A New Combined Method of Segmentation and Classification of Synthetic Aperture Radar Images

In this paper we proposed a hybrid method for segmentation and classification of SAR images. As image segmentation is a primary step of any segmentbased classification, a two level segmentation approach has been proposed. This approach adds value to the polarimetric data analysis by including information on the backscattering behaviour of the objects, extracted by the Freeman-Durden analysis method in the first segmentation level. In this paper neural network is used as the classification engine. We use NASA/JPL data of San Francisco area for our experiment. We proposed a new scheme of supervised classification. The result shows the effectiveness of the algorithm according to the caparison with the present maps of the area.

کلیدواژه ها:

Segmentation ، Classification ، SAR imaging ، Freeman-Durden decomposition ، neural networks