Skip to Main Content
Automatic detection of sustained intensity changes (edges) in SAR images for applications requiring localization and identification of objects is complicated by the nature of the speckle. The speckle is characterized by a high degree of correlation and multiplicative signal dependence. Ratio based detectors like MRoA and RGoA edge detector use predefined thresholds. The modified RGoA detector defines an automatic threshold determining method. But all these edge detectors, apart from detecting the object edges, detect a number of false edges. In this paper, a new ROEWA based algorithm that automatically discriminates the object boundaries and the false edges is proposed. Otsupsilas nonparametric and unsupervised principle for automatic threshold selection is introduced in this classification process. An optimal threshold is selected by maximizing the seperability of the classes in gray level. Real SAR images are used to verify our method and the results are compared with the existing methods for edge detection. Experimental results show that the proposed method is robust and efficient.