Research On The Targets Recognition Based On The Wavelet Transform And Artificial Neural Network

Research on the feature extraction based on the acoustic property of target is the main method to the target classification. Because there are defaults in the classic methods of acoustic signal analysis, a new method based on the wavelet analysis is brought forward in the paper. By the means of the multi-resolution analysis, acoustic signal is analyzed in different scales. The "peak" of the efficiency of the wavelet transform is developed, and the high dimension feature space is reduced into lower dimension ones, thus the target feature is formed. Aiming at classify the targets efficiently, back propagation classifier in the artificial neural networks are designed, and BP algorithm is used in the training process of the network. At last, the result of the experiment is given.