Extraction of the superficial facial vasculature, vital signs waveforms and rates using thermal imaging

摘要

This work deals with non-invasive and non-intrusive measurements of the human facial vasculature from thermal imaging, and estimates the waveforms and rates of the arterial pulse. The paper addresses the issues involved with use of Long-Wave IR imaging in studies of human biometrics, and focuses on measurements of human's heart rate based on a small number of thermal video frames. Infrared videos of 30 subjects were recorded at two distances. Vascular mapping and multi-resolution analysis are used to calculate a subject's heart rate using only 512 Long-Wave IR video frames (17.07 seconds). This semi-automatic process is typically 89–99% accurate, with average accuracies at both distances at approximately 93%. The work is a step in a comprehensive computer vision system for computational studies on human-machine interface.