Optics for Arts, Architecture, and Archaeology (O3A) VIII

Data acquisition and processing is a critical issue for high-speed applications especially for three-dimensional imaging and analysis. Digital holographic tomography is a potential approach that can quantitatively measure the three-dimensional distribution of the refractive index of any phase object or transparent specimen. Generally, tomography is operated by acquiring projections of the sample and numerically mapping those projections onto a 3D representation using an inverse problem, such as the filtered back projection algorithm. From the practical point of view, there are mainly two ways for recording the data. First, the set of data can be acquired when varying the illumination angle. Last, the data can be acquired by the sample rotation. In both approaches, the sample and the optical set-up must be highly stationary whereas the illumination beam or the object is rotated. Another option is to simultaneously acquire the necessary set of data with a single shot acquisition and then to process them. This would have for advantage of permitting 3D imaging of non-stationary targets or transient time-varying object. The use of multiple camera sensors is complicated and not cost efficient. So, this paper presents the proof of concept for a novel approach based on three color digital holography and the use of a single monochromatic sensor. The principle is based on off-axis holography and spatial multiplexing of multi-wavelength holograms. Three wavelengths from three different laser lines are used to illuminate the target at different incidence angles. The reference beams from the lasers are combined into a single three color beam and the spatial frequencies of the reference waves are adjusted so as to allow for the spatial multiplexing of digital holograms with the monochromatic sensor. After de-multiplexing and processing the color holograms, the amplitude and phase of the target along the views are obtained. Further processing in order to compensate for aberrations of the set-up are proposed and discussed. As proof of concept, we provide results for 3D shape of a 3D ball reconstructed using the inverse Radon transform. These first results are adequate to be exploited in the study of the acoustic field of an ultrasound transducer, for a frequency of 40Khz.

Near-eye displays (NEDs) for augmented and virtual reality (AR/VR) are spotlighted because they have a possibility to provide much more immersive experiences never possible before. With the virtue of recent progress in sensors, optics, and computer science, several commercial products are already available, and the consumer market is expanding rapidly. However, there are several challenging issues for AR and VR NEDs to become closer to our lives. Here, we will explore these issues and important topics for AR and VR, and introduce some of the ideas to overcome them: diffractive optical elements (DOEs), retinal projection displays, and 3D display with focus cues. First, unlike VR with a simple optical system, AR that needs to merge an artificial image with an outer scene requires additional optics. The diffractive elements have the merit of being thin and transparent, suitable for the image combiner. Among them, holographic optical elements (HOEs) have great potential as they can record the desired volume grating from the simple lens to the complex wavefront using light interference. Second, in order to wear the NEDs for a long time, it must deal with the visual fatigue as well as the form factor. Retinal projection display can effectively prevent the vergence-accommodation conflict problem even with a simple optical design. In the retinal projection display, the light rays from the display are adjusted to converge into a small point using a lens. It ensures a wide depth range in which the images are clearly visible. Furthermore, it is possible to provide observers with accurate focus cues for the alleviation of visual fatigue via multi-layer displays and holographic displays. Recently, we conceived tomographic NED that can reproduce dense focal planes. We confirm that this system provides quasi-continuous focus cues, semi-original contrast, and considerable depth of field. The experimental results of our prototypes are explained. We also explain the recent activities of using deep learning in holographic NED system.

The goal of SensApp FET-Open project is to develop an innovative super-sensor that will be able to detect Alzheimer’s disease (AD) biomarkers (β-amyloid, Tau and pTAU) in peripheral blood. Considering that nowadays an accurate diagnosis of AD requires the highly invasive withdrawal and analysis of cerebrospinal fluid, SensApp will represent a breakthrough in the field of AD diagnosis thanks to the ability to detect the early stage of the disease by a simple blood collection. We call Droplet-Spilt-and-Stack (DSS) the new technology that will emerge from SensApp. The achievement of SensApp goal will be insured by the interdisciplinary cooperation between different research institutions and one company involved in the key fields of the project, Vrije University of Brussels, VTT Technical Research Centre of Finland, University of Linz, Ginolis Ltd, IRCCS Centre “Bonino Pulejo”, under the coordination of CNR-Institute of Applied Sciences and Intelligent Systems. This communication will illustrate the progress of the activities. Acknowledgments: The authors acknowledge the EU funding within the Horizon 2020 Program, under the FET-OPEN Project “SensApp”, Grant Agreement n.829104.

