Positive and timely identification of targets is critical in any military scenario. Target identification from backscattered electromagnetic energy is an evolving area. The objective of this paper is to study the applicability of Long Short-Term Memory Recurrent Neural Network (LSTM RNN) for High Resolution Range Profile (HRRP) based Radar target classification. Simulated Radar Range Profile data is used here. Three Different Target models are considered in this study. The classification is performed using a LSTM RNN.

LSTM recurrent neural networks for high resolution range profile based radar target classification

In the implementation of E-pulse technique based radar target recognition system, only one E-pulse is constructed for each target and it is considered to be aspect-independent, which is grounded in the knowledge that E-pulse synthesis is based on the poles of targets and they are aspect-independent. But our recent study suggests that the poles obtained by pole extraction algorithm are aspect-dependent although theoretically poles of targets are aspect-independent. This is because the residues of the poles are aspect-dependent. The residues represent the contributions of the poles to the late-time transient scattered field and they may be so small at some aspect angles that the corresponding poles can not be extracted by pole extraction algorithm. Consequently, different poles may be extracted at different aspect angles. If only one E-pulse is constructed using the poles extracted at one aspect angle, it may not work for the transient scattered fields at other aspect angles. In order to solve this problem, we propose a multi-directional E-pulse technique which constructs multiple E-pulses for different angle regions of a target. Numerical results show that the proposed method is very robust and can dramatically improve the recognition rate of E-pulse technique.

Radar Target Recognition Based on Multi-Directional E-Pulse Technique

High-Resolution Range Profile (HRRP) sequence has attracted academic attentions in the field of radar automatic target recognition (RATR) owing to its abundant spatial-temporal correlation between adjacent samples. However, it is difficult in the working state of radar to obtain complete HRRP sequence samples due to various internal and external factors such as ground clutter and systematic error, which poses an enormous challenge to radar target recognition. Therefore, it is crucial to repair the missing HRRPs based on the adjacent samples in the previous frames. In this paper, we discuss the extrapolate method of incomplete samples and propose an improved neural network algorithm named as Vanishing Gradient Mitigation Recurrent Neural Network (VGM-RNN). The lost samples in the sequence can be extrapolated by VGM-RNN, and the problem of vanishing gradient which is possessed in classical RNN can be effectively mitigated. The proposed method in this paper can be divided into two parts, as sample extrapolation and sequence recognition, in which sample extrapolation is the core method. Experimental results on Moving and Stationary Target Acquisition and Recognition (MSTAR) dataset show that our proposed model exhibits higher accuracy and efficiency, as well as excellent anti-noise performance, compared with traditional methods. It is suggested that our proposed model can be effectively applied to radar system.

/pdf/vgm-rnn-hrrp-sequence-extrapolation-and-recognition-based-on-5ebt7i4sue.pdf

VGM-RNN: HRRP Sequence Extrapolation and Recognition Based on a Novel Optimized RNN

Target recognition and classification are challenging problems in radars. Radars can generate variety of measurements depending on the capabilities of specific radars. Out of the various types of the radars, High Resolution Radar Range Profile (HRRP) measurements are one of the important radar target signatures that can be used for target classification. In this paper, a study has been performed to compare Ran dom forest algorithm and Deep learning networks for HRRP measurements based radar target classification. The motivation of the study is to assess the performance of random forest based classification against deep learning methods due to their inherent explainability nature, which could be preferred choice in mission critical applications. Popular Convolution Neural Network (CNN) and Long Short Term Memory – Recurrent Neural Network (LSTM-RNN) are considered for this comparative study. The study showed that Random Forest algorithm performance is comparable to the Deep Learning techniques in HRRP based target classification problem.

Comparative Study Between Deep Learning Techniques and Random Forest Approach for HRRP Based Radar Target Classification

Support Vector Machines (SVMs) are an emerging machine learning technique that has found widespread application in various areas during the past four years. The success of SVMs is mainly due to a number of attractive features, including a) applicability to the processing of high dimensional data, b) ability to achieve a global optimum, and c) the ability to deal with nonlinear data. One potential application for SVMs is High Range Resolution (HRR) radar signatures, typically used for HRR-based Automatic Target Recognition (ATR). HRR signatures are problematic for many traditional ATR algorithms because of the unique characteristics of HRR signatures. For example, HRR signatures are generally high dimensional, linearly inseparable, and extremely sensitive to aspect changes. In this paper we demonstrate that SVMs are a promising alternative in dealing with the challenges of HRR signatures. The studies presented in this paper represent an initial attempt at applying SVMs to HRR data. The most straightforward application of SVMs to HRR-based ATR is to use SVMs as classifiers. We experimentally compare the performance of SVM-based classifiers with several conventional classifiers, such as k-Nearest-Neighbor (kNN) classifiers and Artificial Neural Network (ANN) Classifiers. Experimental results suggest that SVM classifiers possess a number of advantages. For example, a) applying SVM classifiers to HRR data requires little prior knowledge of the target data, b) SVM classifiers require much less computation than kNN classifiers during testing, and c) the structure of a trained SVM classifier can reveal a number of important properties of the target data.

Using support Vector Machines as HRR signature classifiers

In this paper range profiles offered by the high range resolution (HRR) radar are used as features and a multilayer feedforward neural network (MFNN) based classification scheme is presented for radar target identification (RTID). Typical experimental examples for identifying aircraft-type targets have shown that the range profiles can provide stable feature available for RTID, even in the situation where the change range of observing aspect angle is large. In addition, with a low ratio of learning patterns to test patterns, the trained network has good generalization capability.

High range resolution radar target identification using multilayer feedforward neural network

Since micro-motion of a target will affect the positions and phases of scatters in high resolution range profiler (HRRP), the methods of micro-motion extraction from the signals of range gates cannot be used. According to the variation features of scatter position, we treat the HRRPs as image sequence and an improved Viterbi algorithm is used to track and isolate the range-gate histories of scatters. Then the time-frequency transform and spectral analysis methods are performed to analyze the micro-motion characteristics of each scatter. The proposed algorithm is verified via simulation data and its performance under noisy environment is analyzed. (5 pages)

Chen Da-qing

Papers

High range resolution radar target identification using multilayer feedforward neural network

Micro-motion Analysis Based on HRRP Sequence