Detecting Signboard Information of Shops for Revising Car Navigation Database Using Vlms (vehicle-borne Laser Mapping System)

Map database for car navigation systems accommodates a variety of urban information such as gas stations and convenience stores to satisfy growing users need. Though map databases are revised once or twice a year, it may not be very satisfactory not only for car navigation services but also for future passenger navigation services, because many new shops are constantly emerging while many of the others disappearing through severe competition for survival. Nevertheless, actual map manufacturing work still depends on labor-intensive methods. Field survey is carried out based on the claims from customers. There's a real need of improving efficiency to serve "fresh" data. VLMS (vehicle-borne laser mapping system) is developed to acquire three dimension data in urban area reliably, quickly and at high resolution for 3-D urban GIS database. It consists of three laser scanners, six line cameras, INS and GPS. The platform vehicle can be operated at 40km/hr. In addition, the high-resolution imagery and spatial data can be applied to acquire sign-board information of urban facilities, for example, convenience stores and gas stations and so on. Since signboards of certain types of shops, such as convenience shops have "standardized" color, shape and logo, and are mounted on wall at specific height to attract someone's notice, it's possible to automatically identify them with image-processing methods. This paper describes methodologies and the results of applying VLMS imagery for the automated revision of signboard information.