SAR polarimetric parameters for land use classification

Abstract

Polarimetric SAR offer full complex S matrix measurements at a high resolution across a two dimensional land surface area. Many algorithms have been developed which aim to use the information within the scattering matrix or the derived covariance matrix to obtain terrain classification. Irrespective of whether the classification is carried out in the measurement domain, using some form of clustering, or in the spatial domain, using image segmentation, the efficiency of any algorithm is greatly increased by using the smallest number of uncorrelated paarmeters. The puropose of this paper is to consider the utility of some parameters for the purpose of terrain classification. Polarimetric SAR data recorded with DLR's E-SAR system will be used in the analysis.