基于EEMD的中药三维荧光光谱去噪方法研究

Research on denoising method of three dimensional fluorescence spectrum of traditional Chinese medicine based on EEMD

针对中药三维荧光光谱信号特征提取中的噪声问题,基于集合经验模态分解(EEMD)的去噪原理,对中药溶液的三维荧光光谱信号进行去噪研究。利用相关系数和箱线图筛选固有模态分量(IMF),将筛选后的IMF进行重构得到去噪后的数据。并探讨了EEMD分解时添加的高斯白噪声的幅值及设定相关系数的大小对去噪效果的影响。结果显示,当高斯白噪声的幅值为原信号标准差的0. 2倍,相关系数大于0. 5筛选IMF分量时,EEMD方法能很好地消除中药白术溶液三维荧光光谱信号中的噪声。进一步将该方法与经验模态分解(EMD)及小波去噪方法进行比较,结果表明,EEMD方法对中药溶液三维荧光光谱数据去噪效果优于EMD及小波去噪方法。

Based on the denoising principle of ensemble empirical mode decomposition(EEMD),the three-dimensional fluorescence spectrum signal of traditional Chinese medicine solution is denoised.The intrinsic mode function(IMF)is filtered by correlation coefficient and box diagram,and the filtered IMF is reconstructed to get the data after denoising.The influence of the amplitude of Gauss white noise added by EEMD decomposition and the setting correlation coefficient on the denoising effect are also discussed.The results show that when the amplitude of Gauss white noise is 0.2 times that of the original signal standard deviation and the correlation coefficient is greater than 0.5 to screen the IMF component,the EEMD method can effectively eliminate the noise in the three dimensional fluorescence spectrum signal of traditional Chinese medicine Atractylodes macrocephala.The method is compared with the method of empirical mode decomposition(EMD)and wavelet denoising.The results show that the denoising effect of EEMD method on the three-dimensional fluorescence spectrum data of traditional Chinese medicine solution is better than that of EMD and wavelet denoising.