拉曼光谱预处理中几种小波去噪方法的分析

Denoising of raman spectra based on wavelet transform

拉曼光谱分析中,噪声的存在常影响分析的准确度和检测限.以钙长石的拉曼光谱为研究对象,探讨小波变换在拉曼光谱信号去噪方面的应用,分别采用移动窗口最小二乘多项式平滑、移动窗口中位数平滑、非线性小波软硬阈值法和小波变换模极大值法对加噪后的拉曼光谱进行去噪并对去噪效果进行比较.结果表明,小波变换模极大值光谱去噪法得到了较高的信噪比,小波软硬阈值法次之,其他2种方法去噪效果较差.小波变换模极大值法能够有效去除光谱噪声,并很好地保留了光谱信号特征,为拉曼光谱的校正模型的建立奠定了良好的基础.

During the spectrum analysis process, noise usually influences the analytical accuracy and the detection limit. Taking the Raman spectra of anorthite as the research object, the application of wavelet denoising to Raman spectra was discussed. Savitzky-Golay smoothing, moving window median smoothing, the nonlinear wavelet soft and hard threshold denoising method and wavelet transform modulus maxima method were applied respectively to pure Raman spectra with added noise, the performance of these wavelet denoising methods was compared. The results show that the wavelet transform modulus maxima method obtained a high signal to noise ratio, followed by the soft and hard threshold wavelet method. The other two methods were less effective. Wavelet transform modulus maxima method is able to eliminate spectroscopic noises and interferences while reserving major information. It contributes to the foundation of Raman spectra correction model.