Real-time contactless respiration monitoring from a radar sensor using image processing method

The respiratory rate and waveform play an important role on reflecting health conditions. Some of the diseases, such as Biot’s Breathing, Chyne-Stokes respiration, require long-term continuous respiration monitoring which is normally conducted by hospitals or sleep centres. Portable sensor devices are required to help patients that are difficult to attend such settings. There are several solutions including piezoelectric sensors, watch type monitoring devices, ultrasound technologies, radar sensors, and so on. In terms of these choices, the radar sensor can provide a more comfortable and contactless experience. Most of the existing methods are analysing the 1-D radar signal. The 2-D radar signal, such as the radar image, contains more spatial information which is easier to be interfaced with further pose or gesture recognition functions to extract useful information. This paper presents an image processing approach with radar images generated from an Ultra-wideband (UWB) radar sensor to produce respiration waveform and estimate respiratory rate in real-time. The raw signals from the radar sensor is converted to a 2-D heatmap image, and then an image processing algorithm is employed to extract respiration information including the respiration waveform from the heatmap. Experimental results show that the proposed system can reliably measure respiration motion in static situations such as lying in the supine or lateral position with an average precision of 95.02%.