Machine Recognition of Arabic Handprinted Scripts

Kambiz BADIE; Masamichi SHIMURA

Machine Recognition of Arabic Handprinted Scripts

Kambiz BADIE, Masamichi SHIMURA

Full Text Views

0

Cite this

Summary :

This paper presents some methods for recognition of handprinted Arabid scripts. Arabic scripts are generally structured of curves and hence our recognition method is based on the identification of the component curves in the script. In the paper, first we propose a method for recognition of single handprinted Arabid characters, using the concept of contour tracing and identification of the curves in the character. The identification of a curve is performed through determination of its local directions. Next, we develope the method to the recognition of cursively handprinted scripts, by usage of the identification of the basic shapes. The basic shapes are identified through using the orientation of the curves and the relationships between the feature points in the script. The results of computer simulations show average recognition rates of more than 95% and 90% for singly written and cursively written characters, respectively. The advantage of our method is its powerful recognition of highly distorted shapes, particularly the shapes which are generated through rotation of the original pattern. Although the present method has been developed for the recognition of Arabic scripts, it can be applied equally well for recognizing many other singly and cursively written scripts.

Publication: IEICE TRANSACTIONS on transactions Vol.E65-E No.2 pp.107-114

Publication Date: 1982/02/25

Publicized

Online ISSN

DOI

Type of Manuscript: PAPER

Category: Computers

Authors

Kambiz BADIE
Masamichi SHIMURA

Keyword

Cite this

Copy

Kambiz BADIE, Masamichi SHIMURA, "Machine Recognition of Arabic Handprinted Scripts" in IEICE TRANSACTIONS on transactions, vol. E65-E, no. 2, pp. 107-114, February 1982, doi: .
Abstract: This paper presents some methods for recognition of handprinted Arabid scripts. Arabic scripts are generally structured of curves and hence our recognition method is based on the identification of the component curves in the script. In the paper, first we propose a method for recognition of single handprinted Arabid characters, using the concept of contour tracing and identification of the curves in the character. The identification of a curve is performed through determination of its local directions. Next, we develope the method to the recognition of cursively handprinted scripts, by usage of the identification of the basic shapes. The basic shapes are identified through using the orientation of the curves and the relationships between the feature points in the script. The results of computer simulations show average recognition rates of more than 95% and 90% for singly written and cursively written characters, respectively. The advantage of our method is its powerful recognition of highly distorted shapes, particularly the shapes which are generated through rotation of the original pattern. Although the present method has been developed for the recognition of Arabic scripts, it can be applied equally well for recognizing many other singly and cursively written scripts.
URL: https://global.ieice.org/en_transactions/transactions/10.1587/e65-e_2_107/_p

Copy

@ARTICLE{e65-e_2_107,
author={Kambiz BADIE, Masamichi SHIMURA, },
journal={IEICE TRANSACTIONS on transactions},
title={Machine Recognition of Arabic Handprinted Scripts},
year={1982},
volume={E65-E},
number={2},
pages={107-114},
abstract={This paper presents some methods for recognition of handprinted Arabid scripts. Arabic scripts are generally structured of curves and hence our recognition method is based on the identification of the component curves in the script. In the paper, first we propose a method for recognition of single handprinted Arabid characters, using the concept of contour tracing and identification of the curves in the character. The identification of a curve is performed through determination of its local directions. Next, we develope the method to the recognition of cursively handprinted scripts, by usage of the identification of the basic shapes. The basic shapes are identified through using the orientation of the curves and the relationships between the feature points in the script. The results of computer simulations show average recognition rates of more than 95% and 90% for singly written and cursively written characters, respectively. The advantage of our method is its powerful recognition of highly distorted shapes, particularly the shapes which are generated through rotation of the original pattern. Although the present method has been developed for the recognition of Arabic scripts, it can be applied equally well for recognizing many other singly and cursively written scripts.},
keywords={},
doi={},
ISSN={},
month={February},}

Copy

TY - JOUR
TI - Machine Recognition of Arabic Handprinted Scripts
T2 - IEICE TRANSACTIONS on transactions
SP - 107
EP - 114
AU - Kambiz BADIE
AU - Masamichi SHIMURA
PY - 1982
DO -
JO - IEICE TRANSACTIONS on transactions
SN -
VL - E65-E
IS - 2
JA - IEICE TRANSACTIONS on transactions
Y1 - February 1982
AB - This paper presents some methods for recognition of handprinted Arabid scripts. Arabic scripts are generally structured of curves and hence our recognition method is based on the identification of the component curves in the script. In the paper, first we propose a method for recognition of single handprinted Arabid characters, using the concept of contour tracing and identification of the curves in the character. The identification of a curve is performed through determination of its local directions. Next, we develope the method to the recognition of cursively handprinted scripts, by usage of the identification of the basic shapes. The basic shapes are identified through using the orientation of the curves and the relationships between the feature points in the script. The results of computer simulations show average recognition rates of more than 95% and 90% for singly written and cursively written characters, respectively. The advantage of our method is its powerful recognition of highly distorted shapes, particularly the shapes which are generated through rotation of the original pattern. Although the present method has been developed for the recognition of Arabic scripts, it can be applied equally well for recognizing many other singly and cursively written scripts.
ER -