Investigating the Complex Signal Kurtosis for SAR Ship Classification

This paper explores the application of the complex signal kurtosis (CSK) for ship classification in single-channel synthetic aperture radar (SAR) images. Preliminary results based on the OpenSARShip dataset reveal varying trends among oil tankers, bulk carriers, container ships and fishing vessels. In addition, high-resolution samples from Capella Space were used to examine the contribution of different ship sections, which showed that the deckhouse has the strongest influence on the CSK in the majority of cases. Nonetheless, it is important to note that limitations in either the spatial resolution or the number of available samples restrict the generalisability of the results. Consequently, this paper emphasises the necessity for a large, open dataset consisting of complex-valued, high-resolution samples to comprehensively investigate the role of the CSK, as well as phase information in general, for ship classification.