An Assessment of Arabic Handwriting Recognition Technology

Abstract

Automated methods for the recognition of Arabic script are at an early stage compared to their counterparts for the recognition of Latin and Chinese scripts. An assessment of the technology for Arabic handwriting recognition is provided based on the published literature. An introduction to the Arabic script is given followed by a description of algorithms for the processes involved: segmentation, feature extraction, classification, and search. Existing corpora for Arabic are described together with a design for corpus collection. The paper is concluded by identifying technology gaps and providing a bibliography of the recent literature on Arabic recognition.