Recently a new trend towards a more systematic use of Reflectance Hyperspectral imaging (HSI) has emerged in major museums. Extensive acquisition of HSI data opens up new research topics in terms of comparative analysis, creation and population of spectral databases, linking and crossing information. However, a full exploitation of these big-size data-sets unavoidably raises new issues about data-handling and processing methods. Along with statistical and multivariate analysis, new solutions can be borrowed from the Artificial Intelligence (AI) area, using Machine Learning (ML) and Deep Learning (DL) methods. In this work different algorithms based on multivariate analysis and Artificial Intelligence methods are comparitevely applied to process HSI data acquired on three Picasso’ paintings from the Museu Picasso collection in Barcellona. By using a “data-mining approach” the HSI-data are examined to unveil new correlations and extract embedded information.

Sketch extraction is of great value for historians to copy and study historical painting styles. However, most of the existing sketch extraction methods can successfully perform extraction only if the sketches are well preserved, but for paintings with severe conservation issues, the extraction methods need to be improved. Therefore, we propose a sketch extraction method using spectral imaging and deep learning. Firstly, the spectral image data is collected and the bands sensitive to the sketches are extracted by using the prior knowledge of the sketches (e.g. near infrared bands will be chosen if the sketches are made of carbon ink). A publicly available image dataset of natural scenes is used to pre-train the bi-directional cascade network (BDCN). The network parameters in the model are then fine-tuned by using the sketches drawn by experts based on images of painted cultural objects, so as to solve the problem of insufficient sketch dataset of painted cultural objects and enhance the generalization ability of the model. Finally, the U-Net is used to further suppress unwanted information, to make the sketch clearer. Experimental results show that the proposed method can extract clear sketches even with faded paintings and the presence of unwanted information or instrumental noise. It is superior to the other six advanced extraction methods in visual and objective comparison. The proposed deep learning method is also compared with an unsupervised clustering method using Self-Organising Map (SOM) which is a ‘shallow learning’ method where pixels of similar spectra are grouped into clusters without the need for data labeling by experts.

Recent advances in technology have brought major breakthroughs in deep learning techniques. In this work, we elaborate on such techniques for output data of image processing performed on craquelure patterns in historical paintings. Historical painted objects, especially panel paintings, with their long environmental history, exhibit complex crack patterns called craquelures. These are cracks in paintings that can be referred to as ‘edge fractures’ as they are initiated from the free surface. The analysis has been conducted on the set of selected craquelure patterns on which recent deep learning methods i.e. Neural Networks algorithm is implemented and the results of such self-learning process are discussed.

The work discusses the results of Terahertz Time-Domain Spectroscopy (THz-TDS) imaging carried out on cultural heritage assets. Hyperspectral THz images are corrupted by several degradation effects and their processing pose therefore major challenges. In this work, the limits of THz-TDS are addressed by a twofold computational strategy: (i) removal of the surface warping and (ii) application of a fast joint deblurring-denoising approach for image restoration. The reduction of the main degradation effects is illustrated with the aid of experiments conducted on ancient silver coin and a contemporary painting, highlighting the advantages for cultural heritage applications of imaging spectroscopy in the far-infrared frequencies (3-300 cm-1).

