心率的长期监测对心血管疾病的预防和治疗具有重要意义。当前心率检测常用的监护仪、心电图机、智能手表和运动手环等均属于接触式测量装置,长期佩戴易产生压痕,甚至给使用者带来不适。在非接触测量方面,远程光电容积脉搏波描记法 (remote photoplethysmography,rPPG)可以通过分析面部视频获取心率,是一种很有潜力的心率长期监测方法。目前,绝大多数对于rPPG的研究都在使用电脑做数据分析,体积过大不易摆放,难以满足医学临床和家庭日常使用的需求。针对这一问题,本文尝试在嵌入式平台上依据rPPG原理实现心率监测。监测系统主要由树莓派4B开发板、相机和触摸屏组成。采用AdaBoost算法实现人脸识别与追踪,选取额头和脸颊作为感兴趣区域(range of interest, ROI),利用巴特沃斯带通滤波去噪,根据POS模型提取BVP波形,对来自不同ROI的BVP波形做盲源分离得到最终的脉搏波,最后利用能谱分析计算心率。实验结果表明本文所研究的系统具有与PC端相同的心率检测准确性和鲁棒性。本文的研究成果可以为心率长期监测设备的小型化和普及做出自己的贡献,也可以为智慧医疗中的远程监测治疗提供有力的保障。

Long-term monitoring of heart rate is of great significance for the prevention and treatment of cardiovascular diseases.Commonly used equipment for this purpose includes patient monitor,electrocardiograph,smartwatch,sports bracelet,et al.They are all contact devices. Thus,skin indentation and itch are inevitable problems for long-term wearing.The technic named remote photoplethysmography (rPPG) can acquire heart rate from facial videos,which makes it a promising method for long-term heart rate monitoring.However,the data analyses of most rPPG researches rely on computer,which is a stumbling block to the commercialization and expansion of rPPG technic.To solve this problem,this paper attempts to realize heart-rate monitoring based on rPPG technic on the embedded platform.The monitoring system is mainly composed of Raspberry PI 4B development board,camera and touch screen.AdaBoost algorithm was used to realize face recognition and tracking.Forehead and cheek were selected as ROI.Butterworth bandpass filter was used to de-noise.BVP waveform was extracted according to POS model.The final pulse wave was obtained by blind source separation of multiple BVP waveforms from different ROIs.Heart rate was calculated by energy spectrum analysis.Experimental results show that our system has the same heart rate detection accuracy and robustness as PC terminal.The achievements of this paper should be useful for the miniaturization and popularization of long-term heart rate monitoring equipment,and also provide a strong guarantee for remote monitoring treatment in smart medical care.

天津市高等学校大学生创新创业训练计划项目(202110058071)资助项目