Recognition the Arabic Characters Based On the Characteristics of Arabic Language

Abstract

Characters recognition problem is a special case of image analysis, so it has received more attention than any other pattern recognition problem. The only real problem is to design a system extracted the features efficiently and describes the character perfectly. Much research has been done regarding the recognition of Latin, and Chinese characters. In contrast to Arabic language, very little investigation has been performed regarding Arabic characters recognition. This paper seeks to provide a comprehensive list of the characteristics of the Arabic characters, and to address some challenges posed by the nature of the characters.One of the important ways of summarizing a two-valued picture (i.e., binary image) is to replace all the thick lines by suitable thin lines representing the real boundaries. In this paper, thinning process is performed by using the skeleton process, which summarized the character data. Zhang-Suen algorithm is used to find the skeleton of the Arabic characters, after carrying out several developments on this algorithm for keeping the basic forms of the character from decay when using this algorithm in its known form like the group of dotting associated with the dotted letters. In this paper, a technique has been employed to recognize isolated Arabic characters. A chain coding technique based algorithm along with other significant peculiarities such as number and location of dots and secondary components features, and the number of closed loops feature existing in the isolated Arabic character has been used in this study to recognize Arabic characters. Experimental results show the relatively high accuracy of the method developed when it is tested on several standard Arabic fonts.