彩色视频提取心率信号的联合近似对角化算法  

JADE Used to Extract Heart Rate Signals from Color Video

作　　者：詹凌君[1] 彭力[1] 叶小凤[1] 庄丹娴 林顺禄 余林威 

机构地区：[1]华南师范大学,广东广州510006

出　　处：《大学物理实验》2017年第4期7-11,共5页Physical Experiment of College

基　　金：广东省省级科技计划项目(2015A030401086);广东省自然科学基金(2014A030313433)

摘　　要：为了实现从彩色视频中有效提取心率信号,实时检测人体心率,提出使用基于特征矩阵联合近似对角化的盲源分离算法,从RGB彩色可见光视频信号中分离周期信号,为周期信号处理和周期分析得出心率提供数据基础,从而提高非接触式心率测量的准确性。同时利用matlab编写程序进行仿真实验,实验结果表明该算法能有效分离振幅在干扰信号波动范围的0.2及以上的源信号,分离信号与已知源信号相关系数高达0.8。In order to effectively extract the heart rate signal from color video, to detect human heart rate con-stantly ,a blind source separation algorithm based on Joint Approximate Diagonalization of Eigen matrices （JADE） is proposed,which can separate the periodic signal from the RGB color video signal.It provides data for periodic signal processing and cycle analysis to obtain heart rate and improves the accuracy of non-contact heart rate detection.At the same time,use MATLAB to write programs for simulation experiments and the ex-perimental results show that the JADE can effectively separate the source signal with amplitude of 0.2 or more in the range of the interference signal, the correlation coefficient between the separated signal and the known source signal is as high as 0. 8.

关 键 词：联合近似对角化算法 仿真 非接触式心率检测 盲源分离 周期信号 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]