Wavelet Transform Based Transient Detection

Abstract

In this work, the problem of detection of transients in noise is considered. A method for the detection and classification of transients based on the matched discrete orthonormal wavelet transform is proposed. The idea behind the method is to increase the transient/bacground ratio by expanding the signals in the time-scale space, identifying the bands containing transients and using various linear/nonlinear operations, such as the Teager’s energy operator, for enhancement. Matched wavelet filters are designed from typical transient waveforms by using the Frazier-Kumar method in order to increase the correlation between the wavelet family and the transients to be detected. Transients are also analysed via the short-time Fourier transform, the Gabor representation and the Wigner distribution.