Up Close With Van Gogh
The Chronicle of Higher Education, 8.9.5
http://chronicle.com/weekly/v55/i02/02a00602.htm

By MARIA JOSI VIQAS

Beware, art forgers. Museums will soon be armed with a new tool to detect
your fake paintings: a sharp-eyed computer program.

An international team of experts in image processing is developing tools to
help art historians tell originals from counterfeits. The researchers
published their results in the July issue of IEEE Signal Processing
Magazine.

C. Richard Johnson Jr., a professor of electrical and computer engineering
at Cornell University, persuaded officials at the Van Gogh Museum in
Amsterdam to provide his digital-analysis team with 101 high-resolution
scans of the painter's works. To guard against digital forgery, the images
were transmitted in grayscale, a format that represents each color as a
shade of gray. The researchers divided the scans into sections and analyzed
such features as texture and brushstroke patterns.

"Painters can use very different kinds of brushwork styles, even for the
same painting, in a mingling fashion," says James Z. Wang, an associate
professor of information sciences and technology at Pennsylvania State
University who led a group of researchers at his institution. That
heterogeneity made it difficult for the computer to isolate individual
brushstrokes, says Mr. Wang.

Art experts had already unmistakably identified 23 of the works as Van Gogh
paintings, so Mr. Wang and his team developed algorithms to identify the
artist's brushstrokes. They built a statistical model of Van Gogh's style
that allows researchers to compare any painting to his original works. The
software, which is still being refined, will assess the likelihood that the
studied painting is a real Van Gogh.

Mr. Johnson says he hopes his project enhances collaborations among art
historians and image processors. "As engineers," he says, "we like to show
up and help people with their problems."