Digital restoration of colour cinematic films using imaging spectroscopy and machine learning.

Digital restoration is a rapidly growing methodology within the field of heritage conservation, especially for early cinematic films which have intrinsically unstable dye colourants that suffer from irreversible colour fading. Although numerous techniques to restore film digitally have emerged recently, complex degradation remains a challenging problem. This paper proposes a novel vector quantization (VQ) algorithm for restoring movie frames based on the acquisition of spectroscopic data with a custom-made push-broom VNIR hyperspectral camera (380-780 nm). The VQ algorithm utilizes what we call a multi-codebook that correlates degraded areas with corresponding non-degraded ones selected from reference frames. The spectral-codebook was compared with a professional commercially available film restoration software (DaVinci Resolve 17) tested both on RGB and on hyperspectral providing better results in terms of colour reconstruction.

Near-infrared hyperspectral imaging (NIR-HSI) and normalized difference image (NDI) data processing: An advanced method to map collagen in archaeological bones.

In the present study, an innovative and highly efficient near-infrared hyperspectral imaging (NIR-HSI) method is proposed to provide spectral maps able to reveal collagen distribution in large-size bones, also offering semi-quantitative estimations. A recently introduced method for the construction of chemical maps, based on Normalized Difference Images (NDI), is declined in an innovative approach, through the exploitation of the NDI values computed for each pixel of the hyperspectral image to localize collagen and to extract information on its content by a direct comparison with known reference samples. The developed approach addresses an urgent issue of the analytical chemistry applied to bioarcheology researches, which rely on well-preserved collagen in bones to obtain key information on chronology, paleoecology and taxonomy. Indeed, the high demand for large-sample datasets and the consequent application of a wide variety of destructive analytical methods led to the considerable destruction of precious bone samples. NIR-HSI pre-screening allows researchers to properly select the sampling points for subsequent specific analyses, to minimize costs and time and to preserve integrity of archaeological bones (which are available in a very limited amount), providing further opportunities to understand our past.

Non-invasive characterisation of molecular diffusion of agent into turbid matrix using micro-SORS.

This study proposes a non-invasive analytical method to study the molecular diffusion of a chemical agent into a turbid matrix with an emerging analytical technique, micro-Spatially Offset Raman Spectroscopy (micro-SORS). Here, the micro-SORS concept has been extended from the analysis of chemically distinct stratified layers to the studies and monitoring of the absorption and diffusion processes, addressing a key analytical need in a number of areas including polymer, pharmaceutical, forensic and biomedical sciences. In Cultural Heritage the knowledge of the penetration depth of a polymer used to consolidate or to protect an object, or the absorption depth of solvents used during a cleaning procedure is crucial for the performance evaluation of restoration methods and their safety towards the work of art. To date the most common protocol for obtaining this type of information comprises the application of stratigraphical analysis on cross-sections prepared after taking a small amount of sample from the work of art. This approach is destructive and may lack of statistical meaning, since the analytical information is limited to the micro area of sampling. To overcome these drawbacks, in this study micro-SORS was successfully used, for the first time, to non-invasively characterise the penetration of a polymer and of a viscous solvent into a gypsum substrate, permitting the reconstruction of the diffusion trends of the products into the matrix and the evaluation of their performances.

Analyses of trace amounts of dyes with a new enhanced sensitivity FTIR spectroscopic technique: MU-ATR (metal underlayer ATR spectroscopy).

The identification of organic dyes is a challenging task in all the fields such as the forensic and conservation sciences, especially in cases where the amount of sample is extremely small. In this paper we propose a new enhanced FTIR method (MU-ATR metal underlayer ATR spectroscopy), which we believe is the first of its kind, for the analysis of a few ng of dyes. With this method, dyed fiber micro-extracts can be analyzed using a commercial FTIR microscope with a fixed incident angle, obtaining the same separation between the different classes of dyes investigated as we obtained analyzing pure dyes in transmission mode. Moreover, the new enhancement method has been validated on a real sample dated back to the 1893, showing how it can be promising for the analysis of trace amounts of organic substances in artistic samples such as dyes in paintings or textiles, varnishes and organic residues on archaeological objects.

New Frontiers in Application of FTIR Microscopy for Characterization of Cultural Heritage Materials.

We present an overview of recent advances in the application of Fourier Transform Infrared (FTIR) microscopy for analysis of complex, multicomponent, and multilayer samples such as those typically encountered in the field of heritage materials. This technique is particularly useful since it allows identification and localization of both organic and inorganic (if IR active) compounds. New improvements have been possible thanks to the introduction of ad hoc sample preparation methods to obtain either thin or cross sections that allow both avoidance of contamination from organic embedding resin and improvement of the quality of the acquired spectra. Moreover, integrated use of spectra registered in the near-infrared (NIR) and mid-infrared (MIR) regions allows better comprehension of cross section composition. Data interpretation has been improved thanks to the development of chemometric methods for elaboration of hyperspectral data. A new and very promising field is the development of enhanced FTIR methods for detection of trace components in microextracts. These systems, allowing detection of extractable organic compounds from about 0.1 mg of sample, will be extremely useful in the future for analysis of natural and synthetic colorants, varnishes extracted, for instance, from cotton swabs used during cleaning of paintings, and organic residues on archeological remains.

Single and multiplexed immunoassays for the chemiluminescent imaging detection of animal glues in historical paint cross-sections.

