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 Reviewed