Signaturanalysen polarimetrischer SAR-Daten von Radarsat-2 und F-SAR

Abstract

In this paper, the potential of two polarimetric decomposition theorems has been examined, as well as their realization in the software packages PolSARpro and ENVI-SARscape. Within this examination, the focus was on the comparison of the efficiency of the module “SARscape”, which is part of the commercial software ENVI 4.8, as well as of the noncommercial software “PolSARpro 4.2.0” (implemented at the University of Rennes, France; funded by the ESA), when applied in my own programming of the used algorithms in IDL 8.1. With physical and mathematical principles of fully polarimetric radar systems as the basis, I conducted examinations in various areas of interest of selected, fully polarimetric SAR- Data taken from the satellite-based sensor Radarsat-2 and the airborne F-SAR of the DLR. In addition to that, it was important to expand the understanding of the dependency of the physical processes on changeable influencing variables or parameters and, moreover and connected with polarimetric analysis, this thesis focused on the interpretation and derivation of fundamental target characteristics. Within the scope of this thesis, I was able to observe that in order to perform a consistent determination of target characteristics, it is indispensable to use a combined examination of polarimetric signature analysis and the Entropy-Alpha classification and to use the filtering of data. In contrast to PolSARpro and the specially developed programs in IDL, it could be demonstrated that the objective of the calculation of polarimetric signatures in SARscape is the detection of backscattering maxima based on trihedral scattering behavior. While applying the reference algorithm in IDL, frictions between the theoretic principles and the implementation in PolSARpro could be identified and resolved during the analysis of the Entropy-Alpha decomposition. Based on the results of this thesis, which have been accomplished by the application of methods used on data sets, it must be concluded that a qualitative analysis and the verification of polarimetric analyses have to become more important in the future.