Automatic refocus and feature extraction of single-look complex SAR signatures of vessels

Abstract

In recent years, spaceborne synthetic aperture radar (SAR) technology has been considered as a complement to cooperative vessel surveillance systems thanks to its imaging capabilities. In this paper, a processing chain is presented to explore the potential of using basic stripmap single-look complex (SLC) SAR images of vessels for the automatic extraction of their dimensions and heading. Local autofocus is applied to the vessels' SAR signatures to compensate blurring artefacts in the azimuth direction, improving both their image quality and their estimated dimensions. For the heading, the orientation ambiguities of the vessels' SAR signatures are solved using the direction of their ground-range velocity from the analysis of their Doppler spectra. Preliminary results are provided using five images of vessels from SLC RADARSAT-2 stripmap images. These results have shown good agreement with their respective ground-truth data from Automatic Identification System (AIS) records at the time of the acquisitions.

Keywords:

• SAR
• autofocus
• vessel detection
• feature analysis
• surveillance