Abstract
A driver’s state and wellness is often reflected in several physiological measures, prime among which is the cardiac pulse of the driver. In this research, an innovative non-invasive method was investigated to determine cardiac pulse in real-time for on-road driving. The method is based on the principle that human skin color varies slightly with blood circulation because blood absorbs more light than the surrounding tissue. Even though the variation is invisible to the naked eye, it can be used to extract pulse rate from the facial area in a video of the face. Low frequency part information of the facial area in a moving window is used to generate feature vectors as input to a trained neural network to estimate the pulse rate. The estimated pulse rate, in combination with other measures, can eventually be used to determine a physiological signature of a driver to reflect his/her state and condition.
KEYWORDS - wellness, cardiac pulse, heart rate, face video, neural network