Clustering based Ship Classification using Radar Signal and Neuron Network

In this paper, we propose a ship classification method based on surveillance radar signals. To get a robust system, our method needs to handle challenges from the high variance given by the poses of a ship. The reflecting radar signals are highly relevant to the poses rather than the ship categories. Therefore, in this paper, we apply a clustering technique to separate the data into suitable domains. Later, the data is classified by corresponding classifiers. Several neural network configurations have been tested to understand the contribution of the proposed preprocessing method. Comprehensive experiments point out that our method helps to improve the relatively good accuracy of a three-category classification.