for Editing Documents
Yoji Maeda and Masaki Nakagawa* Mitsubishi Electric Corporation 325 Kamimachiya, Kamakura, Kanagawa, 247-8530, Japan e-mail: firstname.lastname@example.org
* Dept. of Computer, Information and Communication Sciences Tokyo Univ. of Agri. &Tech. Naka-cho 2-24-16, Koganei, Tokyo, 184-8588, Japan phone: +81-423-88-7144, fax: +81-423-87-4604 e-mail: email@example.com
Generally speaking, document preparation has become easier due to the development of word processing. Nevertheless, some people dislike typing a keyboard and feel pains to operate PCs.
On the other hand, document OCRs have been widely used to utilize existing documents to prepare new documents. However, their recognition performance is still unsatisfactory. Especially, the Japanese documents are difficult to read because the mixture of Kanji, phonetic characters, numerals and even alphabets makes their segmentation and recognition very hard. Therefore, a user has to correct recognition errors using word processing software. Then, the user needs to learn how to type and how to use the word processor.
We are developing a system that recognizes correction marks written on a printout of a document and apply them to the document automatically, thus allows a user to edit documents without using a keyboard.
So far, some researchers have reported systems for correcting a document without using a keyboard. One of the earliest works was done by Suenaga ( 1980). His system read a document with handwritten correction marks directing deletions and permutations of characters and performed designated