CCESAR: Coastline Classification-Extraction From SAR Images Using CNN-U-Net Combination

Monitoring coastline changes is a critical step in evaluating environmental changes, especially with respect to global warming and melting icecaps [1]. Hence, coastline detection and extraction from remote sensing data is an important problem to solve. Synthetic Aperture Radar (SAR) is a remote sensing technology that has promise in this activity because it can penetrate through cloud cover, thus making data available under all weather conditions [1, 2]. At the same time, coastline extraction from SAR images is relatively new compared to optical images and is an important problem to solve [2]. Both coastline and shoreline are defined as a physical boundary between land and water and are interchangeably used [2].