Facial videoplethysmography provides non-contact measurement of heart activity based on blood volume pulsations detected in facial tissue. Typically, the signal is extracted using a simple webcam followed by elaborated signal processing methods, and provides limited accuracy of time-domain characteristics. In this study, we explore the possibility of providing accurate time-domain pulse and inter-beat interval measurements using a high- quality image sensor camera and various signal processing approaches, and use these measurements to diagnose atrial fibrillation. We capture synchronized signals using a high- quality camera, a simple webcam, an earlobe photoplethysmography sensor, and a body- surface electrocardiogram from a large group of subjects, including subjects diagnosed with cardiac arrhythmias. All signals are processed using both blind source separation and color conversion. We then assess accuracy of IBI detection, heart rate variability estimation, and atrial fibrillation diagnose by comparing to a body-surface electrocardiogram. We present a new heart variability indicator for blood volume pulsating signals. Our results demonstrate that the accuracy of a facial VPG system is greatly improved when using a high-quality camera. Coupling the high-quality camera with color conversion from RGB to Hue provides a level of accuracy equivalent to that of commercially available photoplethysmography sensors, and offers a non-contact alternative to current technology for heart rate variability assessment and atrial fibrillation screening.

Library of Congress Subject Headings

Plethysmography; Atrial fibrillation--Diagnosis; Heart beat; Signal processing

Publication Date


Document Type


Student Type


Degree Name

Electrical Engineering (MS)

Department, Program, or Center

Electrical Engineering (KGCOE)


Gill R. Tsouri

Advisor/Committee Member

Panos P. Markopoulos

Advisor/Committee Member

Jean-Philippe Couderc


Physical copy available from RIT's Wallace Library at RC734.P55 H47 2016


RIT – Main Campus