Now comes the interesting part; the analysis of old paintings. Traditional x-ray images have long been valuable in examining and reconstructing paintings. They can establish the condition of the paint and reveal the techniques that were used by the artist in a painting. The downside of x-rays is that, in double-sided works of art or in cases where canvas has been reused by the artist, they produce images that are hard to interpret.
Artificial intelligence is showing promise in discovering details in old masters paintings and analysis of high-resolution x-ray images. It offers an opportunity to understand how works of art were created and an opportunity for investigators to ensure that such masterpieces are correctly preserved and presented. Artificial intelligence has tools that can decrypt technical and complex images. The algorithms can aid in identifying and diagnosing layers of a painting accurately, faster and with more precision.
The new algorithms allow researchers to understand Old Masters paintings by separating complex x-ray images. The capabilities of AI in studying various patterns in paintings demonstrates that AI-oriented techniques powered by deep learning can be used to solve different challenges in investigating works of art. Deep learning has the ability to reveal hidden features of a painting such as concealed designs.
An example where AI has been used to study images is the one carried out by researchers from the National Gallery, UCLA and Duke University where a complex 15th-century altarpiece (Van Eyck's Ghent Altarpiece) in Belgium’s St Bavo's Cathedral was analyzed. The study demonstrates how AI algorithms can be used to study mixed x-ray images that have features both in front and at the back of a painting. The double-sided panels can then be split into two clear images.
By separating double-sided complex images into two different images, the algorithm allows researchers to study and better understand ancient paintings. It shows that with proper algorithms in place, challenges that emerge when investigating pieces of art can be potentially solved. Further, the development of more sophisticated AI-oriented algorithms has an impact on revealing complex designs that were concealed and cannot be viewed using traditional x-rays.
Discovering how the paintings were designed eases the work because it allows those studying images to have insights into techniques used by an artist. This knowledge is a great way to enable restoration of cultural heritage and provides a wealth of data. Although traditional x-ray is a valuable tool that has been in use in examining and restoring images because they can help in identification of techniques used and other insights, the penetrating nature of these rays means that anything in its path will result in an image. This means that a resultant image may be difficult to interpret and can even be worse in panels that are painted on both sides or in instances where the canvas has been reused. This, however, is no longer the case with AI.
By using the new AI algorithms to separate parts of an image, historians and conservators can understand old paintings and information that is obtained can be instrumental in preserving history.
Last modified on Tuesday, 08 October 2019