An edge detection operator for SAR images

The detection of small linear features from SAR imagery is complicated by the presence of speckle and the nonstationarity of the image data. A feature detector is derived from the contrast ratio edge detector to extract linear features from SAR imagery with a constant probability of false alarm. The output of the feature detector is then passed through a thresholded grey level closure operator to thin the detected features, connect adjacent linear features, and remove isolated noise points and linear fragments. The resulting operators are applied to ERS-1 images to illustrate their effectiveness