Edge-preserving classification of multifrequency multipolarization SAR images

Two edge-preserving segmentation algorithms for multiband images are proposed in this paper. In particular, when dealing with multipolarization SAR data, adaptive neighborhood structures are selected for modelling polarimetric complex amplitudes and region labels, and for achieving detail-preservation. Experimental results show that the novel schemes produce significant visual improvements for detail preservation, and exhibit equivalent or higher classification performance with respect to the classical classification schemes. These results have been obtained from multiband, polarimetric SAR SIR-C data, selected for archaeological application studies.