Neuro-fuzzy techniques in the recognition of written Arabic characters

Author

Bouslama, Faouzi

Author_Institution

Dept. of Comput. Sci., Hiroshima Univ., Japan

fYear

1996

fDate

19-22 Jun 1996

Firstpage

142

Lastpage

146

Abstract

A new method for recognition of handwritten Arabic characters is presented. Characters are recognized by detecting their geometrical features and by conducting some discriminatory tests on their projection data. Most of the chosen features are easy to extract. Some of the features which are not so obvious are inferred from measurements. Fuzzy logic is used to model the uncertainties in the relationships between the variables. The 28 isolated letters of the Arabic alphabet are then classified by a feedforward neural network. The simulation results show the recognition rate is high though only a limited number of features has been involved

Keywords

feature extraction; feedforward neural nets; fuzzy logic; handwriting recognition; image classification; optical character recognition; uncertainty handling; character recognition rate; feature extraction; feedforward neural network; fuzzy logic; geometrical feature detection; handwritten Arabic character recognition; letter classification; measurements; neurofuzzy techniques; simulation; tests; uncertainty modelling; Character recognition; Computer science; Computer vision; Data mining; Fuzzy logic; Handwriting recognition; Histograms; Neural networks; Shape; Testing;

fLanguage

English

Publisher

ieee

Conference_Titel

Fuzzy Information Processing Society, 1996. NAFIPS., 1996 Biennial Conference of the North American

Conference_Location

Berkeley, CA

Print_ISBN

0-7803-3225-3

Type

conf

DOI

10.1109/NAFIPS.1996.534719

Filename

534719