Recognition of typewritten Arabic characters in different fonts

The recognition of Arabic typewritten characters is complicated by the lack of letter separation. In the paper a syntactical method has been developed for the recognition of Arabic cursive words in three different fonts at character level. These typewritten words are entered into an IBM pc via a camera and digitizer. The method is based on segmentation of words into ten primitives. These primitives, together with some other features and a few selected geometric measurements, are used in the final decision on the character