Advanced Workflow for Extracting Characters from the Royal Woodblocks of the Nguyen Dynasty to Construct a Sino-Nom Dataset for Reconstructing Lost Woodblocks

Abstract

The woodblock printing technique, first developed in China, enabled the large-scale production of texts and significantly advanced the spread of knowledge and literacy across many Asian countries for centuries. In Vietnam, the royal woodblocks of the Nguyen Dynasty are considered a national treasure. However, many of these woodblocks have been lost or damaged over time, making it imperative to develop a method for reconstructing them. Therefore, this paper proposes a data processing workflow capable of constructing a Sino-Nom character dataset from existing woodblock collections. The constructed dataset can then be used for reconstructing the lost woodblocks and for further in-depth analysis. Using the 3D collection of the Dai Nam Thuc Luc chronicle as an example, we have created a large Sino-Nom character dataset named SiNoC through our proposed workflow. The SiNoC dataset comprises 90,259 pairs of 3D and 2D Sino-Nom characters. This dataset serves as a foundation for deep learning models and advanced image processing techniques aimed at reconstructing lost woodblocks.