文档介绍：784 IEEE TRANSACTIONS ON WORKS, VOL. 11, NO. 3, MAY 2000
Underwater Target Classification Using Wavelet
Packets and works
Mahmood R. Azimi-Sadjadi, Senior Member, IEEE, De Yao, Qiang Huang, and Gerald J. Dobeck
Abstract—In this paper, a new subband-based classification been studied for the characterization of the surface waves using
scheme is developed for classifying underwater mines and Wigner–Ville distribution. The wavelet transform using a five-
mine-like targets from the acoustic backscattered signals. The cycle cosine modulated Gaussian wavelet approximation was
system consists of a feature extractor using wavelet packets in
conjunction with linear predictive coding (LPC), a feature selec- applied to the impulse response of spherical shell of differing
tion scheme, and a backpropagation work classifier. thickness to examine resonance characteristics of the elastic
The data set used for this study consists of the backscattered target [2]. In [3], a wavelet-based method that uses an artifi-
signals from six different objects: two mine-like targets and four cial work to pute discriminatory infor-
nontargets for several aspect angles. Simulation results on ten mation on a target in the form of locations, size, and weights
different noisy realizations and for signal-to-noise ratio (SNR) of
12 dB are presented. The receiver operating characteristic (ROC) of Gaussian patches in the time-scale is described. De Billy
curve of the classifier generated based on these results demon- [4] used the short time Fourier transform (STFT) to determine
strated excellent classification performance of the system. The the resonance spectrum of submerged elastic cylindrical wires.
generalization ability of the work was demonstrated In [5]–[7], different adaptive and spectral-based schemes for
puting the error and classification rate statistics on a large the isolation of target specular reflections in noisy backscat-
data set. A multiaspect fusion scheme was al