A Review of Arabic Document Analysis Methods

Abstract

Arabic document analysis is essential in geometrical information extraction from complex structures in Arabic documents, which can either be historical or modern. This information can be an organized tree structure containing all the component levels, such as column, paragraph, word, table, figure, and article. In this paper, we provide an analysis of recent works on this topic from various perspectives, describing the most commonly used models on document physical layout detection and document logical structure representations in printed styles, summarizing the limitations of previous approaches, identifying challenges along this line of research, and providing new research directions for future algorithms.