The interpretation of a Raman spectrum is based on the identification of its characteristic molecular bands. However, the assignment of the vibrational modes is often compromised by the presence in the spectrum of an intense fluorescence background that covers the measured spectra. Several techniques have been employed to minimize the presence of this fluorescence in order to resolve and analyze Raman spectra. In this paper a new automated method for fluorescence subtraction is described, based on morphology operations. This method is compared with the most commonly used polynomial fitting methods. Results indicate that the proposed automated method is efficient in fluorescence subtraction and retains the line shapes and positions of the Raman bands in the spectra.

Noise filtering is considered a crucial step for the proper interpretation of Raman spectra. In this work, we present a new denoising procedure which enhances the Raman information whilst reducing unwanted contributions from the most frequent noise sources, i.e. the shot noise and the fluorescence's baseline. The procedure increases the signal-to-noise ratio whilst preserving simultaneously the shapes, positions and intensity ratios of the Raman bands. The method relies on cubic penalized spline fitting and mathematical morphology and requires no user input. We describe the details of this method and include a benchmark to study the performance of the presented approach compared with the most commonly used denoising techniques. The method has been successfully applied to improve the signal quality of Raman spectra from artistic pigments. The reliable results that were obtained make the methodology a useful tool to help the analyst in the interpretation of Raman spectra from pigments in artworks. Copyright © 2017 John Wiley & Sons, Ltd.Peer ReviewedPostprint (author's final draft

No abstract

In an art historical context, pigments identification is exceedingly important for the conservation, interpretation and preservation of art works. In practice, the pigments may have been used in mixtures (or in admixtures) with other pigments to produce special effects or tonal qualities. Binary mixtures are the more frequently mixtures used in paintings; for instance, it is commonly used the mixture of a yellow pigment with a blue pigment to produce green colors. In this situation, the pigments identification is actually done on Raman spectra of pigment mixtures and may turn out to be a complex and tedious task, especially when analyzing spectra with a large number of bands located close together as it is the case of mixtures of synthetic organic pigments. To decrease complexity whilst also speeding up the identification process, a system to automatically identify Raman spectra of binary mixtures of pigments is presented in this work. The system is able to identify the two different pigments in the mixture from spectroscopic signature obtained by Raman spectroscopy. The technique has been proved with mixtures showing its robustness against some of the critical factors that could affect the application of Raman spectroscopy for pigments identification as distortion and wavenumber shifts in key Raman bands due to different measurement environmental conditions. The results led to conclude that the system could be a useful tool to help the analyst to make a decision. Copyright © 2012 John Wiley & Sons, Ltd.

A filter based on the fuzzy logic technique to enhance the signal-to-noise ratio of Raman spectra is presented. The reasoning for generating a fuzzy filter is explained and its performance is evaluated on Raman spectra contaminated with cosmic ray events and shot noise. The filter suppresses noise and simultaneously preserves information without requiring a priori knowledge on the accurate shape of the Raman band and on the statistics of the noise that masks the information signal.