Cybersleuth paints forgers into corner

By STEVE CONNOR

Scientists have created a computer that can tell the mathematical difference between a genuine work of art and a forgery by analysing features invisible to the human eye, paving the way to a new method of art fraud detection.

The computer can also distinguish between the contributions of apprentices who collaborated on a well-known masterpiece officially attributed to a single artist.

Scientists said the technique meant academics could better understand the hidden contributions of lesser-known artists.

The researchers used a mathematical approach, analysing the statistical likelihood that a particular brush or pen stroke was performed by the artist.

A similar mathematical approach powered by computer has been used to analyse the words in famous texts to see whether they were the sole creation of a well-known author.

Henry Farid, associate professor of computer science at Dartmouth College in Hanover, New Hampshire, said the technique could be applied to fine art thanks to the widespread use of high-resolution digital imagery that collects up to 2400 points of light in a single square inch of canvas.

"We have been able to mathematically capture subtle characteristics of an artist that are not necessarily visible to the human eye," he said.

"We expect this technique, in collaboration with existing physical authentication, to play an important role in the field of art authentication.

"Similar methods have been used to analyse works of literature.

"We can find things in art work that are unique to the artist, such as the subtle choice of words or phrasing and cadence that are characteristic of a certain writer."

Scientists program the computer with an artist's personal style of painting or drawing, using digital images of masterpieces known to be the work of the same painter. The machine can then decide whether a new work it subsequently analyses is likely to be a forgery.

Professor Farid and his colleagues analysed 13 drawings that had been attributed - at least at some time - to Flemish artist Pieter Bruegel the Elder. The computer successfully distinguished between eight paintings known to be by the artist and five famous imitations by contemporary artists, including some by artists who intended to commit a forgery.

A second part of the research, reported in the journal Proceedings of the National Academy of Sciences, analysed digital images of the Madonna with Child by Italian Renaissance artist Perugino, who became famous for his altar pieces.

The faces of three figures on the left of the painting were found to be the work of one artist, and the three on the right were different enough to be the work of different artists, probably Perugino's apprentices, a common practice in Renaissance art.

The researchers say analysing brush and pen strokes mathematically will be combined with other techniques, such as x-rays to see underneath a coat of paint, in verifying a painting's authenticity.

- INDEPENDENT