Detection of the Pulse Waveform Characteristic Points by Wavelet Transform Using Multiscale Differential Operator

Abstract

The pulse waveform characteristic point detection is important for non-invasively detecting cardiovascular parameters. A novel designed algorithm based on wavelet transform (WT) is developed. For some pulse waveforms, characteristic points exactly correspond to the zero-crossing of a wavelet with one vanishing moment, while the others incompletely correspond to. And characteristic points also correspond to the local extrema of a wavelet with two vanishing moment. So an algorithm combining a wavelet with one vanishing moment and another one with two vanishing moment is used to improve the detection rate of characteristic points. The results show that automatic identification of characteristics points has a high rate of accuracy.