Table recognition in mathematical documents

[摘要] While a number of techniques have been developed for table recognition in ordinary text documents, whendealing with tables in mathematical documents these techniques are often ineffective as tables containingmathematical structures can differ quite significantly from ordinary text tables. In fact, it is even difficult to clearly distinguish table recognition in mathematics from layout analysis of mathematical formulas. Again, it is not straight forward to adapt general layout analysis techniques for mathematical formulas. However, a reliable understanding of formula layout is often a necessary prerequisite to further semantic interpretation of the represented formulae. In this thesis, we present the necessary preprocessing steps towards a table recognition technique thatspecialises on tables in mathematical documents. It is based on our novel robust line recognition technique for mathematical expressions, which is fully independent of understanding the content or specialist fonts ofexpressions. We also present a graph representation for complex mathematical table structures. A set of rewriting rulesapplied to the graph allows for reliable re-composition of cells in order to identify several valid tableinterpretations. We demonstrate the effectiveness of our technique by applying them to a set of mathematical tables from standard text book that has been manually ground-truthed.

[发布机构] University:University of Birmingham;Department:School of Computer Science

[关键词] Q Science;QA Mathematics