基于提升小波改进阈值的光谱信号去噪研究

Research on spectrum signal denoising based on improved threshold with lifting wavelet

对傅里叶红外光谱信号的预处理进行研究,分析了小波变换在光谱信号去噪中的应用,针对传统小波阈值去噪中存在的缺点,对阈值函数进行改进。改进方法通过保留部分低于阈值的分解系数,克服了硬阈值法不连续,以及软阈值法估计系数和真实系数具有恒定偏差的缺点。实验采用基于提升小波的改进阈值函数法对铜陵市市区上空空气实测光谱信号进行预处理,结果表明,与经典小波变换和Donoho软阈值、硬阈值法相比,相同条件下信噪比、均方差、运行速度均有所提高。

In this paper,the signal preprocessing of infrared spectroscopy was researched by using Fourier transform,and the application of the wavelet transform in spectrum signal denoising was analyzed. For the shortcomings of traditional wavelet thresholding,the threshold functionwas improved. By retaining part of the decomposition coefficients below the threshold,the improved method overcame that the hard-threshold method was not continuous,and the soft threshold hada constant variation of the real shortcomings. The experiments,in which the improved method based on lifting wavelet was used,preprocessed the measured air spectrumin Tongling City urban. The results show that compared with the classical wavelet transform and Donoho soft threshold value and hard threshold value method,the proposed method can improvethe signal to noise ratio（ SNR）,the root mean square error（ RMSE） and the running speed.