Remote real-time heart rate monitoring with recursive motion artifact removal using PPG signals from a smartphone camera

Remote photoplethysmography (rPPG) recorded by low-cost smartphone cameras is a promising method for noncontact monitoring of heart rate (HR). The main challenges of this method are the limited rPPG signal strength and motion artifacts (MAs). To obtain a clean signal while preserving photoplethysmography features, this study proposes algorithms to eliminate the noise and distortions of the extracted rPPG signal. To obtain the best point for the rPPG signal extraction, the highest green channel difference in two consecutive frames is calculated. Mexican hat wavelet transform decomposition is used for MA distortion elimination. Furthermore, we propose a recursive baseline-wander removal algorithm with an adaptive window to effectively remove baseline drift. The peak detection accuracy is enhanced by using an adaptive sliding-window size based on the detected fast Fourier transform beat frequency. Our proposed algorithm was validated using a range of HR values from seven subjects. The results of the algorithm validation showed a real-time operation and accuracy improvements of more than 37.5% for peak detection in the worst case. The proposed method can measure HR remotely from a 0.4 m distance without any additional sensors, achieving a mean absolute error ± standard deviation of 3.58 ± 2.4.