Interpretation of High Resolution Sar Data Using Existing Gis Data in Urban Areas

Within the paper the automatic matching of existing GIS data to airborne SAR scenes for georeferencing is discussed. For this purpose simulated SAR images are generated based on the available GIS objects using a SAR simulator. These simulated scenes are matched against the captured SAR image in order to detect the given GIS objects in the SAR scene. The feasibility of the approach is demonstrated by the automatic localisation of street data, which have been captured originally for car navigation purposes. Since the horizontal accuracy of this data is limited, the georeferencing is further refined in a second step. This is realised by the automatic detection of landmark buildings, which are selected from a 3D city model. Since good initial values are available from the preceding localisation of the street regions, the matching problem can be solved despite the great complexity of SAR imagery if building objects are depicted. The high accuracy requirements, which can be met if the SAR scene is matched against a 3D city model, are for example required if a combination of the SAR scene with data from other sensors like optical images has to be realised. As it will be discussed in the final part of the paper, the presented matching process can in principle also be applied in order to detect inconsistencies between SAR scene and given GIS data. This will allow for potential applications in the context of change detection and update of the existing data.