A New Approach for Noncontact Imaging Photoplethysmography Using Chrominance Features and Low-rank in the IoT Operating Room

Hongwei Yue, Xiaorong Li, Hongtao Wang, Huazhou Chen, Xiaojun Wang, Ken Cai
2019 IEEE Access  
Advancements in image-processing and medical imaging technologies and other research areas have resulted in the growing importance of surgical navigation systems in the field of minimally invasive surgery. However, sophisticated auxiliary equipment is a prerequisite for full heart-rate monitoring in operative systems. A system puts forward to address this issue with a wireless sensor network architecture comprising acquisition modules, router gateways, remote servers, and a medical monitoring
more » ... nter. Factors such as variations in ambient light and face shaking can also easily affect heart rate detection based on face videos, thus resulting in inaccurate estimations of heart rate from the blood volume pulse (BVP) signals. This study proposes to address this concern by employing a novel method for non-contact heart rate estimation to overcome noise interference. First, chrominance features are selected to extract BVP signals, and the low-rank and sparse matrix decomposition methods are applied to overcome the detrimental effects of noise and interference while ensuring that valuable data are preserved.Next, the data are relayed to the server via the gateway. Finally, users can log on to the health-related cloud platform and gain information regarding their health status in real time. Experimental results reveal the advantages of the proposed technique over conventional face-based heart-rate estimation methods, including the capability to decrease dependence on sophisticated auxiliary equipment, avoid direct skin contact that can cause discomfort to patients undergoing surgery, and improve the comfort of surgical operations. Moreover, the proposed heart-rate measurement technique can contribute to improve the construction of smart cities. INDEX TERMS Smart city, photoplethysmography, chrominance features, heart-rate detection, low-rank and sparse matrices.
doi:10.1109/access.2019.2932204 fatcat:iddg4ar4lnc6jesd2luln2qga4