【摘要】 人体心电信号在采集过程中掺杂着各种噪声信号,而常用的小波阈值法在心电信号去噪中存在缺陷,为此在小波阈值法的基础上进行改进,得到新的阈值去噪方法,该方法能够较好保护心电信号特征,有效保持各种心电波形的幅度,并在一定程度上有效抑制脉冲噪声.此外,由于传统小波阈值法去噪都是基于离散小波变换进行的,所以在幅度较小的Q,S波处会产生Pseudo-Gibbs现象,而文中以改进的小波阈值法为基础对心电信号进行平移不变的去噪处理,有效地克服了这个问题.利用美国麻省理工学院的MIT-BIH心电数据库对以上方法进行验证,取得了良好的去噪效果.实验结果表明,所提出的算法与传统的小波阈值法比起来能够更好的保持心电信号的几何特征,且具有更高的信噪比.更多还原

【Abstract】 The electrocardiogram signals may mix various kinds of noises while being gothered and recorded.Wavelet thresholding de-noising method proposed by Donoho et al.is often used in de-noising of the electrocardiogram signals.Due to the defects of Donoho’s method in de-noising of electrocardiogram signal,an improved thresholding de-noising method was proposed.The advantage of the improved de-noising method is that it may not only remain the geometrical characteristics of the original electrocardiogram signal and keep the amplitudes of various electrocardiogram waveforms efficiently,but also suppress impulsive noise to some extent.Furthermore,the traditional wavelet thresholding de-noising method causes Pseudo-Gibbs phenomena in Q and S waves of the electrocardiogram signal due to discrete wavelet transform.In order to suppress Pseudo-Gibbs phenomena in Q and S waves,a new de-noising method combining above improved wavelet thresholding method with the translation-invariant de-noising was proposed.Finally,the American MIT-BIH database was used to validate the methods.The experimental results indicate that the proposed methods in this paper are better than traditional wavelet thresholding de-noising methods in aspects of remaining geometrical characteristics of electrocardiogram signal and the signal-to-noise ratio.更多还原