Ship Classification in Single-Pol SAR Images Based on Fuzzy Logic

This paper presents a new ship classification methodology that uses single-pol synthetic aperture radar (SAR) images to categorize targets based on a fuzzy logic (FL) decision rule. As such, the method tries to overcome the lack of an operational solution that is able to reliably classify ships with one SAR channel. The method has the following three main stages: 1) radar signature isolation; 2) parametric vector (P) estimation; and 3) decision rule. The first part analyzes the reflectivity histogram of the ship signature to iteratively cluster the pixels of interest. Then, P is calculated by estimating the values of some macroscale features such as length, breadth, and radar cross section profile along the ship signature. Finally, the decision rule is evaluated with FL so that the measured vector P is correlated with the vectors associated with a set of reference categories. These categories have been defined based on user feedback and have been characterized with accurate simulation studies. Specifically, the values of P for each reference ship have been derived with the SAR simulator GRECOSAR. The classification method has been tested with several ENVISAT images acquired for the surroundings of the Strait of Gibraltar. Ground truth has been retrieved via transponder polls, which reveals a preliminary ratio of positive classifications close to 70%. Although this value is not definitive and more tests are needed, it is a good starting point.