Radar Target Recognition Based On The Statistic Model Of High Resolution Range Profile

Radar automatic target recognition technology has great application value in military and civilian. High-resolution range profile (HRRP) is the amplitude distribution of returned echoes along the radar line-of-sight, and it contains target structure information. Moreover, HRRP has the advantages of easy acquisition and processing, thereby radar target HRRP recognition is of great value in the RATR community. At present, most of radar target HRRP recognition methods are based on statistical model. In this framework, different establishment method of target characteristics mathematical model represents different classification algorithm.This thesis focuses on the statistical modeling of returned echo in each HRRP range cells. According to the complex statistic distribution of returned echo in each HRRP range cells, the radar HRRP recognition approaches respectively based on parametric probability density estimation method, nonparametric probability density estimation method and semi-parametric probability density estimation method are studied. Firstly, the theoretical analysis of traditional parametric method and nonparametric method is studied, then applied it to the radar target HRRP recognition. Secondly, through the experimental analysis, the advantages and disadvantages of the parametric method and parametric method are directly compared. At last, based on semi-parametric probability density estimation, combining Gamma model with nonparametric methods, such as the Parzen Windows and SLC, two approaches of radar target recognition based on the semi-parametric density estimation are proposed. The two recognition approaches unify the statistic distribution model, and both advantages of parametric method and nonparametric method are merged in the semi-parametric density estimation. Simulation results based on the HRRP dataset of five aircraft models demonstrate the effectiveness of the proposed approach.