基于小波变换和改进的瞬态独立成分分析融合算法的心电信号降噪方法

Electrocardiogram noise reduction based on fused algorithm of wavelet transform and improved independent component analysis

本文提出一种基于小波变换与独立成分分析融合的信号处理方法,该方法用于抑制多通道同步采集的心电信号包含的噪声。首先利用小波变换对各路同步采集的原始心电信号进行八尺度分解,获得低频逼近信号与高频细节信号,通过设定阈值的方法去除属于低频噪声部分的逼近信号。然后对保留的细节信号进行反变换实现信号重构,再利用包含预同步功能的瞬态独立成分分析改进算法从重构的信号中分离出高频噪声与心电信号独立成分。最后采用信噪比与均方根误差作为信号质量评价指标,将融合算法与单独使用瞬态独立成分分析算法的处理结果进行对比,结果表明融合算法进行降噪处理这一方法具有更高的信噪比和更低的均方根误差,本文提出的融合算法具有良好的心电信号降噪性能。

A signal processing method based on the fused algorithm of wavelet transform and independent component analysis was proposed in the paper. The fused algorithm was used to reduce the noise of electrocardiogram（ECG） signals.Wavelet transform was firstly applied to split the acquired ECG signals into eight scales, obtaining the low- frequent approximation signals and high-frequent detail signals. And the low-frequency noise which was belonged to approximation signal was removed by setting a threshold. Secondly, the maintain detailed signals were inversely transformed to reconstruct the signals, and the independent component analysis algorithm with pre- sync function was applied to separate the independent components which included high- frequent noise and ECG signals from reconstructed signals. Finally, signalnoise ratio and mean square error were taken as evaluate indexes of the signal’s quality to compare the result of noise respectively reduced by using transient component analysis and fused algorithm. The result showed the fused algorithm had higher signal- noise ration and lower mean square error. The proposed fused algorithm has a satisfactory noise reducing performance in ECG signals processing.