Deep CNN Supported Recognition of Ship Using SAR Images in Maritime Environment

Abstract

In this paper, we have proposed a method to recognize the ship using Synthetic Aperture Radar(SAR) images. The ability of SAR to form radar images independent of any weather conditions and with large swath width made this monitoring technique very well suited for maritime surveillance. The dataset used for the recognition of ships using SAR images consists of three classes for classification. We have used a Deep Convolution Neural Network(CNN) for the recognition of ships from SAR images. The network is trained rigorously and after testing the sample data with a well-trained network we achieved an accuracy of 90 percent.