A modular framework for the automatic reconstruction of shredded documents
Razvan Ranca · National Conference on Artificial Intelligence · 2013
The unshredding problem is of interest in various domains such as forensics, investigative sciences and archeology, and has therefore been approached in many different ways. This paper tries to bridge the gap between previous, disparate, efforts by proposing a modular, probabilistic, solution. Novel approaches to two of the independent subproblems are presented and shown to both have good theoretical properties and to empirically outperform previously published methods.