CWRU researchers develop AI to identify counterfeit paintings
(TNS) – art forgers of the world, take note.
A team of art historians and scholars from Case Western Reserve University has developed a computer technique that can almost certainly determine which artist painted a particular painting with just bare tiny details of brushstrokes that the artist cannot control and cannot see Eye.
The method combines data from the precise, three-dimensional mapping of the surface of a painting with analysis by artificial intelligence – a computer system based on the human brain and nervous system that can learn to recognize and compare patterns.
The CWRU team, first reporting its findings in November in Heritage Science magazine, believes its work is breaking new ground and could be used in the future to identify counterfeits by detecting differences in tell-tale markings that are so small like the width of a brush bristle.
CWRU physics professor Kenneth Singer, who works on the project, said such traces were indicators of “an artist’s unintended style.”
“I wouldn’t say it’s foolproof; I am a scientist. But I’d say it’s a powerful tool, ”said Singer, Faculty Director of MORE, the University of Otpo / Electronics Research and Education Materials.
Michael Hinczewski, associate professor of physics at CWRU who is also on the research team, said in a press release that the new algorithm is so precise that it is “almost like a fingerprint”.
Elizabeth Bolman, chair of the art history division at CWRU, said the new method has the potential to greatly improve the attribution of works of art.
This is a critical point in the art market at a time when millions of dollars could depend on expert opinion about the authenticity of a given object.
“We’re at the point where we just figured out the basics of a concept, and our first attempt was spectacularly successful when we exceeded our wildest dreams,” said Bolman. “Where that can be done from here, we can all dream.”
In the experiment published in Heritage Science, Singer and Bolman and their colleagues were able to determine with greater than 95 percent accuracy which of four art students at the Cleveland Institute of Art used the same brushes to paint almost identical paintings of a yellow flower blossom, paints and canvas.
The analysts scanned the surfaces of the paintings and digitally divided them into grids of tiny squares from half a millimeter to a few centimeters wide. The randomized data was then examined by the machine learning software, which made comparisons and then identified the four artists with high accuracy.
The CWRU project is not the first to use artificial intelligence to analyze works of art. In 2017, Rutgers researchers published a study in which they collected data on more than 80,000 individual strokes in 300 drawings by Pablo Picasso, Henri Matisse and Egon Schiele and other artists and reliably identified forgeries.
However, the CWRU team said it was the first project to combine the three-dimensional surface topography of works of art with machine learning analysis.
In a new phase of work to be published, the CWRU team used the new technology to correctly identify which parts of a portrait of Cardinal Tavera from the early 17th century were being restored after the painting was cut into pieces during the Spanish Civil War.
Next, the team would like to compare two nearly identical versions of the Crucifixion of Christ by El Greco to see which parts were painted by the artist himself, which were painted by his son Jorge Manuel, and which were painted by members of the artist’s workshop or later by restorers treated.
One version is owned by the Cleveland Museum of Art and the other is owned by the Institute for Spanish and Hispanic Art in Bishop Auckland, England.
“The El Greco project is examining different scans of paintings to see if we can identify the workshop process and different hands,” said Bolman. “Did he work on them? How much did his son Jorge work on it? These are hotly contested topics. ”
In order to meet the high demand for their paintings, artists such as El Greco, Peter Paul Rubens and Rembrandt employed large workshops and sometimes made several versions of the same picture. Scientists have been embroiled in widespread debates about the attribution of such works for decades, as distinct from attempts by modern counterfeiters to deceive buyers by selling counterfeits.
The first results of the CWRU project seem to open up the possibility that computers could eliminate the need for connoisseurs, a branch of art history devoted to identifying who made what.
But Lauryn Smith, a Ph.D. Art history candidates at CWRU and digital art history fellows at Frick Museum in New York, who helped shape the experiment published in Heritage Science, said the use of artificial intelligence is a logical next step in connoisseurship history, no the end of it.
ROOTS OF ENCRYPTION
The field was made in the late 19th century. Feet or ears.
Morelli and scholars he trained, including the American art historian Bernard Berenson, used the method to sift through original paintings by Italian Old Masters from the works of assistants or lesser masters.
More recently, art historians trying to determine the authenticity of works of art have combined connoisseurship with scientific data based on the age or composition of pigments, canvas, wood, or other materials.
Smith said the new machine learning techniques take Morelli’s concept of “invariant” detail to a new, higher level of scientific specificity.
“There’s a bit of scare tactics in this process,” she said. “It really shows that by working with scholars, art historians and curators and all of those stakeholders, you can create phenomenally useful projects that can move the field forward.”
Smith said she came up with the idea for the project a few years ago with Michael McMaster, then a Ph.D. Candidate in physics.
As their relationship turned into a romance, Smith and McMaster decided to submit a paper to a conference on art and science where they proposed applying machine learning technology to analyzing the topography of the surface of a painting – the tiny ridges and bumps that arise when an artist applies paint to the canvas.
Their paper was not accepted for the conference, but colleagues in the CWRU’s art history and physics departments were intrigued and encouraged the couple to continue the project that continued after Smith and McMaster’s wedding in 2019.
“It’s our story from lab rats to lovers,” said Smith.
McMaster came up with the idea of using a chromatic confocal profilometer, a widely used scanning device, to analyze the CIA students’ paintings.
Now the team is increasing its analytical skills. Bolman arranged for the non-profit Factum Foundation for Digital Technology in Conservation, based in Madrid, Spain, to bring their proprietary Lucida 3D scanner to the Cleveland Museum of Art in November to scan El Greco’s crucifixion. The technology captures data that is smaller than a micrometer, Bolman said.
A similar process will soon be underway at El Greco in England, she said.
The Factum Foundation said in a statement on its website that the implications of the research conducted to date by the CWRU team are “far-reaching and groundbreaking. Connoisseurs will soon have a new tool that will help with tricky attribution questions for many paintings. “
Smith said there was a lot of interest when she and McMaster lectured at academic conferences. They had inquiries as to whether the technology could be applied to the study of coins, textiles, sculptures, painted musical instruments, and other objects.
“We’re way ahead of what everyone else is doing,” said Bolman. “This is a completely different way of approaching the visual material culture.”
© 2021 Advance Local Media LLC. Distributed by Tribune Content Agency, LLC.