New concept for art and antiquities identification based on craquelure pattern analysis

Abstract

In this paper, it is shown that the craquelure patterns of paintings and other antiquities can be considered as fingerprints so that the authenticity of any given artwork can then be verified against prior works. The authors propose to extract craquelure features from photographic images and use them in the implementation of a unique matching procedure to address the art forgery issue as well as to monitor the health conditions of the objects concerned. This feature extraction strategy is robust against illumination, scale, rotation and perspective distortion. The craquelure extraction system developed, called multi-scale multi-orientation morphological processing (MMMP), performs analyses in each sub-band. A comprehensive craquelure image dataset has been constructed from a variety of different types of painting and other art objects. The results show significant improvement and compare favourably with the current best results in the market.