Research On Wavelet Ridge-extraction Algorithms And Its Applications

A proper signal representation is one of the core tasks of signal and information processing, for many practical signals, such as frogs, birds’sound, there are a variety of marine organisms. Biological signal and the production of life such as machine equipment damaged by abnormal signal, the time-frequency information was the important characteristic of the signal, is the basis for many subsequent tasks such as feature extraction, pattern recognition.Wavelet as a nonstationary signal analysis tools, powerful, can contour characteristic signal, can also get details characteristic signal. Continuous wavelet transform signal although there are a lot of redundancy, but it contains all the information of the instantaneous frequency of signal. Show that the stationary phase principle:when the signal to meet the incremental demand, wavelet coefficients show "ridges" feature in the time scale plane, coefficients on the ridge contains all the information signal, can be used to reconstruct the signal. Since the90’s, there have been many kinds of ridge extraction and signal reconstruction algorithm based on ridge.Based on the previous research, proposed one kind based on the "image" segmentation of the ridge line extraction algorithm, because the modulus of wavelet coefficients more often distributed in the time scale plane region (ridges), the wavelet coefficients modulus matrix as an "image", the segmentation then, for each region segmentation is determined from the extreme position can be obtained from wavelet ridge. By using wavelet ridge region can be well soon reconstructed signal. The traditional algorithm this algorithm, accuracy, speed is greatly increased.