摘要
光谱在采集过程中,时常伴随随机噪声,为获得高质量的样品光谱信息,必要的光谱降噪处理成为光谱预处理环节的重要组成部分,目前,针对高维激光拉曼光谱的构建与降噪研究相对较少。以羊奶粉为例,提出了以激光强度为外扰的二维相关高维拉曼光谱的构建,实现了样品的三维拉曼光谱分析,展现了更为丰富的光谱信息;其次,提出了以相关系数为评价指标的高维拉曼光谱降噪策略评估方法,研究了Savitzky-Golay滤波,Haar小波,Daubechies小波和Biorthogonal小波处理拉曼光谱数据的降噪效果。结果显示,拉曼光谱数据蕴含丰富的样品化学特征信息,通过光谱降噪处理可有效降低随机噪声的影响,不同降噪函数的光谱处理效果存在差异,以相关系数为评估指标可量化分析降噪函数的适用性,针对实验体系,SavitzkyGolay滤波降噪处理效果较好。
The spectra are often accompanied by random noises during the collection process,and in order to obtain high-quality spectral information of the sample,the necessary spectral noise reduction is an important part of the spectral pre-processing.At present,there are relatively few researches on the construction and noise reduction of high-dimensional laser Raman spectroscopy.Taking goat milk powder as an example,a two-dimensional correlation high-dimensional Raman spectroscopy with laser intensity as the external disturbance was proposed,and the sample was analyzed in three-dimensional Raman spectroscopy,which showed more abundant spectral information.Secondly,the evaluation method of high-dimensional Raman noise reduction strategy based on correlation coefficient was proposed,and the noise reduction effects of Savitzky-Golay filter,Haar wavelet,Daubechies wavelet and Biorthogonal wavelet on Raman spectral data were studied.The results show that Raman spectral data contains abundant sample chemical feature information,the effect of random noise can be effectively reduced by spectral noise reduction.Spectral noise reduction can effectively reduce the influence of random noise,and the spectral processing effects of different noise reduction functions are different.The applicability of the noise reduction function can be quantified by using correlation coefficient as the evaluation index.For the experimental system,the Savitzky-Golay filter has better noise reduction effect.
作者
张正勇
桂冬冬
马蕴文
沙敏
王海燕
Zhang Zhengyong;Gui Dongdong;Ma Yunwen;Sha Min;Wang Haiyan(School of Management Science and Engineering,Nanjing University of Finance and Economics,Nanjing,J iangsu210023,China;Z School of Management Engineering and Electronic Commerce,Zhejiang GongshangUniversity,Hangzhou,Zhejiang 310018,Chin;Jiangsu Province Institute of Quality and SafetyEngineering,Nanjing,Jiangsu 210023,China)
出处
《应用激光》
CSCD
北大核心
2018年第3期468-473,共6页
Applied Laser
基金
国家自然科学基金资助项目(项目编号：61602217,71433006,91746202)
江苏省高校自然科学研究面上资助项目(项目编号：16KJB150015)
关键词
激光拉曼光谱
降噪
高维数据
奶粉
二维相关
laser Raman spectroscopy
noise reduction
high-dimensional data
milk powder
two-dimensional correlation
