Real-time Video-based Heart Rate Measuring Method

The heart rate, as one of the most important physiological parameters, plays anirreplaceable role in human health monitoring. The content of oxyhemoglobin in human facialcapillaries changes with the contraction and relaxation of the heart. As with oxyhemoglobinabsorption of light, skin color will also vary slightly, which cannot be observed by naked eyes.This paper discusses an algorithm of face location and tracking and a kind of effective videoenhancement method, and proposes a real-time heart rate measurement algorithm based onvideo by using computer pattern recognition and image processing technology. The algorithmhas good operational, equipment cost is low, and the advantages of non-contact measurement.There are three major steps:1. Face localization and tracking. By using the computer pattern recognition technology,the face in the video is located and the feature point is tracking to extract the face region.2. Skin color detection and enhancement. We extract the face region of normalized afterskin color detection, then according to the principle of video enhancement to enhance thecolor of skin.3. Signal denoising. We convert the enhanced color of skin from the space domain totime domain signals, using the human body special heart rate range of signal filtering toremove noise and the heart rate can finally be obtained by extracting the number of signalwave.In order to improve the measurement precision and robustness of the algorithm whileachieving the real-time requirement, the green channel and the method of calculating mean intime windows is utilized.The heart rate measurement algorithm can be performed in real time and meets theapplicable requirement with an average error of only3.46times/minute, has an applicationprospect in daily health monitoring and telemedicine.