The characterization of the organic components in a complex, multilayered paint structure is fundamental for studying painting techniques and for authentication and restoration purposes. Proteinaceous materials, such as animal glue, are of particular importance since they are widely used as binders, adhesives and for gilding. Even though proteins are usually detected by chromatographic and proteomic techniques, immunological methods represent an alternative powerful approach to protein analysis thanks to the high specificity of antigen-antibody reactions. Our previous studies demonstrated that ovalbumin and casein could be localized in paint cross-sections with high sensitivity and good spatial resolution (i.e. within the single painting layers) by using chemiluminescent (CL) immunochemical microscope imaging. In the present research work, we describe for the first time the immunolocalization of collagen (the main protein of animal glue) in paint cross-sections by CL imaging microscopy. Two different analytical protocols have been developed, allowing either the detection of collagen or the simultaneous detection of collagen and ovalbumin in the same paint sample. The assays were used to detect collagen and ovalbumin in cross-sections from model samples and historical paintings (a wall painting dated to 1773-1774 and a painted wood panel of the Renaissance period) in order to achieve information on paint techniques and past restoration interventions.

Development of innovative embedding procedures for the analyses of paint cross sections in ATR FITR microscopy.

We report the development of innovative embedding procedures for the analysis of paint cross sections by attenuated total reflection (ATR) Fourier transform IR microscopy. This technique was chosen because it is widely employed for the characterization and spatial location of organic and inorganic components in artistic samples. Moreover, the performance of the technique may be critically affected by sample preparation in terms of surface morphology and the presence of contamination. First, we evaluated the use of KBr as a barrier to contamination by the embedding synthetic medium. In this way, the sample cross section can be polished by means of a sample holder, which allows a controlled pressure to be applied to the sample, thus improving the reproducibility and quality of the surface cross section. In addition, argon ion milling was used for the polishing of samples embedded in KBr, and provided very promising results in terms of surface planarity and reduction of superficial contamination by KBr. Finally, the use of NaCl as an alternative to KBr was proposed thanks to its advantages in terms of hygroscopicity, cost, and toxicity. In addition, cross sections embedded in NaCl were characterized by greater hardness, a feature that allowed us to obtain improved contact with the ATR crystal.

A round robin exercise in archaeometry: analysis of a blind sample reproducing a seventeenth century pharmaceutical ointment.

Chemical analysis of ancient residues of pharmaceutical or cosmetic preparations such as balms or ointments is made problematic by the high complexity of these mixtures, composed of organic and inorganic materials. Consequently, a multi-analytical approach and special caution in the interpretation of the results are necessary. In order to contribute to the improvement of analytical strategies for the characterization of complex residues and to reconstruct ancient medical practices, a replica of a pharmaceutical formulation of the seventeenth century was prepared in the laboratory according to a historically documented recipe. In a round robin exercise, a portion of the preparation was analysed as a blind sample by 11 laboratories using various analytical techniques. These included spectroscopic, chromatographic and mass spectrometric methods. None of the laboratories was able to completely reconstruct the complex formulation, but each of them gave partial positive results. The round robin exercise has demonstrated that the application of a multi-analytical approach can permit a complete and reliable reconstruction of the composition. Finally, on the basis of the results, an analytical protocol for the study of residues of ancient medical and pharmaceutical preparations has been outlined.

New advances in the application of FTIR microscopy and spectroscopy for the characterization of artistic materials.

Fourier transform infrared (FTIR) spectroscopy is one of the most widely applied techniques for the investigation of cultural heritage materials. FTIR microscopy is well established as an essential tool in the microdestructive analysis of small samples, and the recent introduction of mapping and imaging equipment allows the collection of a large number of FTIR spectra on a surface, providing a distribution map of identified compounds. In this Account, we report recent advances in FTIR spectroscopy and microscopy in our research group. Our laboratory develops, tests, and refines new and less-studied IR spectroscopy and microscopy methods, with the goal of their adoption as routine analytical techniques in conservation laboratories. We discuss (i) the analysis of inorganic materials inactive in the mid-IR region by means of far-IR spectroscopy, (ii) the development of new methods for preparing cross sections, (iii) the characterization and spatial location of thin layers and small particles, and (iv) the evaluation of protective treatments. FTIR spectroscopy and microscopy have been mostly used in the mid-IR region of 4000-600 cm(-1). Some inorganic pigments, however, are inactive in this region, so other spectroscopic techniques have been applied, such as Raman spectroscopy. We suggest an alternative: harnessing the far-IR (600-50 cm(-1)). Our initial results show that far-IR spectroscopy is exceptionally useful with mural paintings or with corrosion products from which larger sample quantities can generally be collected. Moreover, the inorganic composition of a sample can be characterized by the presence of several compounds that are inactive in the mid-IR range (such as sulfides, oxides, and so forth). Stratigraphical analyses by FTIR microscopy can be hindered by the process of cross section preparation, which often involves an embedding organic polymer penetrating the sample's porous structure. Here, the polymer bands may completely cover the bands of organic compounds in the sample. However, a correct methodological approach can prevent such limitations. For example, it is always advisable to analyze the sample surface before preparing the cross section in order to characterize the preparation layers and the varnish layers, which are generally applied to the surface of a painting both to protect it and improve the color saturation. Furthermore, the innovative use of IR-transparent salts as embedding material for cross sections can prevent contamination of the embedding resin and improve detection of organic substances. Another key point in the use of FTIR microscopy in artwork analysis is spatial resolution. The high-energy output of a new integrated FTIR microscope enhances the ability to characterize and spatially locate small particles and thin layers. Moreover, the new configuration proves extremely useful in the evaluation of protective treatments, because larger areas may be analyzed in less time in comparison to traditional systems, allowing the collection of more statistical data.