PPG signal motion artifacts correction algorithm based on feature estimation

In order to reduce the interference of human motion artifacts in PPG signals, when a reflective detector is used to extract signals, a novel motion artifacts correction algorithm based on Gaussian function decomposition and minimum mean square error estimation is proposed. Firstly, band pass filter and notch filter are used to reduce signal burr and interference of power frequency signal to get more accurate and smooth PPG signal, at same time, Kurtosis, Skewness, Standard Deviation are used to evaluate whether motion artifacts exist in a long series of PPG signals. Secondly, an improved sliding window method is used to detect the wave peaks and troughs of PPG signal and, then, take them as features of the PPG signal. Thirdly, according to extracted temporal features of quality PPG signals, the characteristics of the disturbed PPG signals by motion artifacts are estimated employing minimum mean square error estimation and corrected high-quality PPG signals are synthesized using the Gauss function. The experiment are carried out with human physiological signal extraction system designed by ourselves to show that the algorithm proposed in this paper can eliminate motion artifacts in PPG signal sequence well and human physiological parameters are calculated more accurate.