An edge-enhanced segmentation method for SAR images

A major problem in processing synthetic aperture radar (SAR) images is the presence of speckle noise which is multiplicative in the sense that the noise level increases with the magnitude of radar backscattering. Past researchers have proposed filtering schemes for removing speckle based on local statistics. However, these usually do not use edge information in their analysis. In this paper, an edge-enhanced local mean and median filter is proposed to smooth SAR speckle noise while preserving edges. Based on the iterative application of this filter, an unsupervised segmentation scheme is provided to divide a SAR image into homogeneous regions with respect to gray level intensity. The simplicity and efficiency of this scheme are demonstrated by application to airborne SAR images.