基于光照矫正的面部局部区域心率检测

Heart rate detection based on illumination correction and local region of face

针对基于视频的非接触式心率信号易受光照干扰的问题,提出一种利用光电容积描记法(PPG)结合光照矫正和面部区域分割的非接触式心率信号检测方法。首先,对在正常光照环境下采集到的包含人体面部区域的视频进行人脸和背景分割,在面部区域以眼睛为基准按比例选取感兴趣区域(ROI);其次,在RGB颜色模型中分别对背景区域和ROI区域提取绿色(G)通道像素均值获得原始信号并利用归一化最小均方自适应(NLMS)滤波算法进行光照矫正;最后,对去除环境光照干扰的心率信号进行时域滤波和频谱分析得到心率。经实验对比,基于NLMS去除光照干扰的视频心率检测在环境光强变化时检测稳定性好,误差率小于5%。

Concerning the problem that video-based contactless heart rate signal is susceptible to light,a video-based contactless heart rate signal detection method using PhotoPlethysmoGraphy(PPG)combined with illumination correction and facial region segmentation was proposed.Firstly,face area and background of the video collected under normal illumination environment containing face was separated and the Region Of Interest(ROI)was selected in proportion according to eyes location and size.Secondly,original signals were gotten by extracting the pixel mean value of Green(G)channel respectively from the background and the ROI region in RGB color model and illumination interference was corrected by using Normalized Least Mean Square(NLMS)adaptive filter algorithm.Finally,the heart rate was obtained by performing time-domain filtering and spectrum analysis of the heart rate signal with ambient light interference removed.The comparative experiment results show that the video heart rate detection based on NLMS to remove light interference has better stability with error rate was less than 5% when the ambient light intensity changes.