The Art of Identification
A new technique catalogues art based on a detailed view of its texture and color.
Scientists at the Netherlands Institute for Cultural Heritage, a research organization for art conservation, have developed a unique way of identifying works of art. Called FingArtPrint, the project is the result of a collaboration among European universities, museums, and technology companies, and it was funded by the European Union.
FingArtPrint could help make museums and collectors feel secure about loaning their works of art, which can be traced via their “fingerprints.” But the project’s biggest impact is expected to be in slowing the market in stolen art and artifacts, which is the second most lucrative underground business in the world and worth an estimated $6 billion a year, according to the Federal Bureau of Investigation.
“If people are in the business of trafficking, and they know that institutions are starting to identify objects better, then hopefully they’ll have a much more difficult time getting rid of their inventory,” says Sharon Little, a Canadian textile conservator and the coordinator of a working group of the International Council of Museums Committee for Conservation (ICOM-CC) that explores ways to prevent trafficking. A related use of FingArtPrint might be to identify objects in archaeological sites that can’t be excavated or cataloged because of lack of funding or dangerous working conditions. Once those objects show up on the market, curators and collectors will know that they’ve been looted.
FingArtPrint works by creating a unique fingerprint of the surface of paintings, sculptures, coins, books, tapestries, and ceramics. First the researcher selects a predetermined part of the object to inspect measuring one square centimeter. The color of each pixel in the area is mapped with a scientific-grade digital camera that is typically used in industry to ensure the crispness of liquid crystal instrument displays or to examine flat panels for defects.
The roughness of the area is also measured, using a powerful microscope known as a white-light confocal profilometer. Typically, this tool is used in industrial applications to scan surfaces for damage or erosion. The microscope scans micron by micron; at such a high resolution, it effectively turns what appears two-dimensional to the unaided human eye into a three-dimensional landscape. “At that scale,” says William Wei, the director of the FingArtPrint project, “such information is unique to that object.”
Together, the color and roughness data make a fingerprint that can be stored in a computer database. A customs official, museum curator, or potential buyer who questions an item’s authenticity could capture a fingerprint and try to match it with one in the database.
But can FingArtPrint work on all the objects that a museum might collect–or that a smuggler might sneak through customs? Last fall, Wei and his team arranged a set of case studies for 30 objects from various museums and galleries, including an Egyptian stone medallion, a Roman glass bottle, books, an oil painting, several cast-bronze sculptures, and a wooden mask. Most of the objects lent themselves to fingerprinting with this technique. Brush strokes, the crackling of varnish, a layer of printed ink, the pattern of rust, and other irregularities in the surface of wood and ceramic–the profilometer generated unique patterns for all. Even two bronze sculptures taken from the same mold generated two distinct roughness profiles.
One potential limitation of the fingerprinting technique is how natural rates of change might alter the fingerprint itself. “What happens as a painting ages, and the microstructure changes: will changes due to aging then look like a forgery?” asks Hany Farid, a computer scientist at Dartmouth University who specializes in imaging.
The researchers own evaluation found that the process doesn’t work well with textiles and objects that decay rapidly. But it could still be useful for a wide variety of other items.
Currently, owners try to keep track of fine works of art by photographing the entire object or by marking it or attaching a tag or sticker to it. But photographs can be forged, engravings alter the surface, and adhesives and inks might disrupt the delicate chemistry of old objects. “Whenever you have an intervention to an object–say, to put a serial number on it–you want it to be reversible,” says Little. “But it’s complicated to make something reversible and yet permanent at the same time.”
Research into fingerprint durability is ongoing, says FingArtPrint’s Wei. He adds that fragments of fingerprints can be as valuable in art identification as they are in detective work. Other changes, such as the yellowing of a varnish, could be calculated into the fingerprint information.
Wei says the first FingArtPrint prototype should be completed this year.
AI is here. Will you lead or follow?
Join us at EmTech Digital 2019.