Feature Extraction Accuracy Improvement of Acoustic Signals Based on Reassigned Wavelet Scalogram
Based on the wavelet scalogram obtained by Morlet wavelet transform and hard threshold de-noising filtering for typical acoustic emission signals, region segmented location method is introduced to get the number and accurate values of the characteristic frequencies, therefore the error induced by misjudgment and misreading can be avoided effectively. Then considering the weakness of large characteristic frequency error in Morlet wavelet scalogram, the feature extraction accuracy has been improved by combing region segmented location method and reassigned wavelet scalogram. Simulation results show that the proposed method has the merits of well rapidity, high reliability and briefness, hence can realize high precision feature extraction and has great practical value.
P. Li et al., "Feature Extraction Accuracy Improvement of Acoustic Signals Based on Reassigned Wavelet Scalogram", Applied Mechanics and Materials, Vols. 55-57, pp. 2065-2068, 2011