A multi-font character recognition based on its fundamental features by artificial neural networks

Author

Neves, E.M.de A. ; Gonzaga, Adilson ; Slaets, Annie France Frère

Author_Institution

Inst. de Fisica de Sao Carlos, Brazil

fYear

1996

fDate

9-11 Dec 1996

Firstpage

196

Lastpage

201

Abstract

Neural networks present an alternative approach for the character recognition problem. This paper describes the development of a recognition system of multi-font character using topological feature extraction to recognize capital isolated letters. By properly specifying a set of features such as vertical, horizontal, and slant strokes, curvature, open and closed areas, called here “fundamental features”, the recognition was performed using a backpropagation neural network

Keywords

backpropagation; feature extraction; image classification; neural nets; optical character recognition; topology; artificial neural networks; backpropagation neural network; capital isolated letters; curvature; fundamental features; horizontal strokes; multi-font character recognition; slant strokes; topological feature extraction; vertical strokes; Artificial intelligence; Artificial neural networks; Character recognition; Feature extraction; Histograms; Humans; Neural networks; Optical character recognition software; Psychology; Text recognition;

fLanguage

English

Publisher

ieee

Conference_Titel

Cybernetic Vision, 1996. Proceedings., Second Workshop on

Conference_Location

Sao Carlos

Print_ISBN

0-8186-8058-X

Type

conf

DOI

10.1109/CYBVIS.1996.629463

Filename

629463