The Research Of Pulse Signal Processing Algorithm In Strong Background Noise

The blood oxygen signal is interfered by a variety of noises in measurement process inevitably. In particular the motion artifact, which is caused by the movement of the patient's body, leads to erroneous measurement results and can not accurately reflect the actual blood oxygen saturation. Low frequency interference caused by the motion artifact is often overlapped with the spectrum of blood oxygen pulse signal which can't be processed by traditional filters. In view of the pulse signal in strong background noise, the paper proposes a new de-noising method, so that the signal and noises have been separated effectively.The thesis analyzes the power spectrum of pulse signal containing different noise firstly. The spectrum distribution of pulse signal and noise are given. Secondly, this thesis studies the pulse signal processing algorithms through modern signal processing technology. It analyzes the advantages and disadvantages of several de-noising methods of wavelet transform and EMD de-noising method in the pulse signal de-noising. A new de-noising method is proposed based on advantages of these methods. Then the proposed method is compared with wavelet de-noising and the EMD de-noising in qualitative and quantitative. The best parameter is received through analysis of different parameters influence on the new method. At last, the new method is verified using 6 different set of actual measurement data. It is proved that the proposed method can restrain noise effectively.The method is innovative and makes up the lack of hardware filtering. The simulation results demonstrate that it can extract the characteristics of pulse signal in the intensive noises environment. And it also provides an effective approach for the calculation of pulse rate and blood oxygen saturation.