Target identification based on diagonal slice of the complex bispectrum

Ultra WideBand(UWB) Impulse Radio (IR) has huge advantages in target identification for its broad band, good concealment and strong ability to penetrate obstacles. We apply the UWB-IR technology theory in target identification under the foliage environment. It is different from the traditional radar and wireless sensor network (WSN), for it combines forward communication and target identification at the same time. Feature extraction is the most important step in target identification. In the study, a novel method based on diagonal slice of the complex bispectrum is proposed. The diagonal slice of the complex bispectrum, which can be computed by selecting arbitrary transform path, is obtained by enlarging the diagonal slice of bispectrum into the z-plane via Chirp-z transform (CZT). The separability features beyond the bispectrum can be obtained. Then the extreme learning machine (ELM) is applied to classify the targets. According to experiment, the results show that it is quite effective for target identification.