Automatic SAR simulation techniques for object identification in complex urban scenarios

Kurzfassung

Most of the existing SAR simulators provide simulated images only for visual interpretation. This paper presents a new approach for supporting the automatic interpretation of very high resolution SAR images in complex urban scenarios. To this end, an automatic processing chain for generating and geocoding of simulated images has been developed, based on the SAR simulator RaySAR and digital surface models as geometric information for urban scenes. Various simulated images and different layers (layover, shadow, double bounce, and ground) are generated for whole scenes as well as for individual buildings, with consideration of neighborhood influences. For evaluating the proposed method, a LiDAR digital surface model of central Munich is used as input data. The simulated images are compared to a TerraSAR-X spotlight image and show good geocoding correctness, reasonable global layers and precise individual building layover contours.