Facial tracking method for noncontact respiration rate monitoring

Abstract

A non-contact method of monitoring respiration rate is developed. The approach involved tracking a facial region of interest (ROI) associated with respiration in thermal images. Image processing techniques were used to enhance the recorded thermal images and to remove unwanted noise. The skin surface area centered on the tip of the nose was specified by a circle that covered the region affected by respiration process. Feature extraction and signal processing methods were applied to this region to compute the respiration rate. It was demonstrated that the method could successfully track the ROI for both regular and random head movement types and could determine respiration rate in a noncontact manner. Further work is in progress to enhance the algorithm so that it can cope with very large head movements.