Non-contact time varying heart rate monitoring in exercise by video camera

In order to monitor the time varying heart rate (HR) of human in exercise effectively, a noncontact detection method based on video camera is realized in this paper. Based on the principle of Photoplethysmography (PPG), the camera records the regular changes of the skin surface in human face due to their blood volume pulse (BVP). After a series of preprocessing including facial recognition, band-pass filter, trend removal, and reconstruction of source signal, the BVP waveform was retrieved from the video signal. In this way, the extraction of HR could be re-formulated as the problem of extracting the frequency of the BVP signal, which is in a traditional digital signal form. In this paper, five classical frequency extraction methods are compared to find the most proper one. The simulation results show that the frequency extracted from the BVP signal could match the time varying heart rate detected by professional equipment and the approach of calculating the mean value of interbeat intervals (IBI) has the best performance in frequency extraction, especially in the stage of postexercise.