A multi-layer classifier for recognition of unconstrained handwritten numerals

Author

Wang, Gwo-En ; Wang, Jhing-Fa

Author_Institution

Dept. of Electr. Eng., Nat. Cheng Kung Univ., Tainan, Taiwan

Volume

2

fYear

1995

fDate

14-16 Aug 1995

Firstpage

849

Abstract

A hierarchical architecture for recognition of the unconstrained handwritten numerals is proposed. In the first stage of preclassification, a set of structural features named four-zone codes is adopted to preclassify the numerals. Due to the large degree of data and distortion of characters, it is possible to classify two different numerals with same features into a class. A secondary preclassification that utilizes topological stroke features is presented to solve this ambiguity. In order to promote the recognition rate to be a practical OCR system, a three layer Bayesian neural network with 20 dimensional global feature vectors is designed for fine classification of the confusing classes. Experimental results show that the recognition rate of the proposed hierarchical OCR system for handwritten numerals is over 99.82% based on 15423 samples

Keywords

Bayes methods; handwriting recognition; image classification; multilayer perceptrons; neural nets; optical character recognition; 20 dimensional global feature vectors; fine classification; four-zone codes; handwritten numerals; hierarchical OCR system; hierarchical architecture; multilayer classifier; practical OCR system; preclassification; secondary preclassification; structural features; three layer Bayesian neural network; topological stroke features; unconstrained handwritten numeral recognition; Bayesian methods; Character recognition; Computer vision; Feature extraction; Handwriting recognition; Neural networks; Optical character recognition software; Shape; Spatial databases; Testing;

fLanguage

English

Publisher

ieee

Conference_Titel

Document Analysis and Recognition, 1995., Proceedings of the Third International Conference on

Conference_Location

Montreal, Que.

Print_ISBN

0-8186-7128-9

Type

conf

DOI

10.1109/ICDAR.1995.602034

Filename

602034