In the last 100 years concrete and steel have been the most common building materials, so that they constitute the core of the structures on which the architecture of the 20th century is based [1]. Both materials present high resistance and durability [2] but the recent collapse of Minardi Bridge in Italy shows that it is absolutely necessary carry out real-time monitoring of the conservation status. In this sense, the building of the Agencia Impositiva Federal in the city of Paraná, built in the 70s, represents a clear example of brutalist architecture in Argentina. After 40 years of its construction and during remodeling tasks, serious problems in the concrete structures were discovered that forced the evacuation of the building. In order to develop an in situ monitoring system [3] that allows the premature detection of structural problems in 20th century building heritage, it was implemented a novel scheme based on digital holographic interferometry (DHI) which includes fiber optics to guide light from a semiconductor laser the optical set up made by 3D printing, that allows to register the dynamic behavior of the concrete based structure. This system was tested on concrete specimens of different composition which were submitted to loads in the range 1000 kg – 28 000 kg by using an Amsler hydraulic press. The obtained results, though preliminary, showed that it was possible to measure the radial deformation on normalized cylindrical specimens in the range of 4-10 micrometers. References: [1] Bell, T. R. (2017). A concept made concrete : conserving a Brutalist icon through architectural intervention. An explanatory document. A research project submitted in partial fulfilment of the requirements for the degree of Master of Architecture (Professional). Unitec Institute of Technology, New Zealand. https://hdl.handle.net/10652/4353 [2] Bolla, G.; López P.; Facendini S. Concrete with recycled aggregates for roads. 16 IRF World Meeting. 2010, Lisbon, Portugal. [3] Kumar R., Dwivedi G., Singh O.; Portable digital holographic camera featuring enhanced field of view and reduced exposure time, Optics and Lasers in Engineering, Volume 137,2021 https://doi.org/10.1016/j.optlaseng.2020

A method for artworks monitoring is proposed, able to perform simultaneously speckle interferometry and speckle photography, towards the mapping of both out-of-plane and in-plane displacements. The system is simple and tailored to some specific needs of artwork diagnostics.

Structural delamination in mural paintings is a complex phenomenon and is considered among the most frequent types of damage. In conservation practice, the most common technique to identify structural detachments in wall paintings is the percussion method, otherwise known as the ‘tap and listen’ approach. Full-field optical techniques based on interferometry, such as shearography, can potentially provide a more scientifically substantiated evaluation of the condition of heterogeneous structures of wall paintings. The empirical nature of the percussion method was observed during the condition assessment of two medieval wall paintings in Maria Church, Nisse, Netherlands. Consequently, the need to obtain more knowledge regarding the structural condition of these wall paintings was established. Furthermore, to allow the formulation of specific treatment needs for structural delamination in wall paintings, accurate defect mapping is needed. The application of shearography coupled with thermography was believed to provide an holistic representation of the structural condition of the wall painting depicting St. Christopher in Maria Church. A comparison of the practical execution and the interpretation of data between shearography and the percussion method was conducted to deliver a description of both techniques. Initial comparison of the two methods confirmed what was already hypothesised: I) the percussion method can provide an approximate indication of delaminated areas. Results, which are considered both non-qualitative and non-quantitative, are dependent on the experience of the conservator; II) shearography can produce qualitative and quantitative results regarding the presence of delamination in complex structures of wall paintings. Analysis of the shearography data together with the percussion method may result in a reliable way to map the defects for future conservation activities. The next steps include optimisation of the thermal excitation for shearography to maximise the defect detection and adapt the shearography output for onsite interpretation by conservators and professionals in the cultural heritage.

The visual appearance of a painting is the result of variations in pigments, layers and layer thicknesses. An important role can be played by ground colour. The ground is the first preparation of the support before the actual painting. Medieval painters worked on white grounds. Coloured grounds appeared during the late 15th century in Italy before spreading to the northern part of Europe around 1550. Painters like Rembrandt or Rubens exploited dark or light coloured grounds to create spectacular, realistic effects. But how can we understand how they do this? To answer this question and add some quantifiable analysis, one needs to resolve the spectral reflectivity profile of a painting in three spatial dimensions. In addition, as the main intent is to relate the optical effects with colours, one needs to operate using visible light. The answer to these challenges is visible-light Optical Coherence Tomography (vis-OCT). In this study, we investigate the use of vis-OCT to measure the reflectivity profile of multi-layered paint samples in three dimensions. Using mock-up paint samples we prepared with two thin paint layers on a black and a white paper support, we measured the reflectivity profiles of the separate colours and of the superposed colours over either the black or white (back)ground. This paper first describes the signal processing involved in rebuilding the spectral reflectivity in three dimensions. Secondly, we will present the results obtained from the mock-up samples and draw conclusions on the possibility to discriminate colours in a three-dimensional context. All reflectivity measurements will be compared to a classical, commercially available hyperspectral imaging setup. This study is executed in the context of the NWO Down to the Ground Project. The results of the OCT measurements will be used by technical art historians and conservators for investigations into coloured grounds.

