Research Article
Measuring Contour Similarity Based on Improved Hausdoff Distance for the Automatic Conjugation of Irregular Fragments of Ancient Manuscripts
@INPROCEEDINGS{10.1007/978-3-031-71716-1_15, author={Zixuan Yin and Yutong Zheng}, title={Measuring Contour Similarity Based on Improved Hausdoff Distance for the Automatic Conjugation of Irregular Fragments of Ancient Manuscripts}, proceedings={Machine Learning and Intelligent Communication. 8th EAI International Conference, MLICOM 2023, Beijing, China, December 17, 2023, Proceedings}, proceedings_a={MLICOM}, year={2024}, month={9}, keywords={Ancient manuscript fragments Dunhuang manuscripts Contour similarity Automatic conjugation model}, doi={10.1007/978-3-031-71716-1_15} }- Zixuan Yin
Yutong Zheng
Year: 2024
Measuring Contour Similarity Based on Improved Hausdoff Distance for the Automatic Conjugation of Irregular Fragments of Ancient Manuscripts
MLICOM
Springer
DOI: 10.1007/978-3-031-71716-1_15
Abstract
Ancient manuscripts have precious value in historical, cultural, social, and artistic. It is the common wealth of humanity. Suffering the long-term erosion and wears, ancient manuscripts were damaged and even completely separated into fragments scattered throughout the world. Fragment conjugation is an important part of the research and protection of ancient literature, However, the most advanced technology has not been adopted. Because we don’t yet know how to make the entire work automated, and the design and use of similarity assessment indicators, which play a crucial role in automation. In this paper, we propose a contour similarity evaluation based on improved Hausdoff distance, and a novel model which can make entire conjugation fully automated. We took Dunhuang ancient manuscripts as an example, verified on fragmented datasets and compared with expert results. The results indicate that our method can work automated, and the success rate of conjugation has been significantly improved compared to existing methods for similar tasks; In the analysis of contour similarity evaluation, our method based on improved Hausdoff distance can effectively avoid missed matching and improve the success rate of matching.
