Robust heart rate estimate from PPG signal utilizing the harmonic analysis of motion signal

Heart rate (HR) estimation using wrist-type Photoplethysmography (PPG) sensor can play an important role to monitor the physiological status of human body during physical exercise. However, the PPG signal is vulnerable to motion artifact (MA) which makes HR estimation difficult. In fact, the MA generates a number of false peaks in the frequency spectrum of the PPG signal. In this work, we analyze the frequency spectrum of the motion signal acquired from the three-axis accelerometers and figure out a correlation between the peaks of PPG and motion signals´ spectra. Thus we are able to locate the correct HR-peak in the PPG spectrum. The proposed method is computationally efficient and robust to severe motion artifact induced by various physical exercises. The experimental results show that our method works better in terms of average absolute error than the available techniques.