基于相关性的小波熵心电信号去噪算法  

Wavelet entropy denoising algorithm of electrocardiogram signals based on correlation

作　　者：王晓燕[1] 鲁华祥[1,2] 金敏[1] 龚国良[1] 毛文宇[1] 陈刚[1] WANG Xiaoyan LU Huaxiang JIN Min GONG Guoliang MAO Wenyu CHEN Gang(Institute of Semiconductors, Chinese Academy of Sciences, Beijing 100083, China Center for Excellence in Brain Science and Intelligence Technology, Chinese Academy of Sciences, Shanghai 200031, China)

机构地区：[1]中国科学院半导体研究所,北京100083 [2]中国科学院脑科学与智能技术卓越创新中心,上海200031

出　　处：《智能系统学报》2016年第6期827-834,共8页CAAI Transactions on Intelligent Systems

基　　金：中国科学院战略性先导专项(xdb02080002);中国科学院国防实验室基金项目(CXJJ-16S076);青年自然科学基金项目(61401423)

摘　　要：针对心电信号的基线漂移、工频噪声、肌电噪声,本文提出了基于相关性的小波熵去噪算法。算法首先根据基线漂移的低频特性,确定小波分解的层数,置零近似系数,去除基线漂移;再对相邻尺度的高频小波系数进行相关处理,依据小波熵自适应地计算全局阈值去除工频和肌电噪声;最后将置零的近似系数和阈值处理后的小波系数重构得到有效信号。该算法能够在一次小波分解、重构的过程中,同时滤除心电信号中的3种主要噪声。对MIT-BIH数据库数据和模拟数据的仿真实验结果也表明该算法的去噪效果显著优于其他算法。In view of the baseline drift, power line interference and muscle noise of electrocardiogram （ECG） signals, the wavelet entropy denoising algorithm of ECG signals based on correlation was proposed. First, ECG signals were decomposed using wavelets to determine the number of scale of wavelet decomposition, and the lowest approximation coefficients were each set to zero, so as to remove the baseline drift. Then, the high-frequency wavelet coefficient of adjacent scales was processed by adaptively calculating the global threshold with the correlation coefficients between the adjacent scales, to remove the power line interference and the muscle noise. Last, the denoising signals were reconstructed using zero approximation coefficients and processed wavelet coefficients. Using this method, three kinds of noise were removed in one process of wavelet decomposition and reconstruction. Experiments using the MIT-BIH database and simulative data prove that the algorithm is much better than others in ECG denoising with low complexity.

关 键 词：心电信号 去噪 相关性 小波熵 自适应 

分 类 号：R540.4[医药卫生—心血管疾病] TN911.7[医药卫生—内科学]