Signals Of Fetal Heart Rate's De-noising Based On Wavelet Analysis

The original fetal heart rate signals obtained by the ultrasonic Doppler monitor, it contains mother's disturbances, noise produced by transport process. In order to token fetal heart rate better, put forward a de-noising method of wavelet analysis.We discussed the standard period signal and white noise signal under three kinds of wavelet bases, and discussed the synthesized signal of the standard period signal add with white noise signal;Establishment a standard of wavelet analysis by analysis standard deviation of every sub-signal, handles to provide to according to the de-noising of the fetal heart rate signal.For the acquiring process, we do an experiment on a computer, and emulate it by the MATLAB software, in this way we gains emulation signals and autocorrelation curve as well as emulation heart rate. It draws an AR signal model for power spectrum estimates of fetal heart rate signals to indicate statistic characteristics in the aspect of frequency of the signals. We also find auto-similarity (what in other word is fractal characteristic) in the signal by the means of continuous wavelet transform.Heart rate signal is a kind of changeably signal, and is best for wavelet transform. We select the db4 wavelet base, wavelet transform resolves the signal in each frequency, noisy signal is high frequency, which locates in high frequency bands, and heart rate signal locates in low frequency band. So we handle each frequency respectively, we deal with the high frequency coefficient according to the character of the noise, adopt the method of compulsory de-noising, acquired threshold de-noising and given soft-thresholding de-noising respectively, use given soft-thresholding de-noising signal to token fetal heart rate.