基于人脸视频的心率检测研究综述

Survey on heart rate estimation based on facial video

心率是反应人体生理状态的重要指标。目前,以接触式为主流的心率检测方法需要直接皮肤接触,无法进行远程测量,易导致不卫生、不舒适等问题。近年来,基于远程光学体积描记术(Remote photoplethysmography,rPPG)的心率检测方法受到广泛关注,该方法无须接触即可远程测量,解决了不卫生、不舒适等问题。介绍了人脸视频心率检测的背景;总结了通过人脸视频进行心率检测的基本原理和一般流程;归纳了目前主流的检测方法、常用数据集和评价指标,并从兴趣区域(ROI)的选取、滤波去噪方法和心率信号分离算法等方面进行了实验比较;最后指出了该方向存在的问题及未来研究趋势。

Heart rate is an important index,which reflects the physiological state of human body.At present,the main methods of heart rate detection require directly to touch on skin and cannot measure remotely,which might easily lead to unhygienic and uncomfortable experience.In recent years,the heart rate detection based on remote photoplethysmography(rPPG)has attracted wide attention.This method can be used for contact-free measurement,which overcomes the problems of unhygienic and uncomfortable experience.In this paper,the background of remote heart rate detection is introduced,then the principle and basic flow of heart rate detection based on face video are presented,and then we summarize the mainstream methods,data sets and evaluation metric,and conduct experimental analysis from different aspects,such as ROI selection,filtering de-noising algorithm and heart rate signal separation algorithm.Finally,the existing problems and future research trends of heart rate detection based on face video are discussed.