Debris detection in SAR imagery using statistics of simulated texture

This paper presents a preliminary approach for debris detection in SAR images based on simulated training are-as. For this purpose radiometrically correct simulations of heaps of debris are produced. Based on statistics of the first and second order they are analyzed for their textural characteristics. The resulting feature information is used for the localization of de-bris-like signatures in real SAR imagery. Results are presented and discussed.