An approach to automated spiral eddy detection in SAR images

Satellite derived synthetic aperture radar (SAR) imagery frequently contains manifestations of coherent structures (eddies) of different spatial scale. However, due to strong dependence of SAR imaging on the near-surface wind speed during the SAR acquisition, such eddy manifestations can be at great extent masked by the signatures of other, mostly atmospheric, phenomena. In the present paper, we propose a method for an automated detection of eddy manifestations visualized by surfactant films presenting on the water surface. The method proposed based on sequential application of image transformations aimed at masking atmospheric phenomena and highlighting the surfactant filaments manifesting eddies in SAR images. Thus extracted dark patches are being fitted by circles, and close co-location of several such circles would be considered an eddy manifestation.