Abstract: Since many materials behave as heterogeneous intimate mixtures with which each photon interacts differently, the relationship between spectral reflectance and material composition is very complex. Quantitative validation of spectral unmixing algorithms requires high-quality ground truth fractional abundance data, which are very difficult to obtain.In this work, we generated a comprehensive hyperspectral dataset of intimate mineral powder mixtures by homogeneously mixing five different clay powders (Kaolin, Roof clay, Red clay, mixed clay, and Calcium hydroxide). In total 325 samples were prepared. Among the 325 samples, 60 mixtures were binary, 150 were ternary, 100 were quaternary, and 15 were quinary. For each mixture (and pure clay powder), reflectance spectra are acquired by 13 different sensors, with a broad wavelength range between the visible and the long-wavelength infrared regions (i.e., between 350 nm and 15385 nm) and with a large variation in sensor types, platforms, and acquisition conditions. We will make this dataset public, to be used by the community for the validation of nonlinear unmixing methodologies (https://github.com/VisionlabUA/Multisensor_datasets)

Abstract: This study presents a novel method for the detection and quantification of atmospheric corrosion products on carbon steel. Using hyperspectral imaging (HSI) in the short-wave infrared range (SWIR) (900-1700 nm), we are able to identify the most common corrosion minerals such as: alpha-FeO(OH) (goethite), gamma-FeO(OH) (lepidocrocite), and gamma-Fe2O3 (maghemite). Six carbon steel samples were artificially corroded in a salt spray chamber, each sample with a different duration (between 1 h and 120 hours). These samples were analysed by scanning X-ray diffraction (XRD) and also using a SWIR HSI system. The XRD data is used as baseline data. A random forest regression algorithm is used for training on the combined XRD and HSI data set. Using the trained model, we can predict the abundance map based on the HSI images alone. Several image correlation metrics are used to assess the similarity between the original XRD images and the HSI images. The overall abundance is also calculated and compared for XRD and HSI images. The analysis results show that we are able to obtain visually similar images, with error rates ranging from 3.27 to 13.37%. This suggests that hyperspectral imaging could be a viable tool for the study of corrosion minerals.

Abstract: As part of its conservation-restoration, the 13th century stained-glass panel ‘the Annunciation’, was examined at the micro- and macro level. This window, since 1898 in the collection of the Museum Mayer Van den Bergh (Antwerp, B), was formerly a part of the southern Rose window of the Notre Dame Cathedral (Paris, F). The insigths emerging from a first phase of the analysis, comprising non-invasive analysis techniques such as optical microscopy combined with macroscopic X-ray fluorescence (MA-XRF) and X-ray diffraction (MA-XRPD) mapping, were used to select sampling positions for the second phase of investigation that involved micro-invasive analysis, namely scanning-electron microscopy coupled to energy-dispersive X-ray analysis (SEM-EDX). The aim of the investigation was fourfold: (1) to assess the applicability of MA-XRF scanning for the characterisation of stained glass windows prior to any conservation or restoration procedure, (2) to assess the applicability of MA-XRPD scanning to identify the degradation products formed on the surface of stained glass windows, (3) to establish a method to limit the set of sampled glass fragments taken from a glass panel for quantititive analysis while maintaining sufficient representativeness and (4) to distinguish the original glass panes and grisaille paint from non-original glass panes that were inserted during various past interventions. Most of the panes in this window proved to consist of medieval potash glass, consistent with the 13th c. origin of the window while a limited number of panes were identified as non-original infills, with divergent glass compositional types and/or colorants. Most panes derive their color from the pot metal glass (i.e. homogenously colored) they were made of. Some of the panes that originally had a red flashed layer on their surface, completely or partially lost this layer due to weathering. Three main compositional glass families with similar color could be defined. With the exception of the yellow and orange panes, the chromophoric elements responsible for the dark(er) and light(er) blue (Co), green (Cu), purple (Mn) and red colors (Cu) were identified. Two different grisaille paints were encountered, part of which were restored during the 19th century. On the basis of this information, all missing pieces were replaced by glass panes with appropriate colors and the panel could be successfully conserved to its former glory. On the surface of several panes, typical glass degradation products such as calcite, syngenite and gypsum were identified, together with lead based degradation products such as anglesite and palmierite. In addition, the presence of hematite and melanotekite in the grisailles was observed; also the presence of Zn, uncorrelated to Cu, in the grissailes on the right side of the window became apparent.