On Friday December 16th, I had the pleasure of assisting with the hyperspectral imaging of three paintings in the Cantor Arts Center’s collection, including Andre Derain’s Still Life With Fruit (1938), Dirck van Delen’s Solomon Receiving the Queen of Sheba (1642), and Hieronymus Bosch’s The Last Judgment (1500s). Joyce Farrell and Torbjorn Skauli of the Stanford Center for Image Systems Engineering led the advanced imaging procedure, which employs a high-resolution MCT (mercury cadmium telluride) detector that produces digital, high-resolution true-color renderings, performs near-infra-red scans to uncover any underdrawings, and provides information about the pigmentation and other materials in the medium. Both the Louvre and the National Gallery have also been using this system to study and photograph their collections. (See http://www.hyspex.no/pdfs/HySpex_Art_scanner_web.pdf.)
What is hyperspectral imaging? First developed for remote sensing applications, hyprspectral imaging collects information about a scene or an object from across the electromagnetic spectrum. It is is a type of spectroscopy that collects information as a set of images. Each image represents a range of the electromagnetic spectrum, known as a spectral band. Hyperspectral imaging deals with taking snapshots using narrow spectral bands over a continuous spectral range to produce a visual rendering of the spectra of all pixels in the scene. To examine art, the hyperspectral imager gleans optical information over visible (400-700nm) and near-infrared wavelength ranges (700nm to > 1micrometer), as some oils and pigments don’t fluoresce in the visible light range. Because certain objects and materials leave unique fingerprints across the electromagnetic spectrum, the imaging technique has been applied in a multitude of ways in the fields of geology, agriculture, resource management, and medicine – to image faces, large fields, forested areas, and even pig organs.
The processing of these high-resolution, spatially-resolved data sets is rather complicated. In fact, Brian Wandell and Joyce Farrell are teaching a class next quarter (http://white.stanford.edu/~brian/psy221/syllabus.html), and they hope some students from the class will be interested in analyzing the images.