小波域变步长ICA算法提取胎儿心电信号

摘要

Fetal ECG ( Electrocardiogram ) is a primary objective indicator which reflects fetal physiological activity.Separating fetal ECG from mixed signal collected from the maternal body surface is a hot issue in signal processing field . In this paper , we transform mixed signal to wavelet domain and then combine simulated annealing method with ICA to extract independent components ;finally we transform these extracted inversely and choose the FECG component .Actual data experimental results verify the superior performance over conventional ICA algorithm .%胎儿心电信号是反映胎儿宫内生理活动的一种最主要的客观指标，从母体体表混合信号中分离出胎儿心电信号是近年来信号处理领域研究的热点问题。该文利用小波变换将采集到的各导联混合信号变换到小波域；在小波域结合模拟退火法变步长进行独立分量的提取；将提取到的各独立分量进行小波逆变换，从中选择胎儿心电信号成分。通过对临床实际数据进行实验，实验结果提取出了清晰的胎儿心电信号。