Javastylosis

Abstract

To solve some critical issues raised by anastylosis techniques, such as the deterioration of the original fragments due to continuous handling, and to compare alternative matching solutions for choosing the best one for real restoration, the pattern recognition and computer vision communities are proposing approaches that work on virtual copies of the fragments, providing new methodologies that help in the matching task. In this paper, we present Javastylosis, a tool that can support restorers for reassembling 2D images of damaged frescoes. With this computer-assisted methodology the user can apply chromatic and semantics principles. Each virtual fragments can be moved and rotated, and the same operations can be applied on groups of fragments. During the matching task, spurious elements can be discarded, and the final result is a digital solution of the 'puzzles' made by the artwork fragments, providing a table of coordinates useful to the restorers to correctly reposition the real fragments. The graphical interface, designed to be easy and intuitive, has been studied for and tested by people belonging to the autism spectrum but with very acute visual skills, so that inclusion opportunities can be given to this group of disadvantaged people.