Heart beat detection and analysis from videos

Authors:

Abstract
Camera-based photoplethysmography provides a mean to monitor cardiovascular parameters remotely. This contribution tackles heart beat detection from remote photoplethysmograms (rPPG) by using a Wavelet-based detection algorithm and time delay estimation. Using experimental recordings of 18 healthy volunteers under resting conditions we demonstrate that depending on the used region of interest detection accuracies of 95% to 99% can be obtained. Time delay estimation is shown to improve the temporal exactness of the detections. Future work will be directed at the detection in non-resting periods and the automated identification of usable signal segments.