The paper presents the results of investigation of the art subjects materials micro structure that is the most important in studying the art subjects. The main diagnostic tests were performed using the OCT microscope (OCM) model EX1301 (Michelson Diagnostics, UK), which implements the swept-source Fourier-Domain OCT method. The device design uses the multi-beam OCT optics and a laser swept source model HSL-2000-11-MDL with central wavelength at 1305 nm. The laser wavelength sweep range is about 150 nm with radiation power 15 mW. The axial resolution (in tissue) is about 10 mkm and lateral resolution is 7.5 mkm. The OCM device has been successfully applied to investigate the techniques of khokhloma painting for the background and upper drawing, the vase executed in wood intarsia, jewelry encrusted by the mother of pearl. The experiments of investigation of porcelain, ceramic, natural and artificial origin stone were conducted. The OCM was used for studying thick multi-layer paint structure and varnish coatings and for evaluating properties of natural and imitation skin, canvas and tissues. Results of the study give a possibility to solve diagnostics problems. These include investigation of layered structures like dyes, varnishes, glues, materials with strong scattering and absorption in the analyzed spectral region, the local paint and varnish defects in the layers, areas of occurrence of pollution, to study cracks, splits of the material surface.

We present a practical demonstration of the overall capabilities of high resolution CT technique for the analysis of painted art samples, together with a developed comprehensive methodology. Those samples are made of a support material (panel, wall, yarn), preparatory layer(s) and pictural layer(s). They are multi-material and brittle, which makes them complicated to manipulate and analyze. We defined the visible and extractable features from acquired high resolution CT data (grains, material density, layering, growth lines, cells, yarn pattern). Its ease of access, achievable spatial resolution, non-destructiveness, 3D analysis, followed by overall effectiveness make the high resolution CT a perfect tool for the analysis of painted samples.

This present joint research, undertaken in two different hemispheres, is an effort to address the challenge of early structural and sub-surface assessment of heritage marble architectures, like the Taj Mahal, using two complementary non-contact, non-invasive imaging techniques in the THz range. This unique combination of broadband Terahertz Time Domain Imaging (THz-TDI) and highly sensitive, fast, THz-Laser Feedback Interferometry (THz-LFI) holds immense possibility in large-scale architectural restoration projects as they collectively provide accurate structural depth profile up to several centimeters in the volume of the marble including the strain generated within the structure leading to potential cracks.

This work introduces a novel fast hyperspectral image deblurring and denoising approach tailored to archaeological applications of remote sensing. Hyperspectral data recorded by means of airborne or satelliteborne sensors can be used to detect buried archaeological deposits as the latter have a localised impact on the physical and chemical properties of the soil and the vegetation located above them, contributing to make them structurally different from the surrounding elements. By processing and analysing hyperspectral images, archaeological photo-interpreters can detect subtle changes in the properties of ground elements that can be attributed to the presence of subterranean archaeological sites. Hyperspectral imagery, while rich in content as far as the spectral characteristics of ground elements, often lacks in spatial resolution and contains blurring degradation and noise, prominent especially in some spectral regions. The influence of blur and noise highly effects not only the quality of the visual appearance of the represented objects and compromises the interpretation process, but impacts also further processing of imagery, limiting consequently the detection of targets of interest. The methodology here presented is based on the low-rank properties of hyperspectral images and fully exploits a sparse hyperspectral data representation linked with the self-similarity characteristics of image patches (small image parts). The restoration procedure additionally includes a bend-dependent formulation of blurring degradation. The preliminary results show high performances and reduced computational complexity, and that the proposed approach is able to cope with Gaussian and Poisson noise and band-dependent blur. By removing severe noise and blur, the accurate detection and interpretation of buried structures in different shapes and sizes is thus improved. The proposed approach significantly increases the number of hyperspectral bands that can be used for further image processing and analysis, providing new avenues for features of interest discoveries in bands where they normally obscured by noise.

