Automatic Recognition of Artistic Arabic Calligraphy Types

In this paper, we propose a new approach to recognizing artistic Arabic calligraphy types, which are handwritten scripts written by special calligraphy pens. The nature of the structural composition of Arabic calligraphy makes it challenging to create such a recognition system: Difficulties include similarities among different types, overlap between letters, and letters that themselves assume different shapes. A new off-line technique of font recognition based on extracting distinctive features of each type is presented in this work. Features of selected types of artistic Arabic calligraphy are extracted to construct the features vector. The features vector is used in the classification process to recognize the type. A genetic algorithm is used to optimize the number of features and the image size that will be considered in the classification stage. In the classification stage, we used a neural network module. The approach is tested on two different datasets. One is a local dataset of three different Arabic handwritten calligraphy types, Thuluth, Reqaa, and Kufi. The other dataset is a public dataset of 10 different computer-generated fonts. The recognition error rate for the local and public datasets was 8.02% and 7.55%, respectively.