JAIC special issue on “Reflectance hyperspectral imaging to support documentation and conservation of 2D artworks”

Abstract

This is the first time the Journal of American Institute for Conservation dedicates a special issue to the emerging topic of “Reflectance hyperspectral imaging to support documentation and conservation of 2D artworks.” As the organizers, we chose to focus on hyperspectral instead of multispectral imaging to emphasize the added information that is attainable by collecting enough reflectance images in narrow spectral bands to allow reconstructing the diffuse reflectance spectrum at each point in the scene. Multispectral imaging (MSI) in general is the collection of reflectance images in a few to tens of spectral bands that need not be contiguous. MSI studies related to use on works of art, including paintings and works on paper, were first published in the 1990s (Saunders and Cupitt 1993). These early studies demonstrated the potential of spectral imaging to spatially separate pigments (Casini et al. 1999), improve upon broadband infrared reflectography (Walmsley et al. 1994), and reveal lost text in palimpsests (Knox et al. 2001). Hyperspectral imaging (HSI) by definition is the acquisition of reflectance images in hundreds of narrow spectral bands to allow the reconstruction of the diffuse reflectance spectra at each pixel in the scene. The reflectance spectra can be used to find and map artists’materials including pigments and paint binders. The earliest reflectance HSI papers in the field of art conservation occurred at the beginning of the twenty-first century (Casini et al. 2005) and the first demonstration of mapping pigments in an entire artwork in 2010 (Delaney et al. 2010). Interestingly, the technique also has been found to provide an improved understanding of paint changes in works of art. With continuous improvements in imaging technology, the line between MSI and HSI is becoming blurred. Recently, the AVIRIS research groups that helped coin the term “hyperspectral imaging” (https://aviris.jpl.nasa. gov/) have urged a change in nomenclature from “hyperspectral” to “imaging spectroscopy” and “imaging spectrometer data.” It is interesting to note the adoption of imaging spectroscopy occurred early in the cultural heritage field (Casini et al. 1999), when a multispectral imaging system based on a Vidicon camera was used to identify and map pigments in paintings. With rapid development and adoption of macroscale scanning x-ray fluorescence, one can expect an increase in the different types or modalities of imaging spectroscopy utilized for the study of cultural heritage objects. In fact, this can be seen in the papers in this special edition. The initial goal of this special issue was to focus on the use of HSI reflectance cameras in the analysis of artworks, mainly paintings and works on paper. The outcome is somewhat broader as the scope of articles cover the methodologies for the operation of HSI reflectance cameras, the investigation of new image processing algorithms to extract the most meaningful information from the large data sets, as well as examples on how the HSI reflectance results can provide insight into the working methods of artists. The first two articles address methodology. “Hyperspectral imaging of art: Acquisition and calibration workflows” presents procedures to calibrate hyperspectral imaging data through a defined set of data acquisition and calibration workflows for hyperspectral cameras. Next, “Short-wave infrared reflectance hyperspectral imaging for painting investigations: A methodological study” illustrates how different spatial and spectral sampling can affect the documentation of underdrawings and the identification of artists’ materials when data in the 950–1650 nm range are considered. The following article, “A John White Alexander painting: A comparison of imaging technologies for resolving a painting under another painting,” compares the information obtained using low cost versus high cost imaging instrumentation. The authors also offer ways to make these technologies robust for non-experts, as well as provide practical tips for undergraduate and graduate education. The fourth article continues on this theme of reused canvas and looks at the added utility of the spectral information to obtain clear images that relate to the working process of an individual artist. “Rembrandt’s An Old Man in Military Costume: Combining hyperspectral and MA-XRF imaging to understand how two paintings were painted on a single panel” expands beyond providing a clear image of the portrait beneath the Old Man in Military Costume to describe how Rembrandt painted over this earlier composition before beginning the painting we see today. The paper also reveals portions of earlier drawing associated with the first painting. The proceeding two articles explore how to identify and map pigments and their distribution in paintings starting from HSI data. In the article titled “A Hyperspectral Imaging Spectral Unmixing and Classification Approach to Pigment Mapping in the Gough & Selden Maps,” the authors present a novel pigment-mapping tool for