一种基于广义特征值最大化的胎儿心电盲提取算法  

A fetal electrocardiogram blind extraction algorithm based on generalized eigenvalue maximization

作　　者：苏博妮[1] 谢宏[2] 化希耀[2] 

机构地区：[1]上海海事大学信息工程学院,上海201306 [2]塔里木大学信息工程学院,新疆阿拉尔843300

出　　处：《微型机与应用》2014年第9期92-94,97,共4页Microcomputer & Its Applications

基　　金：国家自然科学基金(61162018);上海科学技术委员会项目支持(12441903500);中国交通部应用基础研究项目(2012319810190);上海海事大学研究生创新基金(2013ycx039)

摘　　要：从母体腹部混合信号中分离胎儿心电信号是一项复杂的信号处理问题,获取清晰的胎儿心电图对围产期胎儿监护至关重要。胎儿心电信号具有周期性的时间结构,根据胎儿心电的自相关性,以及胎儿心电信号与其他源信号之间统计独立,本文最小化周期差分的方差与期望提取信号的比值建立有噪条件下的盲提取目标函数,推导出了基于广义特征值最大化的胎儿心电提取方法。该算法在一个动态的延迟范围内计算自相关矩阵,只需要计算广义特征值和特征向量,不需要迭代计算,算法简单。仿真实验表明,该算法能够提取到清晰的胎儿心电,性能稳定。Extracted fetal electrocardiogram （FECG）from the maternal abdominal signal is a complex signal processing problems, obtained the clear FECG is a great significance for perinatal fetal monitoring. FECG has periodic time structure , according to the autocorrelations of FECG, and statistical independence between FECG and other sources, this paper minimize the ratio between the varianceof the difference periodic and expected to extract the signal , and established theobjective function of blind extraction under the condition of noises.Then derived the FECG extraction method of maximizing generalized eigenvalue. The algorithm calculates the autocorrelation matrix with a dynamic delay, it does not need to be iterative calculation but calculate generalized eigenvalue and eigenvector, algorithm is simple. Simulation experiments show that the algorithm can extract the clear FECG , and the performance is stable.

关 键 词：盲提取 胎儿心电 广义特征值 自相关矩阵 动态延迟 

分 类 号：TN911.7[电子电信—通信与信息系统]