Multi-look polarimetric SAR image filtering using simulated annealing

Based on a previously published algorithm capable of estimating the radar cross-section in synthethic aperture radar (SAR) intensity images, a new filter is presented utilizing multilook polarimetric SAR images. The underlying mean covariance matrix is estimated from the observed sample covariance matrices, and by applying a set of small orientation-dependent filters in an iterative scheme, the input image becomes highly filtered while maintaining most of the structures in the: scene. Results using multi-look polarimetric C-band data from the Danish airborne polarimetric SAR, EMISAR, are presented.