Automatic Classification of Remote Sensing Data for Gis Database Revision

Geographic information systems (GIS) are dependent on accurate and up-to-date data sets. The manual revision of GIS data is very cost and time consuming. On the other hand more and more high resolution satellite systems are under development and will be operational soon thus high resolution remote sensing data will be available. In this paper a fully automated approach for verification of GIS objects using remote sensing data is presented. In a first step a supervised maximum likelihood classification is performed. It is necessary that the training areas for the supervised classification are derived automatically in order to develop a fully automated approach. The already existing GIS data are used to compute pixel masks (which represent the training areas) for each object class. In order to find inconsistencies between the GIS data and the remote sensing data the result of the classification has to be matched with the GIS data. It is shown that different approaches are needed when dealing either with area objects or with line objects. Examples on both approaches are presented. The automatic verification was tested with ATKIS data sets and DPA high resolution remote sensing data. ATKIS is the German topographic cartographic spatial database and DPA (Digital Photogrammetric Assembly) is an optical airborne imaging system for real time data collection. This paper shows the results of the automatic verification of ATKIS objects represented in DPA data.