Ship target recognition based on superstructure matching

Abstract

The Automatic Target Recognition(ATR) of ship targets based on Inverse Synthetic Aperture Radar(ISAR) images plays a significant role in naval warfare.To address the issue that existing superstructure-based recognition methods do not fully use superstructure shape information,we propose a novel automatic ship recognition method based on superstructure matching using the Dynamic Time Warping (DTW) algorithm.Firstly,the superstructures of each class are extracted from ISAR images after preprocessing and preliminary calibration,then the DTW algorithm is used to create the superstructure templates, and eventually the target's class is determined by the DTW distance from the templates.The proposed method is robust to ship deformation from changing aspect angles and requires few training samples per class.The experimental results of applying the technique on real ISAR data demonstrate the effectiveness of the proposed method.