Use of the Dempster-Shafer algorithm for the detection of SAR ship wakes

An automatic detection algorithm (ADA) for synthetic aperture radar (SAR) ship wakes, based upon the Dempster-Shafer algorithm, is described and applied to a large set of SEASAT wake imagery. The SEASAT images were chosen specifically to test the limitations of the ADA. Images exhibiting multiple SAR ship wakes, short or faint wake features, striated ocean backgrounds, or the presence of other naturally occurring linear ocean features, constituted the majority of images. The ADA correctly classified the majority of the images, with misclassifications linked largely to very short or faint linear features. The ADA performed very well in striated ocean regions and in areas where the wake features were competing with naturally occurring linear ocean structure