Remote radar-camera vital sign monitoring system using a graph-based extraction algorithm

Mm-wave radar and video-based technologies have shown potentials for contactless detection of vital signs. However, state-of-the-art signal processing based vital sign extraction methods are prone to disruptions, such as motion corruption. In this work, we propose a novel graph-based segmentation algorithm for improved accuracy and robustness. We also developed a combined radar-camera system to integrate with the proposed algorithm. The test results on both mmWave radar and camera systems were found to be of high correlations (0.95-0.97) with the golden standard. Our system provides a viable and robust approach for the increasingly popular remote cardiovascular sensing and diagnosing posed by Covid-19.