基于改进色度模型的非接触式心率检测与估计  

Non-Contact Detection and Estimation of Heart Rate by Improved Chrominance Model

作　　者：张家成 邱天爽 马济通 

机构地区：[1]大连理工大学电子信息与电气工程学部,辽宁大连116024

出　　处：《中国生物医学工程学报》2017年第6期641-646,共6页Chinese Journal of Biomedical Engineering

基　　金：国家自然科学基金(61671105;81241059);国家科技支撑计划项目(2012BAJ18B06-04)

摘　　要：成像式光电容积描记技术能够实现低成本、非接触的远程生理参数检测与估计。其中,基于色度模型的视频图像心率估计算法计算简单、效率高,且具有较强的抗运动干扰能力,但并未考虑实验中光照变化产生的信号干扰。针对该问题,提出一种基于改进色度模型的非接触心率估计算法,通过定义感兴趣(ROI)区域和背景区域的反射光模型,利用两种区域颜色通道信息的比值来消除光照干扰,从而得到消除光照干扰的色度模型,并进行脉搏波的提取与心率估计。对10名在校大学生采集到的人脸图像进行分析,发现该算法估计心率的相对误差为3.52%±2.53%,原始色度模型算法的相对误差为7.23%±2.82%,两者对比有显著性差异(P<0.01)。该算法能有效抑制光照变化所引起的干扰,使所提取脉搏波信号的信噪比平均提高1.52 d B,这对更精确地进行非接触式心率估计与脉搏波信号提取具有重要意义。Camera-based image photoplethysmography method enables a low-cost,non-contact way to estimate physiological parameters. The method based on chrominance model has a high computational efficiency and motion resistance. However,this model ignores the interference caused by the illumination fluctuations.Considering this problem,this paper proposed an improved chrominance model. In this model,we defined the reflection model of the ROI area and background area and subtract the illumination interference by using the ratio of these two area. So we can get a new chrominance model without illumination interference and use it to extract the pulse signal and estimate the heart rate. By using the face images from 10 subjects,we found the average relative error of heart rate estimated by this method was only 3. 52% ± 2. 53% and the average relative error of heart rate estimated by previous chrominance method was 7. 23% ± 2. 82%,which has significant difference（P 0. 01）. It also improved the signal-to-noise of the extracted pulse signal by 1. 52 d B compared to the chrominance model. It has an important significance on the non-contact estimation of heart rate and extraction of pulse signal.

关 键 词：心率估计 非接触式测量 色度模型 光照干扰 

分 类 号：R318[医药卫生—生物医学工程]