Sar Edge Detector with High Localization Accuracy
Qianru Wei, Yuke Wang, Pengyi Xie · 2019
In order to improve the localization accuracy of edges, an edge detector is proposed for synthetic aperture radar (SAR) images. The proposed detector, named as C-RBED, combines the Canny detector with the ratio-based edge detector (RBED). The proposed detector shows almost unbiased localization accuracy for the edges in SAR images, which benefits from the unbiased localization ability of Canny. On the other hand, C-RBED keeps the property of the constant false alarm rate as the result of RBED component. C-RBED obviously improves the localization accuracy of edges and hence achieves a high true positive detection rate. The experimental results on both simulated and real-world SAR images show that C-RBED can achieve better performance than traditional SAR edge detection methods in the sense of localization accuracy and edge continuity.