ISRAEL: Israeli academic says algorithm may identify authenticity of artists' paintings

Record ID: 395427

  • Title: ISRAEL: Israeli academic says algorithm may identify authenticity of artists' paintings
  • Date: 10th December 2007
  • Summary: (L!3) TEL AVIV, ISRAEL (RECENT) (REUTERS) WIDE OF PEOPLE VISITING TEL AVIV MUSEUM OF ART VISITORS OBSERVING PAINTING MORE OF VISITORS LOOKING AND POINTING AT PAINTINGS
  • Embargoed: 25th December 2007 12:00
  • Keywords:
  • Location: Israel
  • Country: Israel
  • Topics: Arts / Culture / Entertainment / Showbiz,Light / Amusing / Unusual / Quirky
  • Reuters ID: LVADP5ZHE5SKFMBQ1IAZW410ELL4
  • Story Text: A newly formulated algorithm in computer vision may take artificial intelligence one step further and help art collectors identify the authenticity of a painting.

    "I was curious to see whether a computer can do what humans can do," said the developer, Professor Daniel Keren of Haifa University, in northern Israel. "I was very curious to see whether we can formalize this notion of style and use it to identify painters."
    Professor Keren says he has developed a computer formula that can recognise an artist's distinct style enabling it to assess whether a specific painting was painted by a particular artist.
    He contends that this type of algorithm can be further developed to determine whether or not a specific painting is authentic.
    The mathematical method developed by Professor Keren identifies micro-structures in images, allowing the computer to determine if the painting in question corresponds with the artist's style.
    Once the algorithm familiarizes itself with a painting by Van Gogh in mathematical terms, it can then recognize Van Gogh's unique style in any other painting.
    But Keren says at present the algorithm can only determine the probability of a painting being authentic.

    "If the algorithm tells you that the probability for a certain painting to be a Van Gogh is two percent, then don't touch this painting. If it tells you that it's sixty percent, then you are risking your money. If it says that it's ninety five percent then go ahead and buy the painting,"
    said Professor Keren.
    The development marks a step forward in the field of computer vision and may pave the way to helping art experts and historians in detecting a forgery, according to Professor Keren.

    "I believe that there's a definite possibility that this will serve to detect forgeries".
    But Doctor Doron Lurie, Curator for Old Masters and the Chief Conservator for the Tel Aviv Museum of Art says though the idea is feasible, it would take at least another few decades to develop a formula that answers a wide range of questions to determine a painting's origin.

    "I think it's too naive and too premature," he said. Lurie said he believed that maybe in fifty to one hundred years, somebody could come up with such a formula, "but for the time being I am rather sceptic," he added.
  • Copyright Holder: REUTERS
  • Copyright Notice: (c) Copyright Thomson Reuters 2011. Open For Restrictions - http://about.reuters.com/fulllegal.asp
  • Usage Terms/Restrictions: None
Register