Carbon content is one of the most influential factors of the mechanical behavior of iron and steel and a time marker of the fabrication period. The quantitative assessment of carbon content distribution in archaeological artefacts gives us insights on their properties and nature. We performed micro-LIBS analysis on archaeological steel in order to reconstruct its carbon content distribution map. The quantitative mapping of the carbon distribution enabled us to infer the different phases of the metallographic structure without time-consuming Nital attack treatment. In addition, LIBS carbon content mapping revealed phases that are invisible with metallographic method.

The digital reproduction of a historical movie should resemble as much as possible the analog film projection at the time of the movie release. Nowadays, practices in digital image capture of films do not properly consider the fundamental elements and conditions of the original film projection. The typical rigid RGB capture cannot adapt to the multitude of color film types to be digitized, and the diffuse illumination on the film generally used by standard digital scanning devices is unable to guarantee the proper visual rendition of the original analog projection of film prints. In order to overcome these problems, we designed and built a novel multispectral imaging system that illuminates the film with a condensed light beam. The light comes from the exit port of a custom-made integrating sphere, whose interior is illuminated by a set of ten LEDs that are switched on in succession for multiple image capture. The light coming out from the sphere is shaped by two distinct condensers chosen to minimize the light loss and to maximize the illuminated area. To obtain accurate transmittance data, a special flat-fielding operation had to be developed to compensate for the unstable temperature-dependent luminous intensities emitted by the LEDs. The new imaging system and the computational pipeline were tested on an assorted set of photographic slides. The level of accuracy of the transmittances obtained with the new multispectral imager guarantees color differences that fall below the color discrimination threshold of the human eye. In view of these results, the new optical design represents a promising solution for the creation of a new generation of motion picture film scanners.

Infrared methods are of great importance in nondestructive testing of artworks, allowing a remote and wide-field imaging of interesting hidden features. Here we discuss a workflow based on thermal imaging in the mid infrared 3-5 micron range for the evaluation of subsurface defects in frescoes. Particular attention is payed to obtaining a high resolution (submillimetric) localization of the defects. The transfer of diagnostics techniques into real world applications, is discussed through the proof of concept of the proposed workflow on frescoes at the Sforza Castle (Milan, Italy).

Degradation of paper and the consequent loss of its chemical and mechanical properties has been studied for many decades. For conservators an understanding of the loss of mechanical properties of paper during degradation is highly relevant from a practical point of view. Considering the fibrous microstructure of paper, the mechanical properties of single cellulose fibers and their change in time has been of interest in paper degradation studies. With current advances in experimental micromechanics, the characterization of cellulose fibers is achievable at small scales and with very high accuracy. Such detailed analyses on naturally aged paper in combination with accelerated aging experiments lead to valuable insight in the degradation of paper. The current study represents a novel methodology for the accurate mechanical characterization of micro-scale fibers, such as cellulose fibers. To this end, in-situ mechanical tests are performed using high resolution optical profilometry (white light interferometry) in combination with Digital Image Correlation (DIC) to attain detailed strain measurements. Given the curly geometry of cellulose fibers, there is considerable out-of-plane deformation during a micro-tensile test, making is necessary to use profilometric images. An in-house built setup is used for application of micro-scale speckle pattern needed for DIC. Reliable evaluation of the cross-sectional area of fibers is attained by acquiring profilometric images of both the front and back sides of each fiber, using a mirror in a novel self-calibrating setup. Detailed analyses of the results from different samples show the accuracy and reliability of the measurements and the methodology. By these means, an accurate evaluation of the stiffness and strength of cellulose fibers is attained. Performing such measurements on cellulose fibers from aged papers is expected to result in better understanding of the effect of cellulose degradation on the mechanical properties of paper.

