A “dynamic” land masking algorithm for synthetic aperture radar images

A novel approach to land masking in synthetic aperture radar (SAR) images is designed and implemented. The developed algorithm takes as input an archived shoreline from a public domain database and modifies it to draw the actual shoreline on SAR images by analysing backscatter values. Starting with data from the GSHHS (the global self-consistent hierarchical high-resolution shoreline database), coastline positioning is improved by evaluating the radiometric intensity gradient in coastal areas of the SAR image. A further enhancement is obtained by applying the Canny edge detection algorithm. The methodology is tested on Envisat and ERS images, as well as on ALOS and COSMO SkyMed images.