Arabic-Jawi Scripts Font Recognition Using First-Order Edge Direction Matrix

Abstract

Document image analysis and recognition (DIAR) techniques are a primary application of pattern recognition. OFR is one of the most important DIAR techniques. The information about font type indicates important information to support human knowledge and other document analysis and recognition techniques. In this paper, a new optical font recognition method for Arabic scripts is proposed based on the First order edge direction matrix, which is an effected simple feature extraction method for binary images. The proposed methods based on several recent methods in pre-processing and feature extraction stages. The performance of the proposed method is compared with the previous OFR methods that based on texture analysis methods in the feature extraction stage. The results show that the proposed method presents the best performance than of other methods in terms of computation time and accuracy. © Springer-Verlag Berlin Heidelberg 2013.