Event Detail

Event Type: 
Wednesday, December 6, 2006 - 08:00
KEC (Kelley Engr Ctr) 1001

Speaker Info

Kyushu University (Fukuoka, Japan)

We are developing an OCR software for mathematical documents, called InftyReader. Beta copies of the software are downloadable from http://www.inftyproject.org. The recognition process is composed of several steps:

Pre processing (Noise reduction, etc.)
Layout analysis (Detection of graphics areas, table areas, etc.)
Separation of textual areas into math or text portions
Character/symbol recognition and Math structure analysis
Verification of character/symbol recognition based on structure analysis
Logical structure analysis of the document (Title, Headers, etc.)
Conversion of the results into some standard format for math documents(LaTeX, MathML, Braille, etc.)
In the lecture, I will report briefly the current state of the software in each step, and discuss related future problems from a practical point of view. This talk is jointly sponsored by CS, Math, and Physics