Priority levels and heuristic rules in the structural recognition of mathematical formulae

In the paper an algorithm for an automatic recognition of the structure of mathematical formulae saved in a graphical form has been presented. The described method is based on 2D graph grammars, although the other approaches were mentioned. Moreover, the heuristic rules used during the simplification of the initial data structure, the example graph grammar rules, the priority system used within the production of the rules, and the test application that allows the step-by-step analysis of an entire recognition process, were discussed. The solution introduced in the paper allows to analyse web pages and the other documents containing a math notation saved in graphics formats. It will also help in the development of systems indexing, searching and translating such documents to formats used by visually impaired people.

[1]  Masakazu Suzuki,et al.  Mathematical formula recognition using virtual link network , 2001, Proceedings of Sixth International Conference on Document Analysis and Recognition.

[2]  Z. Sroczyński.  Prezentacja dokumentów internetowych zawierających złożoną notację matematyczną dla potrzeb osób niewidomych , 2003 .

[3]  Z. Sroczyński,et al.  Telepraca i zdalna edukacja osób niepełnosprawnych wzrokowo , 2006 .

[4]  Loïc Pottier,et al.  On-line handwritten formula recognition using hidden Markov models and context dependent graph grammars , 1999, Proceedings of the Fifth International Conference on Document Analysis and Recognition. ICDAR '99 (Cat. No.PR00318).

[5]  Stéphane Lavirotte,et al.  Mathematical formula recognition using graph grammar , 1998, Electronic Imaging.

[6]  James Arvo,et al.  Equation entry and editing via handwriting and gesture recognition , 2001, Behav. Inf. Technol..

[7]  Paul A. Viola,et al.  Ambiguity and Constraint in Mathematical Expression Recognition , 1998, AAAI/IAAI.

[8]  Stephane Lavirotte.  Optical formula recognition , 1997, Proceedings of the Fourth International Conference on Document Analysis and Recognition.