August 18, 2008

Computers Help Detect Art Copies, Forgeries

By Tom Avril The Philadelphia Inquirer

PHILADELPHIA - The painting of the lean-faced, bearded man with the penetrating stare is unmistakably a self-portrait by Vincent van Gogh.

An art historian can tell by looking at the riot of bold, colorful brushstrokes.

Researchers at Pennsylvania State and Princeton Universities, however, use an analytical tool that surely the troubled Dutch master never imagined: the computer.

Their method is far from foolproof, but the two teams, along with a third one in the Netherlands, were able to distinguish dozens of van Gogh's works from those painted by others - including an infamous forgery.

A picture, after all, is more than a thousand words. It can be represented as bits of data, just like a bank account or music on a compact disc, and the researchers have sifted this information through the dispassionate filter of statistics.

The authors, who described their results in a recent issue of the engineering journal IEEE Signal Processing, are quick to say that they don't want to replace art historians. Their methods aren't sophisticated enough to do so even if they wanted to.

"Sometimes, a computer is pretty smart," says Penn State's James Z. Wang, one of the authors. "Other times, it may not be."

Yet he and his colleagues predict the computer will become an important tool alongside other scientific techniques that have long been used in art scholarship, such as chemical analysis of paint fragments or the use of X-rays to count threads in a canvas.

They've already won converts at Amsterdam's Van Gogh Museum, which has the world's largest collection of the artist's work.

"It was much more successful than I would have expected," says Ella Hendriks, the museum's head of conservation.

But before it could happen, there was a big question. How do you get a bunch of engineers and statisticians to communicate with people in the subjective realm of art?

Answer: Start with someone who is a member of both worlds.

C. Richard Johnson never went to an art museum as a child, and he pursued an early interest in the sciences by attending Georgia Tech.

But once there, he did a study-abroad program in Germany that he calls a "life-changing experience."

He spent hours at a museum in Berlin, becoming captivated by the works of Rembrandt. Later at Stanford University, he earned a Ph.D. in electrical engineering but also found time for his newfound love, with a minor in art history.

Yet it was not until 2005, during a sabbatical from his job as a Cornell University engineering professor, that Johnson looked through the literature for ideas on how he could marry his two talents.

He discovered the work of Penn State's Wang and his wife, Jia Li, who were performing statistical analysis of Chinese paintings. At Princeton, math professor Ingrid Daubechies was pioneering the use of statistics to analyze images from various fields of science and medicine, such as MRIs. And at Maastricht University in the Netherlands, computer scientist Eric Postma had started to analyze the works of van Gogh.

Museums around the world had begun to digitize their collections to aid in conservation and research, but the notion of crunching those reams of data was in its infancy.

So Johnson approached the Van Gogh Museum and offered to organize a conference. In exchange for the use of high-resolution scans from dozens of paintings, the three university teams - Penn State, Princeton and Maastricht - would present their research at the event in Amsterdam.

Like most people, the museum officials were unfamiliar with the statistical techniques involved, but Johnson sold the deal.

"He can talk between the two sides," Wang says. "He is serving as a bridge."

(c) 2008 Telegraph - Herald (Dubuque). Provided by ProQuest LLC. All rights Reserved.