Validating a Notch Filter for detection of targets at sea with ALOS-PALSAR data: Tokyo Bay

Abstract

The surveillance of maritime areas is a major topic for security aimed at fighting issues as illegal trafficking, illegal fishing, piracy, etc. In this context, Synthetic Aperture Radar (SAR) has proven to be particularly beneficial due to its all-weather and night time acquisition capabilities. Moreover, the recent generation of satellites can provide high quality images with high resolution and polarimetric capabilities. This paper is devoted to the validation of a recently developed ship detector, the Geometrical Perturbations Polarimetric Notch Filter (GP-PNF) exploiting L-band polarimetric data. The algorithm is able to isolate the return coming from the sea background and trigger a detection if a target with different polarimetric behavior is present. Moreover, the algorithm is adaptive and is able to account for changes of sea clutter both in polarimetry and intensity. In this work, the GP-PNF is tested and validated for the first time ever with L-band data, exploiting one ALOS-PALSAR quad-pol dataset acquired on the 9th of October 2008 in Tokyo Bay. One of the motivations of the analysis is also the attempt of testing the suitability of GP-PNF to be used with the new generations of L-band satellites (e.g., ALOS-2). The acquisitions are accompanied by a ground truth performed with a video survey. A comparison with two other detectors is presented, one exploiting a single polarimetric channel and the other considering quad-polarimetric data. Moreover, a test exploiting dual-polarimetric modes (HH/VV and HH/HV) is performed. The GP-PNF shows the capability to detect targets presenting pixel intensity smaller than the surrounding sea clutter in some polarimetric channels. Finally, the quad-polarimetric GP-PNF outperformed in some situations the other two detectors.