A Map-Search Framework Based on Attributed Graph Matching
Much research has been done in recent years on map digitization and on construction of publicly available digitalmap databases. Digital-map databases are critical components of GIS applications, such as navigation,1,2 route planning,3-5 and travelinformation systems.6 While such digitalmap databases are readily available, one of the major issues that the scientific community must address in this area is the automatic interpretation of satellite and aerial images to extract road network structure and other information directly from terrain-surface images.
While many navigation tools enable searching on these digital maps—with searches typically being based on semantic information, such as street names or points of interest (POIs)—a major problem with the current approach to these systems lies in the fact that the user must know in advance enough semantic information about the area to search and locate POIs or other information. In practice, most people can’t remember all this information. To address this issue, we present a framework for automatically extracting the map graph descriptors for high-level, intuitive search and search by example.