改进的双变量收缩函数模型脑电信号消噪方法

针对传统小波消噪全局阈值处理独立性假设和双变量函数模型对没有父系数的最高层小波系数不做处理的缺陷，提出一种高密度离散小波变换中利用双变量收缩函数对脑电信号进行消噪的方法。子小波系数根据双变量函数实现局部自适应收缩处理。同时根据父系数趋于0时，阈值函数近似于软阈值函数，对最高尺度小波系数进行软阈值法消噪。从实际信号处理效果和客观定量指标两方面进行评价，结果表明这种改进算法都优于软阈值法、硬阈值法以及双变量收缩法。

EEG Denoising Method Based on Improved Bivariate Shrinkage Function

Traditional wavelet denoising methods have an assumption that the wavelet coefficients are independent in global thresholding. Then traditional bivariate shrinkage function model has a deficient in not considering the highest scale wavelet coefficients with no parents coefficients.To tackle these defects,in this paper an EEG denois?ing method is proposed based on high density discrete wavelet transform using bivariate shrinkage function. In this method,the children coefficients will achieve local and adaptive shrinking treatment using bivariate function. Then because of the appearance when the parents coefficients tend to zero,the threshold function approximate to the soft threshold,the soft threshold is used for denoising in the highest scale wavelet coefficients. The results show that this improved method is better than the soft threshold,the hard threshold method and bivariate shrinkage method from two perspectives,the actual signal processing effect and the objective quantitative indicators.