A Heterogeneity-based Ship Detection Algorithm for SAR Imagery

Ship and wake detection from the SAR imagery have been studied for long time. In several situations, ship wake is not available for some reasons, such as sea status, the incidence angle of radar, and so on. Fortunately, ship always imaged in the radar image. In this paper a heterogeneity-based algorithm to detect ship in the open ocean is proposed. Firstly, a heterogeneity test is locally performed to detect the candidate target in the original image, and then a global optimal threshold is selected according to Renyi entropy to get rid of the false object. Thirdly, the segments of the same object for the noise present are merged according to its minimum distance to get the accurate position and number of the ships. Several real SAR images demonstrate its efficiency.