Smartphones are able to determines the accurate position using GPS, which is outdoor positioning system. For indoor environment, where GPS satellites are not visible, light, radio waves and magnetic fields are often used, such as Wi-Fi-based positioning system. The accuracy of positioning using Wi-Fi or Bluetooth is not good (2-100 meters) for indoor environment. This project intends to develop an image based indoor positioning mobile application which is able to determine the position from the real-time images taken by phone's camera. A SIFT (Scale Invariant Feature Transform) based with context enhanced algorithm was designed and implemented to position by matching a serial of real-time images with panoramic images taken by a special camera. Experiments showed with dozens of locations sampled in the database, the application could locate users in less than 10 seconds in 77% of cases. This project proved image based system could achieve real time indoor positioning in current smartphones and it could be further developed for larger and more complex environments. Also it can be associated with Wi-Fi positioning system to achieve better accuracy and speed.