Target Recognition In Ballistic Missile Midcourse Defense Based On Features Of HRRPs

Ballistic missiles take a long flight time in the midcourse.In order to improve the viability of warhead,a large number of baits and debris will be released,jamming the defense systems.Identifying warheads from a large amount of threat target group is one of the most critical problems for the missile defense system.It is of great significance to put a research on ballistic missile target recognition technology to enhance the capability of anti-missile defense system.In this paper,targets in midcourse(warhead,fragment,spherical decoy)recognition is studied by the methods of feature extraction of high resolution range profiles(HRRPs).The range profiles(RPs)of targets are equivalent to the projection to the line of sight(LOS)of the scattering centers,so it can reflect the subtle structural features of targets.Because HRRPs have attitude sensitivity,the orbital motion and the micro motion of targets cause the periodic transformations of the attitude angle,so HRRPs change correspondingly.Being aimed at extracting robust features from HRRPs,a complete algorithm is developed to recognize warhead from baits.Firstly,each HRRP is divided into two classes by the minimum within-cluster variance algorithm,so that the coupling degree of energy distribution between the two classes is the lowest,and the two peaks of HRRP of warhead are in different classes.Then,the weight interval(WI)and the weight variance(WV)are brought in as the quantized features to build the feature space.The distribution of HRRPs in the feature space is calculated for targets at full attitude angles.Finally,the ellipse fitting methods are used to determine the feature distribution region of warhead,and the robustness of the algorithm is verified by computing the recognition rate and the false alarm rate.Experimental results show that the algorithm is efficient to recognize warheads from debris,spherical decoys,and light decoys of high imitation.Certainly,there must be a large number of template data as the training data to create the feature space.For national defense security reason,the size,material,and HRRPs of targets in midcourse are not available.Based on the existing materials,the CST MWS software was used to simulate the target models and obtain their full-attitude angle echo database.Then we carry out the research on feature extraction of HRRPs and ballistic missile target recognition according to above methods.