González-Vidal Juan José, Pérez-Pueyo Rosanna, Soneira María José, Ruiz-Moreno Sergio
Universitat Politècnica de Catalunya, Signal Theory and Communications Department, ETSETB, C/ Sor Eulàlia d'Anzizu s/n, D5, Campus Nord, 08034 Barcelona, Spain.
Appl Spectrosc. 2015 Mar;69(3):314-22. doi: 10.1366/14-07502. Epub 2015 Feb 1.
A new method has been developed to automatically identify Raman spectra, whether they correspond to single- or multicomponent spectra. The method requires no user input or judgment. There are thus no parameters to be tweaked. Furthermore, it provides a reliability factor on the resulting identification, with the aim of becoming a useful support tool for the analyst in the decision-making process. The method relies on the multivariate techniques of principal component analysis (PCA) and independent component analysis (ICA), and on some metrics. It has been developed for the application of automated spectral analysis, where the analyzed spectrum is provided by a spectrometer that has no previous knowledge of the analyzed sample, meaning that the number of components in the sample is unknown. We describe the details of this method and demonstrate its efficiency by identifying both simulated spectra and real spectra. The method has been applied to artistic pigment identification. The reliable and consistent results that were obtained make the methodology a helpful tool suitable for the identification of pigments in artwork or in paint in general.
已开发出一种新方法来自动识别拉曼光谱，无论其对应单组分光谱还是多组分光谱。该方法无需用户输入或判断。因此无需调整任何参数。此外，它会为所得识别结果提供一个可靠性因子，旨在成为分析师决策过程中的有用支持工具。该方法依赖于主成分分析（PCA）和独立成分分析（ICA）的多元技术以及一些指标。它是为自动光谱分析应用而开发的，其中分析光谱由对被分析样品毫无先验知识的光谱仪提供，这意味着样品中的组分数量未知。我们描述了该方法的细节，并通过识别模拟光谱和真实光谱来证明其有效性。该方法已应用于艺术颜料鉴定。所获得的可靠且一致的结果使该方法成为适用于一般艺术品或涂料中颜料鉴定的有用工具。
