Thick 2D relations for document understanding

摘要: We use a propositional language of qualitative rectangle relations to detect the reading order from document images. To this end, we define notion encoding rule and analyze possible formalisms express rules such as LaTeX SGML. Document expressed in rectangles are used build detector for In achieve robusmess avoid brittleness when applying system real life images, thick boundary interpretation relation is introduced. The framework tested on collection heterogeneous images showing recall rates up 89%.