Arabic writer verification based on shape complexity

Writer recognition task has received a lot of interests during the last decade due to it wide range of applications. This task includes writer identification and/or verification. In this paper we propose an original approach for the Arabic writer verification based on shape complexity. In this approach two handwritten words are submitted to the system, which will proceed to a segmentation to each of them into graphemes. A dedicated segmentation module for Arabic handwriting has been developed for this purpose. All the graphemes generate are then analyzed to select the one who embed the most relevant information about the writer identity in each word. This step is performed through a shape complexity measure that determines the most complex shape (grapheme). The decision, if the two samples have been written by same writer or not is based on the similarity of the two most complex shapes extracted from each word. The approach has been evaluated on dataset of 105 and the obtained results are very promising since we succeed to accept correctly in 80 % of cases when we used the whole dataset and reject correctly in 89% of the cases when we use 20% of the dataset.