Automatic real time derivation of breathing rate from thermal video sequences

The breathing rate (BR) is one of most important physiological parameter used for cardiopulmonary arrest prevention and for evaluating respiratory problems as (sleep) apnea, congestive heart, hypo / hyper –ventilation, asthma etc. In this paper, we propose an efficient method for non-contact estimation of BR using thermal imaging. The system is based on computer vision algorithms and sequentially performs: face detection, interest points extraction and tracking, geometric transformation between successive frames and nostril position estimation. The performance of the proposed framework is evaluated against the BR measured using a wired thermistor. The thermistor is placed near the subject nostril and is connected to an acquisition system designed for medical applications. The experimental evaluation validates the proposed methodology, returning high accuracy scores. In terms of the computational complexity, the system performs the BR estimation in real time.