Respiratory detection using non-contact sensors

Abstract Current respiratory detection study are mainly based on contact respiratory detection technology which has the limitation that human body should wear corresponding detection instruments. This paper studies three types of non-contact respiratory detection technologies. (1) Kinect-based respiratory detection method, which uses the depth sensor to obtain the depth value of the chest and abdomen region; (2) Ordinary web camera-based respiratory detection method, which applies optical flow to detect the motion information of the chest and abdomen region; (3) Thermal imager based respiratory detection method, which obtains the temperature change information of the nostril region from thermal video. The advantages and disadvantages of these three non-contact respiratory detection technologies are explored according to the experimental results. Kinect-based respiratory detection method is less affected by environmental interference, but the detection accuracy is lower and the detection range is limited. The detection method based on ordinary web camera is more accurate. However, the environmental illumination requirements are high. The detection method based on the thermal imager is not affected by the ambient light source, but the detection has poor real-time performance.