S.J.S.(1998) A system for multilook polarimetric SAR image statistical classification

Abstract. This paper presents a system for the statistical classification of multilook p o-larimetric SAR images. The methods used are the pointwise Maximum Likelihood (ML), as initial solution, and the contextual ICM ( Iterated Conditional Modes) algorithm. The mu l-tilook SAR data are modelled from the multivariate complex Wishart distribution, and the densities for several important transformations are derived. The system is user-friendly, since it is based upon interactive graphic user interfaces. With this approach, the statistical modelling is hidden to the user. Examples of classifications of SIR-C/X-SAR images is presented