When we visit a museum, we don’t question whether the paintings truly are original pieces by the authors in the tags or not. How do experts really know when a piece is original or a well-done copy?
Traditionally, in order to decide if a painting is original or a copy we needed art experts. Normally, we would find experts in the most famous artists who had learnt, from experience, to distinguish original pieces from a certain author. Then, the Van Gogh Museum would have Van Gogh experts who would recognize this author’s work, and so on. As you can imagine, this work was extremely hard, and informatics has tried to help out.
In 2008, an experiment took place: the Van Gogh Museum asked a local artist to paint a copy of one of Van Gogh’s pieces. Then, they took a picture of the copy, mixed it up with 5 pictures or other original pieces, and handed it over to three teams from three different universities, to see if they could discover which one was the copy. All three teams used computer programs that analyze brush strokes. It seems that it is possible to identify a piece that is different from several others easily, as a copy is usually made with more strokes than an original, and these are slower and less natural. If we analyze the strength and direction of strokes in several paintings, we can find a numerical pattern and identify a painting that does not follow the same pattern as the others.
Like this, informatics helped all three teams distinguish the fake painting from the original pieces.
You can find the whole report here