Radar Target Recognition Using the Differential Power Spectrum

In this paper we discuss the problem about the target recognition by the high resolution radar range profiles. Several feature extraction methods for computing shift invariants are simply reviewed: such as bispectrum, differential cepstrum, then the differential power spectrum (DPS) based features are introduced to this study. A multi-layered feed-forward neural network with simulated annealing resilient propagation (SARPROP) algorithm is selected as classifier. Simulations are presented to identify the range profiles of four different aircrafts. The results demonstrated that the differential power spectrum based features are effective and robust for radar target recognition