Mid-Infrared Fiber-Optics Reflectance Spectroscopy: A Noninvasive Technique for Remote Analysis of Painted Layers. Part II: Statistical Analysis of Spectra

Abstract

Mid-infrared fiber-optics reflectance spectroscopy supported by classification procedures based on the Mahalanobis distance in the principal component space was applied to investigate laboratory samples simulating actual paintings. The spectral data obtained were analyzed by means of principal component analysis (PCA). The application of PCA to first-derivative spectra resulted as a robust method of processing spectral data and made it possible to distinguish the binding medium and/or the pigment/dye, and to classify test samples by means of the Mahalanobis distance discrimination method.