Köhler, Etienne Supervisor: Prof. Gudrun Klinker Advisor: Eghtebas, Nassim (@ga53xoy) Submission Date: [created]
It is becoming more frequent that psychological experiments are being conducted online, meaning that participants can take part from the comfort of their homes. With the recent development of visual heart rate detection algorithms, heart rate and other cardiovascular signals can now be detected using a regular webcam, which allows experimenters to measure valuable physiological information online, without having to connect their participants to expensive heart rate monitors. An issue that presents itself, is that the performance of visual heart rate detection algorithms is highly dependent on the lighting conditions that illuminate the subject. Since experimenters cannot guarantee that every participant is illuminated by the same lighting conditions, this could cause measurements of varying accuracy. To help in understanding how lighting conditions affect the performance of visual heart rate detection algorithms, this study introduces a new public dataset that exposes each subject to 13 different lighting conditions. We discovered that the placement and direction of the light source play a vital role in the quality of a measured pulse signal, and found luminance and colour temperature levels that allow for the most accurate measurements.
[ PDF (optional) ]
[ Slides Kickoff/Final (optional)]