An investigation on server-side object-scene recognition performance using coarse location information and camera phone-captured images

Abstract

This paper presents a solution based on information already residing within a mobile network and aimed at the cultural tourist. It also demonstrates how scene (or landmark) recognition from an image can be achieved by combining local invariant image features, cell location information and classification based on Self-Organizing Map clustering. The proposed server-side approach makes the solution independent of the mobile platform and thus accessible to any camera- embedded mobile station having the Multimedia Messaging Service enabled.