Perceptual grouping for building recognition in high-resolution SAR images using the GESTALT-system

GESTALT is a production system interpreter designed for advanced automatic recognition from difficult pictorial data. Building recognition from leading edge high resolution SAR-data is a good example for such a challenge. This contribution explains the system itself and its application to this particular issue. Perceptual grouping paradigms are coded in the productions in order to discriminate man-made ordered structure from arbitrary clutter. In particular symmetry and repetitive similar structure render promising prospects for this application.