Abstract
The article addresses the challenge of reconstructing 2D broken pictorial objects by automating the search for matching elements, which is particularly relevant in fields like archaeology and forensic science. The authors propose a method to match such elements and streamline the search process by detecting and filtering out low quality matches.
The study delves into optimizing the search process in terms of duration and assembly quality. It examines factors like comparison window length, Levenshtein measure margin, and number of variants to check, using theoretical calculations and experiments on synthetic elements. The experimental results demonstrate enhanced method effectiveness, yielding more useful solutions and significantly reducing the complexity of element comparisons by up to 100 times in extreme cases.