A knowledge-based approach to Chinese archive document understanding

Abstract

The Chinese archive document possesses special geometrical and logical properties due to its construction based upon rectangular field which contain either title strings or data strings related to some other titles. In this paper, we propose a knowledge-based approach to analyze the logical relationship among the fields. After extracting the lines and fields of an archive document image, this procedure can identify fields as the title fields, the sub-title fields (if there exist such tree-structure logical relationship), and the corresponding data fields. This proposed approach enables us to achieve a better performance in information manipulation of archive documents.