Application study on underwater target recognition based on wavelet packet entropy

A method for underwater target recognition based on wavelet packet transform and fisher linear classifier is studied. On the basis of wavelet transform, the wavelet packet transform is developed. It can offer plentiful time-frequency information for nonstationary signals. Firstly, the radiated noise of underwater target is decomposed by wavelet packet. Secondly, the entropy of terminal nodes through wavelet packet is served as feature vectors. Lastly, the piecewise linear classifier which is designed based on Fisher linear classifier is applied for underwater target recognition. Simulation results show that the classification method which uses wavelet packet entropy as feature vectors possesses higher recognition correct ratio.