Real Time Classification of Targets Using Waveforms in Resonance Scattering Region

The classification of similar shaped objects from scattered electromagnetic waves is a difficult problem, as it heavily depends on the aspect angle. The reduction of the adverse effects of the aspect angle is possible by extracting distinguishable features from the scattered signals. In this paper, we propose a target identification method in resonance scattering region using a novel structural feature set based on scattered signal waveform. The feature set carries out a triangularization process to model the hills and valleys of the scattered signal. Once these subwaveforms are identified, their peaks, widths, increase and decrease rates are calculated for each of them. Together with the inter-distance between the sub-waves, feature vector is constructed. Then, cross validation strategies are used to design a classifier using multi-layer perceptron network. The simulations performed by two different target libraries; dielectric rods with different permittivity and small scale aircraft models show very high accuracy of the proposed system in real time