This work presents a novel framework for urban map search. The search capabilities of the existing GIS systems are restricted to text-based search, neglecting significant topological and semantic information. We propose a framework that aims to extend the search capabilities by offering sketch-based search. Initially, the urban maps are processed in an offline step in order to extract the topological information in the form of an attributed graph. In the online step, the user queries the system by sketching the desired network structure. The search algorithm is based on attributed graph matching of the query graph and the attributed graphs of the urban maps and allows both partial and global matching of the query. Experimental results illustrate the excellent performance of the system for intuitive search in maps.

A. Mademlis, K. Kostopoulos, K. Moustakas, D. Tzovaras and M.G. Strintzis, “A Novel 2D Urban Map Search Framework Based on Attributed Graph Matching”, IEEE Multimedia, vol.17, no.3, pp. 24-33, July 2010.