An integrated OCR software for mathematical documents and its output with accessibility

Abstract

This paper describes shortly a practical integrated system for scientific documents including mathematical formulae, named 'Infty'. The system consists of three components of applications: an OCR system named 'InftyReader', an editor named 'InftyEditor' and converting tools into various formats. Those applications are linked each other via XML files. InftyReader recognizes scanned images of clearly printed mathematical documents and outputs the recognition results in a XML format. It recognizes complex mathematical formulae used in various research papers of mathematics including matrices. InftyEditor provides a very efficient interface to correct the recognition results using keyboard. Another feature of InftyEditor is its handwriting interface to input mathematical formulae for users with vision and speech interface for visually impaired uses. The XML files output by InftyReader/Editor can be converted into various formats: LATEX, MathML, HTML and Braille Codes; in UBC (Unified Braille Codes) for English texts and in Japanese Braille Codes for Japanese texts. © Springer-Verlag 2004.