Region Detection in Sar Images by Adaptive Textural Segmentation

Abstract

A new approach to Synthetic Aperture Radar (SAR) image segmentation based on the use of textural data exploitation is presented. As an efficient way to characterize natural surfaces is to measure their textural features, so a number of textural ("virtual") data are extracted from the original pictures by means of some numerical transformations. The segmentation has been carried out by integrating the hierarchical-clustering and the region-growing approaches into a new technique. Preliminary results on JPL/SAR images are reported and discussed.