Automatic reading of bilingual typewritten test

Author

Abdelazim, H.Y. ; Hashish, M.A.

Author_Institution

IBM Cairo Sci. Center, Egypt

fYear

1989

fDate

8-12 May 1989

Abstract

Automatic test reading is an area of research in applied pattern recognition whose ultimate goal is to enhance the process of data entry. The authors describe a recognition system for the automatic reading of bilingual (Arabic/English) text. The presence of Arabic writing in a text imposes a great difficulty on the recognition process due to the cursiveness inherent in the Arabic script. The recognition system is developed based on a combined statistical/structural approach. A segmentation algorithm has been developed to resolve the cursiveness problem. A multistage classifier is used in the recognition phase. The system is software-based using a microcomputer (286 processor) and a binary scanner as the input device. The recognition rate reached is 99%

Keywords

character recognition equipment; pattern recognition; Arabic writing; applied pattern recognition; automatic text reading; bilingual typewritten test; binary scanner; data entry; microcomputer; multistage classifier; segmentation algorithm; statistical/structural approach; Automatic testing; Bibliographies; Books; Character recognition; Classification tree analysis; Image segmentation; Microcomputers; Optical character recognition software; Text recognition; Writing;

fLanguage

English

Publisher

ieee

Conference_Titel

CompEuro '89., 'VLSI and Computer Peripherals. VLSI and Microelectronic Applications in Intelligent Peripherals and their Interconnection Networks', Proceedings.

Conference_Location

Hamburg

Print_ISBN

0-8186-1940-6

Type

conf

DOI

10.1109/CMPEUR.1989.93394

Filename

93394