Recognition system for printed multi-font and multi-size Arabic characters

Abstract EN

This paper describes an automatic recognition system for Arabic writing.

The major problem in automatic reading of Arabic cursive characters is the segmentation of their constituent elements.

First, the text is segmented in lines, then in connected parts, and  nally in characters.

This segmentation is based on the contextual properties of Arabic writing and gives a solution to the problem of over-segmentation of some characters.

The adopted recognition method is a structural one.

It is based on the detection of holes and concavities in four directions (up, down, left, and right), which permits the classi cation of characters.

In addition, secondary characteristics are used in order to differentiate between the characters of each class and to reduce the confusions that are, in practise, encountered in every multi-font system.