A DISCRIMINATION METHOD OF WATER AND SHADOW AREAS BASED ON POLARIZATION ENTROPY OF SENTINEL-1 DATA

Rapid and accurate extraction of water body information is fundamental to disaster assessment. However, in SAR images, shadows often appear similar to the water regions, so the commonly threshold-based water detection methods easily confuse them. The purpose of this paper is to use Sentinel-1 polarimetric SAR data to find a simpler and more effective method to identify shadow areas out of water bodies. Considering a strong dependence on intensity of Wishart classifier, the weak backscattered regions on the SAR images will be categorized into the same class. We firstly use H / (alpha) over bar Wishart to roughly classify the PolSAR data to obtain a relatively complete experimental study area. Then, we further discriminate the shadow regions from the study area based on the difference of the entropy between the water body regions and the surrounding environment. The experimental results based on Sentinel-1 measured data prove the effectiveness of this method.