Allure of Craquelure: A Variational-Generative Approach to Crack Detection in Paintings

TL;DR

This paper introduces a hybrid variational-generative method for detecting cracks in paintings, combining deep generative models and inverse problem techniques to improve accuracy in complex visual scenarios.

Contribution

It presents a novel inverse problem framework that integrates deep generative priors with variational modeling for precise crack detection in artworks.

Findings

Effective crack localization in complex paintings

Improved differentiation between cracks and artistic features

Robustness to challenging visual conditions

Abstract

Recent advances in imaging technologies, deep learning and numerical performance have enabled non-invasive detailed analysis of artworks, supporting their documentation and conservation. In particular, automated detection of craquelure in digitized paintings is crucial for assessing degradation and guiding restoration, yet remains challenging due to the possibly complex scenery and the visual similarity between cracks and crack-like artistic features such as brush strokes or hair. We propose a hybrid approach that models crack detection as an inverse problem, decomposing an observed image into a crack-free painting and a crack component. A deep generative model is employed as powerful prior for the underlying artwork, while crack structures are captured using a Mumford--Shah-type variational functional together with a crack prior. Joint optimization yields a pixel-level map of crack localizations in the painting.