Fuzzy association of vessel targets in heterogeneous remote sensing images

The purpose of this letter is to present a method for data association in spaceborne synthetic aperture radar (SAR) and optical vessel detection reports. Firstly, we analyze the track prediction uncertainty which makes the association problem difficult to be solved. Then in the condition with speed information unknown in detection reports, we use the vessel's sailing angle based on the dead reckoning and multiple features to get a fuzzy correlation matrix. In the matching step, the well-known global nearest neighbor (GNN) criterion is used to get a optimal matching. The experiment results in different condition show that the proposed method is efficient for data association in large-scale maritime surveillance.