Segmentation of Arabic Handwritten Text to Lines

Automatic recognition of writing is among the most important axes in the NLP (Natural language processing). Several entities of different areas demonstrated the need in recognition of handwritten Arabic characters; particularly banks check processing, post office for the automation of mail sorting, the insurance for the treatment of forms and many other industries. One of the most important operations in a handwriting recognition system is segmentation. Segmentation of handwritten text is a necessary step in the development of a system of automatic writing recognition. Its goal is to try to extract all areas of the lines of the text, and this operation is made difficult in the case of handwriting, by the presence of irregular gaps or overlap between lines and fluctuations of the guidance of scripture to the horizontal. In this paper, we have developed three approaches of handwritten Arabic text segmentation, then we compared between these three approaches.