OSIRIS - A brief survey on ship detection, classification and identification from SAR imagery

Salerno E., Martinelli M., Moroni D., Pieri G., Reggiannini M., Righi M., Salvetti O., Tampucci M.

Synthetic Aperture Radar  Maritime Surveillance  SAR-based vessel identification 

This report summarizes some of the recent literature on ship detection and recognition from SAR images. The main features of the SAR technique and the typical target and clutter signatures are first briefly reviewed, with particular reference to marine applications. Then, the recognition issue is split into three fundamental steps: detection- discrimination, classification, and identification. The information drawn from the literature dealing with detection-discrimination is essentially on the type of features used, with some indication on the extraction methods adopted and the results obtained in the experiments reported. This information is useful for classification and identification purposes in that the different features yield different discriminative powers. Since classification and identification are our main concern in the OSIRIS project, this survey considers the proposed strategies as potential starting points for our own implementation, mainly focusing on the number of classes that can be assigned on the basis of the features considered. An analysis of the classification algorithms proposed is only marginal here, as it will be the subject of a future, much more goal-oriented study.

Source: Project report, OSIRIS, 2016