A Real-Time and Non-Contact Approach to Monitor Heart Rate for Night Care Applications

Abstract

Approaches to measure heart rate using a regular color camera has been proposed since their non-contact and low-cost properties. However, in some scenarios, e.g. monitoring the heart rate of the sleeping people at night, are not suitable for the methods using color camera since there is usually no light when people are sleeping. This paper proposed a framework to monitor the heart rate of sleeping people with a camera, an infrared (IR) source, and IR cut filter. The IR sources are attached to the camera, and put on the top of the sleeping people. We computed the intensity variance of ROI instead of mean to extract the pulse signal, followed by Discrete Fourier Transform to find the main frequency as the heart rate. We also addressed the motion interference produced by the sleeping people and alleviated it by an efficient way. We further improved the accuracy by rejecting the unreliable measurements. The experiments show that our method achieved 92% accuracy.