Image segmentation applied to the investigation of craquelure domains in paintings

Old paintings are generally characterized by the presence of cracks patterns, formed during the drying process. This network of cracks, known as craquelure, represents a key feature for the authentication of artworks and the discovery of forgery. In this work, an image segmentation approach is used for a quantitative study of cracks patterns in paintings. We will show how the size distribution and orientation of the craquelure domains can be evaluated using this technique.