Ship wake detection using Radon transforms of filtered SAR imagery

Ship traffic surveillance plays an important role in providing safety of shipping, traffic management as well as treating a great deal of related environmental problems. One of the quite new but promising possibilities for this purpose lies in suing satellite-borne SAR imagery. A moving ship produces a set of waves often appearing in the image as bright or dark linear structures. These structures can provide information on both ship direction and speed. In the work presented here, the possibility of automatic detection of ship wakes was tested by applying the Radon transformation to the area surrounding the ship, followed by a verification of each detected wake by a set of criteria to discern it from other wake-like linear structures which are very often appearing in SAR imagery. Different methods for the improvement of the original image are applied as a preprocessing technique for the Radon transformation. The success of the algorithm implementation was found to depend greatly upon both wake and image appearances. The band-pass filtering together with a non-linear image amplification proved to be of use for the detection of practically invisible wakes.