Research on a Practical Model of a computerized Aloud-Reading System for Materials Including Mathematical Expressions

本研究由两部分组成：1.包含数学表达式的印刷品光学字符识别系统的开发；2.OCR结果计算机朗读系统的开发。通过研究，我们分别取得了以下成果。研究了从扫描图像中分离数学表达式的方法。我们给出了几个关于精确识别数学公式的新步骤。此外，我们还开发了一个实用程序，将OCR结果以各种文件格式输出；例如，还为视力障碍者准备了LaTeX用户界面。我们为日本的Microsoft Windows上的一个通用编辑器开发了一个说话宏。它分析LaTeX文件，并使用常用的屏幕阅读器朗读日语的数学表达式。我们已经执行了几次评估并改进了这个宏。最新型号适用于较高水平的科技材料。将这两种技术结合起来，即使学生有视觉障碍，他们也可以获得印刷的技术材料。

This research consists of two parts : 1. A Development of an Optical character recognition system for materials in print including mathematical expressions and 2. A development of acomputerized aloud-reading system for results obtained by OCR.We have achieved the following through the research, respectively.1. The method to separate mathematical expressions from scanned images is investigated. We give several new procedures concerning precise recognition of mathematical expressions. Furthermore, we develop a utility to output results of OCR as a variety of file formats ; for instance, LaTeX.User interfaces for persons with visual disabilities are also prepared.2. We develop a speaking macro for a generally used editor on Microsoft Windows in Japan. It analyzes LaTeX files and read aloud mathematical expressions in Japanese, using a generally used screen reader. We have performed several evaluations and improved this macro. The newest model works for scientific and technical materials of rather high level. Combining these two techniques, even if students had visually disabilities, they could access technical materials in print.