Abstract
The identification of modern restoration products with portable instrumentation will enable conservators and conservation scientists to gain information in a shorter time and at a reduced cost, compared to standard micro-destructive techniques. This study employs portable Visible Near-Infrared Fibre Optics Reflectance Spectrometry (Vis-NIR-FORS) as a diagnostic tool to identify and classify the considered products. We assembled a large library made of the most commonly used industrial products in the field of cultural heritage. Thirty products were applied in different concentrations on different substrates (e.g., marble, mortar, polychrome plaster, modern mural painting), and mock-up samples were investigated through Vis-NIR-FORS in the 350–2200 nm range. We then applied several statistical techniques and modern machine learning algorithms such as clustering to the entire database, to quantify the degree of correlation among different spectra. We show how using a multivariate, robust statistical approach can increase the signal-to-noise ratio, thus leading to a more precise identification and localisation of these products. Finally, the proposed procedure was also applied to spectra collected in situ on mural paintings.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call