RGB-thermal Imaging System Collaborated with Marker Tracking for Remote Breathing Rate Measurement

Abstract

The pixel variation signal extracted from the nasal region of RGB-thermal images can be used to achieve breathing rate (BR) measurement. However, this method fails when the nasal region is not detected in complicated motion scenarios. In this paper, we develop an RGB-thermal imaging system collaborated with marker sticker to achieve unobtrusive and accurate BR measurement. Pixel variation signal of Regions of interest (ROI) is extracted from the thermal video and chest movement signal is extracted from the RGB video with the assistance of marker stickers. Subsequently, a custom-made time-domain signal processing approach is developed for determining BR. We further propose a method of splicing computation to measure the BR after separate processing of signal segments. We construct an RGB-thermal video dataset with different head and body movements to evaluate the effectiveness of the proposed algorithm. After linear regression analysis, the determination coefficient (R2) of 0.905 has been observed for the estimated and reference BRs, indicating the feasibility of our proposed method in complex motion scenarios.