Fluorescence is a photoluminescence phenomenon where light is absorbed at lower wavelengths and re-emitted at longer wavelengths. For classic artworks, fluorescence gives useful information about varnish and retouches. At the same time, modern artworks may employ synthetic fluorescent pigments because of their special appearance properties, such as increased brightness and vividness provoked by self-luminescence. Hence, it is relevant to investigate the fluorescent signals of cultural heritage objects when studying their appearance. This work proposes a variant to Reflectance Transformation Imaging (RTI) technique, namely Fluorescence Transformation Imaging. Reflectance Transformation Imaging method outputs a single-camera multi-light image collection of a static scene, which can be used to model the reflectance of the scene as a polynomial of the illumination directions. Similarly, Fluorescence Transformation Imaging aims to model the fluorescent signal based on a series of images with fixed scene and viewpoint and varying incident light directions - what changes with respect to RTI is that the wavelength of incident light needs to be shorter than the sensing wavelength. In the literature, there are works that explore the isotropic property of fluorescence in low-dimension multi-light imagery methods (such as Photometric Stereo) to model the appearance of an object with a first-order polynomial. This is because in the fluorescent mode the object gets closer to a Lambertian surface than in the reflective mode where non-Lambertian effects such as highlights are more likely to appear. Nonetheless, this assumption stands for single-object scenes, with uniform albedo and convex geometries. When there are multiple fluorescent objects in the scene, with concavities and non-uniform fluorescent component, then the fluorescence can become secondary light to the object and create interreflections. This paper explores the Reflectance and Fluorescence Transformation Imaging methods and the resulting texture maps for appearance rendering of heterogeneous non-flat fluorescent objects.

This contribution presents some examples of the application of ultraviolet fluorescence (UVF) photography and photogrammetry on cultural heritage objects and paintings. The used approach is based on low-cost systems sustainable in terms of costs for restorers and conservators, but, at the same time, able to supply valid and relevant information for a better knowledge of the preservation state, of previous restoration interventions and of possible superimposed materials. This knowledge is fundamental for planning and addressing the restoration activities, such as cleaning and consolidation, in the most appropriate way as possible. Specific cases studies will be reported concerning large paintings and 3D objects. In the first case, it has been necessary to acquire several images by dividing the surface into various parts, as function of the dimensions of the investigated area, and then by using photogrammetric tools to recompose the images into a single one (orthophotomosaic of the painting). In the second case, the UVF images are acquired all around the 3D objects and then processed in order to obtain a photorealistic model under ultraviolet fluorescence. The image acquisitions were obtained by using a digital camera equipped with different lens and filters. Lighting of the surfaces was obtained by UV projectors. The acquired images have been then processed through photogrammetric tools in order to obtain the final 2D and 3D ultraviolet fluorescence outputs of the investigated artworks that can be explored and profitably used for gathering metrically precise information and material response to UV radiation.

Digital image correlation (DIC) is a well-established technique for the measurement and monitoring of displacements and strains in engineering objects. In technical applications, the surface of an object is modified by applying artificial random structure and monochrome cameras are typically used. The challenge in the measurement of Cultural Heritage objects (CHO) lies in the necessity of using the natural surface texture, which in most cases is far from the ideal random pattern required for DIC. On the other hand, spectrally separated content of natural CHO textures captured by color cameras may provide additional spatial information and support DIC analysis. Therefore the main objective of this work is to gain knowledge on the best color image preprocessing path including the selection of the best demosaicing algorithms for the spectral and monochrome channels with respect to minimizing displacement reconstruction errors.

