Non supervised method for Low-Backscattering area extraction in high resolution SAR image

Low-Backscattering area extraction is important to oil spill, vessel detection or other object recognition. In this paper, we proposed a novel algorithm for low-backscattering extraction in high-resolution SAR images, which combines histogram thresholding and classification. In each sub window the presence of dark areas was determined based on features derived from non-supervised classification. The proposed method can extract the low-backscattering without applying mask to the land area, and the experiment confirmed it is robust.