Radar High Resolution Range Profile Recognition with Small Sample Set Based on Linear Gaussian Mixture Dynamic Model

Abstract

A novel recognition method is proposed to relieve the heavy requirement of training samples in the radar High Resolution Range Profile (HRRP) target recognition. Firstly, the statistical characteristics of HRRP’s frequency spectrum amplitude (FSA) are analyzed. Then a Linear Gaussian Mixture Dynamic Model (LGMDM) is proposed to describe the stationarity and multi-modality of the FSA. Afterwards, the Expectation Maximization (EM) algorithm is derived for model parameter estimation. Finally, experimental results show the proposed method can achieve satisfactory recognition accuracy and rejection performance with only a few training samples.