Automatic Real-Time Fever Screening in a Thermal Video Surveillance System

Automatic identification of elevated human body temperature has become a topic of interest in health-care monitoring systems. The use of a thermal imaging device recently gained in popularity for screening Covid-19 infected subjects in public transportation and public places, but can also be transposed to a variety of domains. Scientific studies support that certain thermal imaging systems can be used to reliably measure surface skin temperature in a stable environment. In this paper we propose a fever detection system that expands the capabilities provided with single thermal sensor by integrating two thermal cameras equipped with nonuniformity correction shutters and a blackbody temperature reference source. The system automatically detects a subject's face and forehead as well as the surface of the blackbody, compensates for the error arising as the person is not on the same plane as the blackbody, and triggers an alarm if the body temperature is higher than a predefined threshold. Experiments demonstrate that the fever detection system succeeds to achieve a confidence margin of ±0.43°C.