A comb feature for the analysis of ship classification in high resolution SAR imagery

In the context of maritime surveillance from high resolution SAR imagery, this paper illustrates a study of ship classification, together with a new feature named `comb' suggested to solve the problem. The proposed comb feature presents the added value in application to distinguish between container ship, tank ship, and cargo ship. It is based on the analysis on radar cross section (RCS) statistic of the ship target related to the ship structure. Besides, a related local RCS (LRCS) is proposed to classify three kinds of ships. The data for experiments is made of TerraSAR-X images.