摘要

The lower-frequency scale spectrum, higher-frequency scale spectrum and scale spectrum with wavelet packets for discrete-time signals were, respectively, defined, with the result that the scale-wavelet energy spectrum is endowed with clear physics meaning. By using the Mallat's multiresolution algorithm and the wavelet packets algorithm, pulse signals of 15 heroin addicts and 15 healthy persons were analyzed. It is found that the ratio between the scale-wavelet energy within a specific scale-time region and the total energy of the pulse signal of heroin addicts is generally higher than that of healthy persons. Using this ratio as characteristic parameter, a critical parameter used for classifying healthy persons and heroin addicts was presented. Thus, 15 healthy persons were correctly detected from a total of 30 subjects. However, two heroin addicts were misjudged in every kind of wavelet transform algorithms. The results show that two methods are really efficient for detecting the abnormalities of the pulse signals of heroin addicts.

  • 出版日期2008

全文