摘要
基于独立成分分析(ICA)和完全总体经验模态分解(CEEMD)方法,从人脸视频中提取脉搏波,进行心率测量.用摄像头非接触地拍摄人脸并从中提取R、G、B通道源信号,即皮肤颜色变化信号,利用ICA对其进行分析得到含噪脉搏信号,再用CEEMD将其分解,提取出脉搏波后用频谱分析计算得到心率.通过实验验证该方法与脉搏血氧仪测量结果具有很好的一致性.利用该非接触式测量方法可以准确地测出人体的心率,操作简单,适用于双盲实验,且自适应的脉搏波提取算法省略了传统线性滤波器的参数选取过程,大大减少了测量者的工作量.
A pulse wave extraction and heart rate measurement method from human facial video was studied based on independent component analysis （ICA） and complete ensemble empirical mode decomposition （CEEMD）. R, G, B channel signals reflecting the change of skin color were extracted from facial video, which was taken by non-contact camera, and then were analyzed by ICA to obtain noisy pulse wave. CEEMD was utilized to decompose the noisy pulse wave into different scale components including the pulse wave which was calculated by spectrum analysis to obtain heart rate. Heart rate can be measured by this method, which has similar accuracy with pulse oximeter. Benefiting from characteristics of simplicity and non-contact, this method can be applied to dual-blind trials and improvement of modern medical equipments. Complex procedure of parameter selection is omitted by this adaptive algorithm, which reduces the workload obviously.
出处
《纳米技术与精密工程》
CAS
CSCD
北大核心
2016年第1期76-79,共4页
Nanotechnology and Precision Engineering
基金
国家自然科学基金资助项目(61178040)
关键词
非接触
人脸视频
心率
独立成分分析
完全总体经验模态分解
non-contact
facial video
heart rate
independent component analysis
complete ensemble empirical mode decomposition
