Digital Change Detection for Map Database Update

In almost all areas of our society there is an increasing need for well maintained and frequently revised digital map databases. Maintenance and revision of geo-spatial databases can be divided into 3 steps: Change detection, classification of changes and registration and updating of the database. Traditionally, three different methods have been used for change detection: field inspections; manual comparison of the map database and up-to-date orthophotos; superimposition of the map database on a stereomodel. All of them are labour intensive and costly. This paper focuses on automated change detection for a “building” layer in a fully 3D geo-spatial map database. The aim is the development of efficient change detection procedures for database maintenance in a production environment. An automatic change detection method, based on unsupervised classification used as input to an supervised (Mahalanobis) classificaion is presented. The method is evaluated on building registrations from the Danish TOP10DK map database, in combination with RGB (colour) and CIR (Colour Infra Red) aerial photos. The test case presented in the paper is from a residential suburban area.