Simultaneous breathing monitoring of multiple persons using thermal and visible imaging

Thermal imaging-based techniques have recently emerged as a potential solution for the contactless measurement of breathing rate. However, these techniques were confined to detecting a single person's breathing rate. In this paper, the simultaneous breathing monitoring of multiple persons is proposed by utilizing the cross-spectral visible-thermal image sequences. A cross-spectral multiple tracking-by-detection approach is used to detect multiple ROI in the visible spectrum and map the corresponding ROI in the thermal spectrum to extract the breathing signal. The linear coordinated mapping is performed by the use of a transformation matrix obtained from cross-spectral image registration. The results have been compared with the contact-based respiration belt for validation. The mean absolute error between estimated and belt-based measurements reached approximately 0.1 bpm (breaths per minute), 0.6 bpm, and 2.8 bpm for single, two, and three persons, respectively. The proposed method has a significant agreement and might be useful as a single measurement unit for contactless detection in the medical care centers, quarantine centers, and influenza screening at the airport.