Radar Target Recognition Based on Polarization Invariant

摘要

With the development of radar full polarization measurement technology, target recognition using polarization information has become a research hotspot. Polarization invariants can be used to characterize a target, which can directly indicate the physical property of targets. Previous target recognition research has focused on missiles or aircrafts, but ground vehicles are also a very important category in military targets. In this paper, two tank models were simulated by using FEKO software. The polarization invariants obtained from simulation data are used as an recognition data set. Comparing the results from three different types of recognition algorithms, the average recognition accuracy based on BP neural network is higher than KNN and SVM methods.