SAR target identification using SAR-COM technique

Abstract

Deep learning techniques give good results on target classification of MSTAR dataset. Most of the techniques for SAR target identification use only the magnitude information of the raw SAR data and discard the phase information. Deep convolution neural network has the ability to automatically learn from the complex image generated using both the magnitude and phase information from the radar data. In this paper we are proposing a new method for generating the complex image dataset (SAR-COM) and a deep learning model which automatically classifies targets from MSTAR raw data.