Non-Contact Respiratory Monitoring Using an RGB Camera for Real-World Applications

Abstract

Respiratory monitoring is receiving growing interest in different fields of use, ranging from healthcare to occupational settings. Only recently, non-contact measuring systems have been developed to measure the respiratory rate (fR) over time, even in unconstrained environments. Promising methods rely on the analysis of video-frames features recorded from cameras. In this work, a low-cost and unobtrusive measuring system for respiratory pattern monitoring based on the analysis of RGB images recorded from a consumer-grade camera is proposed. The system allows (i) the automatized tracking of the chest movements caused by breathing, (ii) the extraction of the breathing signal from images with methods based on optical flow (FO) and RGB analysis, (iii) the elimination of breathing-unrelated events from the signal, (iv) the identification of possible apneas and, (v) the calculation of fR value every second. Unlike most of the work in the literature, the performances of the system have been tested in an unstructured environment considering user-camera distance and user posture as influencing factors. A total of 24 healthy volunteers were enrolled for the validation tests. Better performances were obtained when the users were in sitting position. FO method outperforms in all conditions. In the fR range 6 to 60 breaths/min (bpm), the FO allows measuring fR values with bias of −0.03 ± 1.38 bpm and −0.02 ± 1.92 bpm when compared to a reference wearable system with the user at 2 and 0.5 m from the camera, respectively.

Multi-ROI Spectral Approach for the Continuous Remote Cardio-Respiratory Monitoring from Mobile Device Built-In Cameras

“…Independently of the user–camera distance and the lighting conditions, MAE values are always below 1 breaths·min −1 , which is in accordance with the results obtained in [ 50 ], in which breath-by-breath f R values were computed through an analysis in the time domain. Considering all data collected in the two illumination conditions, at 0.5 m, LOAs of ± 3.43 breaths·min −1 are achieved, which are slightly worse than the results obtained in [ 43 ], in which LOAs of ± 1.92 breaths·min −1 were obtained for breath-by-breath f R values computed with an analysis in the time domain, which required the identification of inspiratory peaks and appropriate outlier exclusion techniques. However, our results are better than those obtained with other contactless techniques based on video [ 59 ], such as NIR cameras (MOD ± LOAs = 2.22 ± 8.0 breaths·min −1 ) or FIR cameras (MOD ± LOAs = 0.78 ± 4.24 breaths·min −1 ).…”

“…A prior transformation from images in RGB space to greyscale images is required to apply the OF to video frames. In this paper, the Horn and Schunck (HS) algorithm is used to extract OF from the frames [ 43 ]. The algorithm assumes that pixels maintain their intensity along their trajectory.…”

“…In this way, a velocity vector for each pixel in the image is obtained. According to [ 43 ], the velocity component along the y -axis (i.e., v(y,f) ) was assumed as the most related to movements of the ribcage caused by breathing. The HS optical flow was applied to all video frames, and the values of the y-velocity vectors within the ROI were averaged to obtain a single value per frame.…”

Improved Touchless Respiratory Rate Sensing

“…The general pipeline in most related PIC-based works [7]- [12] includes steps such as the detection of the region of interest (ROI), 1D profile creation, selection of the best motion signals and transformation it to respiratory wave (RW), RW post-processing and RR estimation either in time or frequency domains.…”

“…Real-time performance is specified in such works as [8], [12], while in [11], in realtime only RW is measured, and RR is estimated offline. The continuous RR measurement is performed in such works as [7], [9]- [12].…”