AUTOMATIC SHORELINE EXTRACTION FROM SAR IMAGERY USING DEEPLABV3+

Shorelines are significantly dynamic that may undergo rapid retreats and advances driven by changing environmental conditions, such as waves, tide, storm surge, and river discharge, etc. To better manage and protect the coastal regions, the ability to repeatedly observe and quantify the changing position of the shoreline is significantly important. Satellite-borne synthetic aperture radar (SAR) has received much attention due to its advantages in large area, all-weather observation capability. However, most previous studies tried to develop shoreline extraction methods based on SAR images at specific coasts. Widely usable extraction techniques for shorelines with variable characteristics do not yet exist. This study aims to develop a widely usable shoreline extraction method based on SAR images by using Deeplab v3+.