Contactless recognition of respiration phases using web camera

The respiration rate provides important information about the health. Recently, contactless sensing devices are becoming popular. Methods based on a video-camera (web-camera) recording and subsequent signal processing are a good example. In this paper, we present a method for the extraction of respiration phases from the videosequence. We implemented a single-step Lucas-Kanade method for obtaining velocities and calculated a signal to noise ratio of pixel blocks for a selection of the tracking blocks. We measured ECG simultaneously with web-camera recording and compared ECG derived respiration with respiration derived by our method. Inspiration - expiration phases were also marked by three independent evaluators and used for validation of the method.