Image Geocoding

Student: Nuno Mourão de Amorim
Supervisor: Gustavo Rocha

Abstract

The Global Positioning System is well known for not reaching indoor environments. Several Indoor Positioning System’s have been proposed, but most of these solutions either deliver high accuracy errors or require expensive material to attenuate positioning errors. In this thesis we propose the use of two image based location recognition routines to compute the position and orientation on indoor environments. These routines are based on Structure from Motion, an incremental algorithm which recovers the 3D structure from related photographs. 3D structures generated are geocoded, compressed and stored in a database. New photographs taken from cheap cameras can be geocoded at any time. By combining Structure from Motion with the already existing Synthetic Views and Prioritized Features algorithms for fast location recognition, we are able to compute indoor GPS coordinates and orientation of new photographs within few seconds.

Thesis Download (PDF)