Evaluation of pictorial layers ageing by HyperSpectral Imaging

The possibility to perform a full characterization and identification of pigments in pictorial layers represents one of the most challenging targets in the field of cultural heritage, especially if their fulfillment is achieved setting up not invasive and not destructive analytical procedures. In this perspective, HyperSpectral Imaging (HSI) based sensing architectures, acting in the VIS-NIR and SWIR wavelength ranges, can represent an important diagnostic tool to realize the monitoring of paintings, as well as to evaluate their preservation status [1]. The goal of this study is to evaluate the alteration of different pictorial layers, as resulting from the combination of different supports, pigments and binders, according to different irradiation times. Starting from the result of recent studies [2], 6 pigments were thus mixed (i.e. red ocher, yellow ocher, chrome yellow, mineral blue, cobalt green and malachite) with four different binders (i.e. Arabic gum, gouache, egg tempera, and linseed oil), applying them on two different media (i.e. wood and canvas). The resulting samples have been aged in solar box and acquired at different aging times. Surface spectra of each sample were then acquired in VIS-NIR and SWIR ranges. Starting from the collected information (i.e. hypercube) chemometric techniques have been utilized to extract and handle pictorial layers information. The sequential application of Principal Component Analysis (PCA) and Partial Least Square-Discriminant Analysis (PLSDA) was then applied in order to study the pictorial layers spectra variations linked to the effects of the applied aging cycles. The sensing architectures have been designed and implemented, both at hardware and software level, in order to perform both in-situ (i.e. directly where the painting is) and ex-situ (i.e. at laboratory scale) analyses. The results showed as the developed modelling procedure allows to correctly predict sample characteristics and composition. Furthermore, it is objective, rapid and non-destructive. The implementation of such approach could allow to perform a full monitoring of the state of conservation and/or the characterization of materials and/or supports not easy to assess with conventional techniques