Video based heart rate estimation under different light illumination intensities

An issue of obtaining accurate heart rates from video frames is the illumination and the minimum duration needed to provide that. This paper examines the accuracy of the obtained heart rates of the same subjects from video frames under different illumination settings. In this study, the facial images of three subjects at rest are captured using a video camera under five different settings. For each setting, 25 readings of video duration of 30 seconds for each subject are captured forming a total of 375 readings to be analyzed. We propose a method that takes into consideration the power of transmitted and incident light on the skin absorbance. The Pearson correction coefficient of the average pixel intensity of red component and green component is used to determine the duration. Experimental results show the proposed method can provide an acceptable result. The root mean square error is about 1.5508 beats per minute (BPM). The average percentage error is 1.1665 % while the correlation coefficient between the experimental readings and actual readings 0.9789.