Noncontact heart rate measurement system on single board computer

Heart rate is important vital signs and has been utilized for medical diagnostic. It is well known that resting heart rate can be regarded as a risk factor of cardiovascular diseases. The standard methods for estimation of heart rate are the electrocardiogram and photoplethysmography, and high-priced dedicated machines are required. However, in order to stay healthy, self-diagnosis of health conditions is very important. In this paper, to promote self-diagnosis, we propose a novel noncontact heart rate estimation system on a single board computer using a RGB camera. Our system is one of the least expensive way to estimate heart rate by using non-dedicated hardware. In our method, most proper measurement area is automatically determined by skin-color detection. Then, the time wave form of the green-channel is regarded as cardiac pulse waves, and heart rate can be estimated by the nonparametric power spectral density estimation and the peak detection algorithm. Experimental results cleared that proposed system has comparable heart rate estimation performance compared with a pulse oximeter.