This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. IEEE JOURNAL OF OCEANIC ENGINEERING 1 On the Capability of Hybrid-Polarity Features to Observe Metallic Targets at Sea Rafael Lemos Paes, Student Member, IEEE, Ferdinando Nunziata, Senior Member, IEEE, and Maurizio Migliaccio, Senior Member, IEEE Abstract—In this paper, target at sea detection is addressed using hybrid-polarity (HP) synthetic aperture radar (SAR) architecture. Wave polarimetry concepts are used to define HP features that are used to observe targets at sea. The sensitivity of HP features to both targets and the surrounding environment is analyzed through a novel objective norm, namely the Relative Sensitivity for Polari- metric Features (RSPolF) index. Detection performance is eval- uated by the novel Dependency of Sea state and Target (DoST) surface characteristics metric, and by the well-known Figure of Merit (FoM). Experiments undertaken on HP measurements em- ulated from Radarsat-2 and ALOS-PALSAR full-polarimetric ac- tual SAR data demonstrate the effectiveness of the proposed HP approach and the different sensitivity of HP features to targets and background sea characteristics. Following those results, a ranking of the HP features performance is proposed which mainly high- lights that HP features complement each other in the detection process. Finally, a constant false alarm rate (CFAR) approach is proposed to exploit two HP features for target detection in an un- supervised way. Index Terms—Hybrid polarization, performance metrics, polarimetric synthetic aperture radar, remote sensing, wave polarimetry, sea surface, target detection. I. INTRODUCTION M AN-MADE targets at sea include a variety of different targets not only in terms of their size. In fact, the most immediate case is the one referring to ships and therefore all matters related to surveillance for navigation security, for anti- piracy activities, oil fields monitoring, and maritime traffic con- trol. In this case, a key issue is not only the spatial surveillance coverage but also the temporal revisit time. In other cases, one may be interested in monitoring very different man-made tar- gets such as oil rigs. Although oil rigs can be of very different sizes, their location is usually known and temporal coverage is a less demanding need. However, in all the cases, a noncoop- erative monitoring system is often of interest such as the one provided by remote sensing. Manuscript received December 22, 2014; revised April 10, 2015; accepted April 15, 2015. Associate Editor: R. Romeiser. R. L. Paes is with the Geointelligence Division, Institute of Advanced Studies (IEAv), São José dos Campos, 12228-001, SP, Brazil and also with the Earth Ob- servation Department, National Institute for Space Research (INPE), São José dos Campos, 12227-010, SP, Brazil (e-mail: rafael.paes.it@ieee.org). F. Nunziata and M. Migliaccio are with the Dipartimento di Ingegneria, Uni- versità degli Studi di Napoli Parthenope, 80133 Napoli, Italy (e-mail: ferdi- nando.nunziata@uniparthenope.it; maurizio.migliaccio@uniparthenope.it. Digital Object Identifier 10.1109/JOE.2015.2424751 Satellite synthetic aperture radar (SAR) is a synoptic active microwave remote sensing system whose observations are independent of solar illumination and mostly unaffected by weather conditions. Basically, targets call for backscattered signals larger than the background sea, due to their metallic structures large with respect to the electromagnetic wave- length, which result in brighter spots in the SAR image plane. However, external and internal factors can increase the back- ground clutter, as well as environmental conditions (sea surface winds, currents, rain cells, etc.) and radar features (spatial resolution, frequency, polarization, incident angle, radar look direction, etc.), respectively. Then, the information provided by a single-polarization SAR is not always sufficient to effec- tively observe them. For this purpose, polarimetric approaches have been proposed that allow taking full benefit of the vec- torial nature of the scattered field. Both physically-based and image-based polarimetric approaches have been developed [1]. In [2], a polarimetric ship detection algorithm was proposed, which is based on the Neyman–Pearson criterion. In [3], a method to solve azimuth ambiguities for ship detection using full polarimetric and high-resolution X-band SAR data was proposed. In [4], a target-to-clutter ratio was analyzed against different polarimetric channels presenting the key role played by an incident angle. In [5], the coherent target decomposition was specialized for target detection purposes. In [6], a detection algorithm based on the Cloude–Pottier polarimetric entropy was proposed. In [7], a multipolarization study was undertaken to define an effective method to detect targets at sea. In [8], a linear polarimetric detector that maximizes the target-to-sea contrast, i.e., the polarimetric matched filter (PMA), was proposed. In [9], a Notch filter was developed which allows detecting features polarimetrically different from the sea back- ground. In [10]–[13], a polarimetric detector that exploits the different symmetry properties resulting from targets and sea clutter was conceived and applied to L-, C-, and X-band SAR data. Nonetheless, full-polarimetric (FP) SAR has a high cost in terms of average transmitted power, on halved swath width and limited range of acceptable incidence angles, with respect to a single-/dual-polarization SAR. All this makes the FP SAR not optimized in terms of spatial and temporal coverage. Such draw- backs are overcome by compact polarimetry (CP) architectures. They provide polarimetric performance close to the one related to FP SARs while avoiding their principal drawbacks. The main idea of CP architecture is to transmit only one po- larization while receiving coherently, according to an orthog- onal polarization basis. Three CP architectures have been pro- 0364-9059 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.