摘要
针对传统的独立成分分析(ICA)算法无法实现实时的信号分离和提取的问题,提出一种基于相关性评估和实时Fast ICA的方法。通过对混合信号分段,并进行Fast ICA处理,再对分离后的信号段进行相关性评估,选取与心电信号相关性最大的信号段进行相位的修正与幅值缩放,实现心电(ECG)信号的实时提取。在分离处理前,对混合信号进行基线漂移的过滤,提高了算法的鲁棒性。实验表明:所提算法的准确率可达96.7%,能够有效实现心电信号的实时提取。
Aiming at problem that traditional independent component analysis（[CA） algorithm can＇t realize real- time separation and extraction of signal, an algorithm based on correlation evaluation and realtimc FastlCA is proposed. The algorithm divides mixed signal into signal segments and FastICA processing is carried out, and conducts correlation evaluation on signal segment. The separated signal segment which is most correlated to electrocardiogram（ECG） signal is selected for phase modification and amplitude zoom to realize ECG signal realtime extraction. Baseline drift removing of mixed signal is carried out before separating processing, improve robustness of agorithm. Experimental resuhs demonstrate that accuracy of the proposed algorithm can achieve 96. 7% , can realize real-time extraction of ECG signal effectively.
作者
高立坤
刘东启
陈志坚
GAO Li-kun;LIU Dong-qi;CHEN Zhi-jian(College of Electrical Engineering,Zhejiang University,Hangzhou 310027,China)
出处
《传感器与微系统》
CSCD
2018年第8期112-115,共4页
Transducer and Microsystem Technologies
关键词
盲源分离
独立成分分析
相关性评估
心电信号
blind source separation
independent component analysis （ICA）
correlation evaluation
electro-cardiogram （ECG） signal
