The Modified Encoder-decoder Network Based on Depthwise Separable Convolution for Water Segmentation of Real Sar Imagery

Abstract

In order to improve the accuracy and generalization ability for water segmentation in real SAR images, this paper proposes a modified feature extraction network and a modified encoder-decoder network based on depthwise separable convolution. Training these networks with a well-labeled water segmentation dataset and a series of adjusted hyper-parameters, this paper attains a well-trained water segmentation model for real SAR images. Segmentation results of spaceborne SAR images demonstrate the high accuracy and good robustness of the proposed method.