Respiratory rate detection using a camera as contactless sensor

Abstract

Aim of this study is to extract near-continuously respiratory rate by a contactless method. An industrial camera was used to record subjects face. Video data were processed offline to derive the video-photoplethysmographic (videoPPG) signal. Three features were extracted from videoPPG and finger PPG signal: pulse rate variability (PRV), pulse amplitude variability (PAV) and pulse width variability (PWV). A combination of these methods has been exploited to estimate the respiratory rate for each time window of 5 second. The results showed relative error with median around 0.5% and interquartile range of 5% both for finger PPG and videoPPG system.