Recognition of Arabic Handwritten Diacritics using the new database DBAHD

Abstract

Recognition of handwritten Arabic characters is gaining momentum and research in this area has increased considerably in recent years. However, research remains modest compared to that performed in other scripts. This is mainly due to the morphology of Arabic writing, in particular its richness in diacritical marks. This signs are generally recognized by adopting structural or morphological measures. However, the difficulty and variability of handwriting can sometimes be misleading, thus influencing the results obtained. This article presents a new database for Arabic handwritten diacritics (DBAHD). It is designed to serve the Arabic handwriting recognition systems based on segmentation and machine learning.