Radar target type classification and validation

Abstract

The main challenge in analyzing the results of persistent scatterer techniques is to associate each coherent radar reflection to a real-world object, referred to as target type classification. In recent years different methods to perform target type classification were studied. In this paper we propose a height-based target type classification method to discriminate radar reflections emanating from the ground and above-ground objects. Data acquired from multiple spaceborne satellites such as ERS, Envisat, and TerraSAR-X covering Amsterdam, the Netherlands spanning over two decades from 1992 to 2012 are processed. The target classification results are validated with highly precise elevation data obtained from an independent airborne laser altimetry technique. In this paper we demonstrate that our target type classification method is accurate and thereby the generated DEM of the ground is of nearly sub-metric accuracy in case of ERS and Envisat, and TerraSAR-X.