Art and conservation experts regularly rely on X-ray imaging to find out what’s going on underneath the surface of a painting. It allows them to learn more about the paint used, the technique, or the painted surface. Now, researchers have found a way to improve this technology with artificial intelligence.
Restorers of the Royal Institute for Cultural Heritage working on the Ghent Altarpiece. Some of the X-ray images of this work have now been further analysed using artificial intelligence. (Credit: JOHN THYS/AFP/Getty Images)
AFP/Getty Images
Even though X-ray imaging is an effective way to study the hidden layers of a painting, it doesn’t discriminate between the actual layer of interest and everything else. The X-rays go right through the entire object. That means that the resulting image might include details that aren’t always relevant. For example, it can show the wooden structure of a canvas frame, which can make it more difficult to look at details of the painting itself in the X-ray image.
This gets particularly challenging if the painting is double-sided. An X-ray image of either side of it will show details of both painted sides, merged together. To distinguish the different images in such a merged X-ray picture, engineer and art experts have now turned to artificial intelligence.
Researchers from University College London and Duke University in Durham, North Carolina, used the visible images from either side of a painted panel to train a neural network to recognize individual images in the mixed X-ray image.
With this technique, they have successfully managed to separate images from a merged X-ray image of panels from the 15th-century Ghent Altarpiece at the St Bavo’s Cathedral in Belgium. The original artists, the brothers Van Eyck, painted these panels on both sides to show different images depending on whether certain sections of the piece were open or closed.
The two double-sided panels from the Ghent Altarpiece along with their x-ray images which include the combined contributions from both sides of each panel.
Saint-Bavo's Cathedral, www.lukasweb.be - Art in Flanders; photos: Hugo Maertens (interior view; Adam & Eve), Dominique Provost (exterior view), KIK-IRPA (x-ray)
The tidied X-ray images have been included as part of a larger conservation project of the altarpiece that started in 2012.
This AI method of tidying up X-ray images isn’t limited to two-sided paintings. It can also be useful in cases where an artist has reused a canvas by painting over an older work, which is quite common. In that case, the training set for the neural network would be different, as the underlying image is not known, but other than that it would use the same principles.
According to Hélène Dubois, Head of the Conservation Project of the Ghent Altarpiece at the Royal Institute for Cultural Heritage (KIK-IRPA), “This new development of the use of the traditional x-ray has great potential for countless applications in conservation of irreplaceable works of art.”
Join us on Facebook or Twitter for a regular update.
#News | https://sciencespies.com/news/both-sides-now-looking-at-two-sided-paintings-with-x-rays-and-ai/
No comments:
Post a Comment