The technical photography (TP) method has been using in identifying pigments in paintings and the results could be used in the early stage before confirming with other methods such as X-Ray Fluorescence and Raman spectroscopy. The TP method employs visible, UV and IR as light sources with different filters. Moreover, the multi-spectrum method, using a continuous light source and various filters, has been used in the same application. This research proposed a method of using narrow multi-colored LEDs as light sources instead of colored filters. It aimed to optimize the number of LED channels for identifying pigments in conservation and restoration applications. Twelve tunable single-wavelength LEDs having wavelengths of 400-700 nm and 2 white LEDs, were used as light sources for capturing images of 357 Kremer pigments, applied on white card paper, without an optical brightening agent, through a visible pass filter, located in front of the single lens reflex Canon 5D Mark II camera that was modified by removing the UV and IR blocks. The 357 pigments including 9 main colors: 50 red pigments, 37 orange pigments, 82 yellows, 38 greens, 38 blues, 17 purples, and a group of greys browns, blacks and whites were used as samples. Subsequently, the pigments were classified, based on the captured images obtained from the individual LED. The redundancy of the CIEL*C*h data of each LED was reduced using the principal component analysis and the number of LEDs was optimized. It was found that all pigments could be classified and some pigments could be identified using single wavelength LEDs. The PCA should be applied to the CIEL*C*h data of pigments having similar shade. The 598 nm 425 nm LEDs were optimized for identifying and classifying blue pigments. The 430 nm, 640 nm and 540 nm were for purple, 403 nm, 540 nm, 660 nm and 672 nm for red, 403 nm, 447 nm, 540 nm and 672 nm for orange, 503 nm, 540 nm and 598 nm for yellow, 503 nm, 598 nm and 640 nm for green and 403 nm, 503 nm, 540 nm and 660 nm for a group of white, grey, black and brown.

During 15th and 16thcenturies, while painters in Flanders or Italy used chalk (calcium carbonate) in ground layers of panel paintings, master painters in Spain employed gypsum (calcium sulphate dihydrate). Thereby, the identification of these minerals in such artworks is relevant for historical studies, especially for periods when there was an active artistic exchange between those countries. Visible and near-infrared reflectance imaging spectroscopy is already applied to characterize pigments and binders in paintings. In addition, there are investigations suggesting that, due to the transparency of some pigments mixed with binder in the near-infrared range, the signal coming from the subjacent substrate may be identified in the reflectance spectra, together with the bands of absorption of pigments and binders. This work proposes a non-invasive methodology involving near-infrared reflectance imaging spectroscopy, for distinguishing between chalk- and gypsum-based ground layers. The presented methodology has been validated using reference samples of oil pigments applied on both preparation layers. Spectral images of samples of gypsum, chalk and pigments, powdered and mixed with binder have been compared with the spectra of the reference samples, revealing the distortions on the reflectance curves of pigments due to the ground layers. The spectral images have been acquired with the VARIM 2.0 system, endowed with an InGaAS detector camera, coupled to an image spectrograph, covering the range from 1000 nm to 1700 nm. Moreover, several real cases have been examined. Study areas takes into account zones affected by retouching, rectifications and conservative interventions, previously identified by non-invasive imaging techniques. The results prove the capability to discriminate gypsum and chalk on ground layers of panel paintings using a non-invasive methodology that involves near-infrared reflectance imaging spectroscopy and without sampling. Consequently, this technique can be a new and important tool to study the origin of panel paintings from the 15th and 16th centuries.

Hyperspectral imaging has become a powerful technique for the non-invasive investigation of works of art, thanks to the possibility to obtain spectral information over the spatial region, allowing the identification and mapping of constituent materials. While hyperspectral imaging has been extensively used on paintings and manuscripts, little work has been done on stained glass. In this paper, a workflow for the imaging and analysis of stained-glass windows is proposed. Contrarily to previous works on the topic, which deal with in-situ applications, the acquisition workflow described in this paper relates to a laboratory set-up. In this configuration, the imaging was carried out in transmittance, under controlled illumination, using a HySpex 1800 camera, which works in the VNIR region. The sample was placed on a translator stage equipped with a diffusing panel, with the light source positioned below the diffuser. To test the validity of the set-up, a mock-up stained-glass panel, originally employed for a project at the Fairford Church (England), was used. A detailed description of the acquisition set-up and the pre-processing steps will be provided in the paper. The processing was carried out with two aims: object visualization and chromophore identification. Regarding visualization, both RGB and false-color images were created by extracting suitable bands from the data-cubes, while the appearance of the glass under different illuminants was rendered using color matching functions. Regarding chromophore identification, the RGB and false-color images were compared, to understand whether similar colors had different compositions. The spectra of each tile were then extracted and compared, and pXRF measurements were carried out to obtain complementary information on the glass composition.