Remote measurement of cardiopulmonary signal using an unmanned aerial vehicle

Abstract

This study proposes a computer vision-based system to reveal the cardiopulmonary signal of humans in both static and dynamic scenarios without the need for restricting or contact. The proposed system extracts the signal based on the optical properties of skin colour variations in the facial region using image sequence analysis captured by a hovering drone while being convenient, safe and also cost-effective. The experimental results showed very good agreement, strong correlation coefficients and acceptable error ratios in comparison to reference instruments (finger pulse oximeter and Piezo respiratory belt). Therefore, the proposed system in this paper may be suitable to be applied to detect human physiological parameters in war zones and natural disasters when the contact with the subject is difficult or impossible.