Prof. Ohad Ben-Shahar

Pairwise compatibility calculation is at the core of most archaeological reconstruction algorithms, in particular those designed to solve different types of jigsaw puzzle problems. However, most existing approaches fail, or aren’t designed to deal with fragments of realistic geometric properties one encounters in real-life puzzles, where irregular shapes or erosion are abundant. In practice, compatibility is typically needed, formulated, and assessed only for fragments with idealized, uniform, or strongly constrained shapes, limiting real-world applicability. In this paper, we propose an efficient approach for computing optimal alignment configurations for pairs of fragments and scoring their pictorial compatibility without any assumptions about their shapes, sizes, or visual content. We also propose evaluation metrics for pairwise compatibility, and introduce an extensive dataset containing pairs of fragments generated via a novel image fragmentation approach and a formal erosion model that mimics real-world archaeological erosion. We benchmark it with existing and new baselines, demonstrating the superior performance of our proposed approach.