Radar signal classification by projection onto wavelet packet subspaces

A scheme for radar information compression and signal classification is presented in this paper. The exemplar function describing a target is decomposed onto wavelet packets from which the best basis is extracted. Then the reduced dimension subspace is obtained to represent approximately the original function. The target is characterized by a wavelet packet subspace called target subspace. An unknown signal is classified by the energy of the projection onto each target subspace. Simulation results show that the method is robust