A Vision Based System for Guided and Collaborative Reconstruction of Fragmented Documents
This paper presents the development and evaluation of a collaborative system for real-time reconstruction of fragmented paper documents in the context of cultural heritage preservation. The developed system includes a collaborative robot, or cobot, that can fully manage the positioning of paper fragments using a specially designed vacuum-based suction attachment. This attachment enables gentle and precise positioning, ensuring the preservation of fragile materials. With this device, we are able to achieve a positioning repeatability of 0.57mm for fragments of 8cm^2. The system offers users the flexibility to choose between manual positioning, with visual guidance, or fully automated positioning performed by the cobot. To further improve the reconstruction process, AI methods for image interpretation, specifically for segmentation and positioning tasks, were applied and evaluated for their applicability to template-based reconstruction of damaged paper fragments. Our investigation provides critical insights into the performance of different local feature matching methods under different document types, taking into account rotation, scale robustness, and the degree of damage to the fragments. With a focus on the reconstruction of damaged and optically altered archival material, SE2-LoFTR, a detector-free local feature matching method, was chosen as the preferred method for the system due to its robust performance in our experiments.