Novel Databases for Arabic Online Handwriting Recognition System

Inthis paper, novel databases for Arabic handwriting recognition, we call them Aswan online Arabic handwritten (AOLAH) databases, are presented. They are designed to help researchers in the field of online Arabic handwriting recognition. The first proposed database consists of 2,520 files, representing 28 Arabic characters written by 90 different writers. Those writers are volunteers from Aswan faculty of Engineering. The other database is extracted from the main database by taking only the strokes from characters. This second database contains 1,530 files for 17 strokes. The uniqueness of the recommended algorithm for data collection is the dealing with every stroke in the character separately. Stroke capturing is done by collecting data points along the path of an input device (stylus pen or mouse) same time those characters are drawn.