Non-Contact Video-Based Assessment of the Respiratory Function Using a RGB-D Camera

A fully automatic, non-contact method for the assessment of the respiratory function is proposed using an RGB-D camera-based technology. The proposed algorithm relies on the depth channel of the camera to estimate the movements of the body’s trunk during breathing. It solves in fixed-time complexity, [Formula: see text] , as the acquisition relies on the mean depth value of the target regions only using the color channels to automatically locate them. This simplicity allows the extraction of real-time values of the respiration, as well as the synchronous assessment on multiple body parts. Two different experiments have been performed: a first one conducted on 10 users in a single region and with a fixed breathing frequency, and a second one conducted on 20 users considering a simultaneous acquisition in two regions. The breath rate has then been computed and compared with a reference measurement. The results show a non-statistically significant bias of [Formula: see text] breaths/min and [Formula: see text] limits of agreement of [Formula: see text] / [Formula: see text] breaths/min regarding the breath-by-breath assessment. The overall real-time assessment shows a RMSE of [Formula: see text] breaths/min. We have shown that this method is suitable for applications where respiration needs to be monitored in non-ambulatory and static environments.