Art restoration using AI

The Van Eyck brothers’ Ghent Altarpiece (1432) is considered as one of history's most celebrated paintings. Almost six centuries later, a team of researchers is now unveiling some of ‘the artwork's best-kept secrets with the help of artificial intelligence (AI)’. The team successfully deconstructed mixed X-ray images of the masterpiece, revealing new details about the two-sided Adam and Eve panels of this polyptych painting.  The Ghent Altarpiece is comprised of 12 highly detailed painted panels, of which eight are painted on both sides of the three-piece folding construction. Depictions of Adam and Eve bookend the altarpiece’s open view; with these two double-sided panels being the subject of this study’s examinations.

In their study, the researchers were required to separate these mixed X-ray images of the double-sided panels into reconstructed images of ‘one-sided’ paintings. In order to successfully achieve this, they had to devise ‘a convolutional neural network (CNN)-based “self-supervised” scheme’ for the task. Specifically, they trained the deep neural network to convert a ‘Red, Green and Blue’ (RGB) visual image onto an X-ray image by leveraging the following sources: the visible RGB image of the front panel, the visible RBG image of the back panel and the mixed x-ray image. Then, they fit a CNN model to take the visual RGB images as input and produce two separated X-ray images as output. The neural network effectively ‘learned’ the algorithm by minimising the differences between the sum of the reconstructed X-ray images (the output) and the original mixed X-ray image. The team suggests that applying AI techniques such as this to art history may open new doors for investigating, conserving and understanding works of art in an innovative way.

Students planning to apply for History of Art can reflect on how this type of AI-assisted analysis may potentially prove useful for countless art applications, including material identification, brush-stroke technique analysis and perhaps even visualisation of previously unidentified, hidden designs and underdrawings.