A Stepped Frequency Modulated Radar Imaging method based on Convolutional Neural Network

In this paper, we propose a new high resolution range profiles (HRRP) imaging method of stepped frequency modulated (SFM) radar based on convolutional neural network (CNN). In the traditional SFM radar imaging methods, the incomplete motion compensation may incur some phenomena, e.g., position shift, amplitude loss and false peaks. In order to solve these problems, and motivated by the great merits of deep learning, we make an interesting attempt to use CNN to learn the mapping from SFM radar echo to HRRP. The basic principle of SFM radar is given firstly. Then the CNN structure and the training process are detailed. Finally, simulation experiments are presented to demonstrate the efficiency of the proposed method.