Study On Radar Target Recognition Using High Resolution Range Profiles

Radar target recognition as an important development direction of radar technology has become an important part of fire control system in future battlefield. High resolution range profiles obtained from the high resolution radar target includes more feature information on the target shape. And using the basic theory of geometry, feature extraction and classification of high resolution radar range profile are studied. kernel technic is introduced to NFL and NFS methods, to make them become two nonlinear methods. And two enhance methods are proposed. And experimental results on radar target recognition with measured data, demonstrated the effectiveness of the proposed methods. The main content is as follows:1. The basic principles of HRRP and the concept of target scattering centers are studied. target characteristics and inherent flaws of HRRP is analyzed. Data preprocessing has been done in two ways. Three aircraft measured data are described in this article.2. The feature extraction methods and classification of feature line are used in radar target recognition. The feature extraction capability was analyzed for between class information and within class information. Contrast feature line classifier with other traditional classifiers, Performance advantage is easy to get. The complexity of the method was analyzed and the computational cost of calculation was assessed. The Uncorrelated Discriminant Nearest Feature Line Analysis(UDNFLA) algorithm is introduced in high resolution range profile. Kernel function and a weighting coefficient is adapted to adjust the proportion between within-class and between-class information to get an optimal effect. And deduced the method of enhanced kernel uncorrelated discriminant nearest feature line analysis(EKUDNFLA) in the high dimensional feature space. Compared with other traditional kernel feature extraction methods, the performance advantages of this algorithm is quantitative analyze. Compared with other features line methods methods, the proposed algorithm was verified to have effective recognition performance. But also the influence between kernel function and the weighting coefficient were analyzed profoundly.3. feature extraction method based on nearest feature space was studied on radar target recognition. For the sake of high computational complexity and unstable result of NFL, the method of NFS would overcome these problems. NFSA and DNFSA methods are introduced into the field of HRRP on radar target recognition. The method of NFS is used to make feature extraction in the high dimensional space using kernel technic. Compared with other traditional kernel NFS feature extraction methods, the performance advantages of this algorithm exhibits superior performance.4. classification method based on nearest feature space was studied on radar target recognition. An extent nearest feature space(ENFS) classification method is proposed, This method was compared with other traditional minimum distance classifier through simulation experiments. Finally, the improved algorithm is applied to other types of data, achieved the desired effect. The algorithm has been shown to have a strong general performance.