Towards automated map updating: detecting houses with new digital data-acquisition and processing techniques

Up to now large scale mapping (scales >1:10000) is done nearly exclusively with aerial photographs. New digital camera systems will replace these analogue systems in the near future. This paper describes an approach for automated detection of houses, using a data set of the High Resolution Stereo Camera-Airborne (HRSC-A). The system provides multispectral information with a resolution of 15 cm (3000 m flight altitude) and Digital Surface Models with a resolution of 50 cm, depicting elevation in steps of 10 cm. The elevation and spectral information supplied by the sensor was used for this study. New object-oriented approaches allow the interpretation of these high resolution data sets. The combination of a new approach (multiresolution segmentation, hierarchical networks) and the multispectral high resolution data of HRSC-A with its accurate Digital Surface Model shows very promising results