W. Li and B. Li (USA)
Interaction design for people with disabilities, formula recognition, MathML
Current optical character recognition (OCR), speech synthesis, and video technologies have made it possible to build reading aid devices for blind and visually impaired readers. However, most existing systems can read only plain text from printed materials. For scientific books such as mathematical or engineering textbooks that are full of equations, formulae, tables, and graphical charts, a conventional OCR-based system will not be able to provide the desired assistance. This paper presents a completely automatic method for converting a scanned or captured mathematical expression into speech. In addressing the challenges in automatic symbol recognition, we present several new techniques that make the proposed system more robust. The symbol-to-speech conversion is facilitated by utilizing a MathML-compliant player.
Important Links:
Go Back
