A new algorithm for iceberg detection with dual-polarimetric SAR data

Icebergs are hazards to lives and goods during navigation in cold waters. In the context of iceberg detection with SAR images, an extensive work was carried out for the detection of large icebergs, but the identification of small bergs or target embedded in sea ice is till difficult. In this work, a new detector is propose to tackle this issue based on dual-polarimetric incoherent (i.e. detected) images. The algorithm is based on the principle that small icebergs are contained in a limited area and they are supposed to have a volume contribution that is higher compared to the sea or sea-ice background. The detector is tested on RADARSAT-2 quad-polarimetric data, where only the multi-looked intensities of the HH and HV channels are used.