Automatic recognition of Arabic characters using logic statements. Part I. System description and preprocessing

Abstract

This article presents an automatic method of reading Arabic printed characters developed in the image processing laboratory of the electrical engineering department of King Saud University. The technique uses a special normalization procedure prior to the development of individual Boolean codes for each character. In an experiment using this method, 100% recognition results were obtained for a set of 58 typewritten characters. In part I of the article the recognition scheme is described and the normalization procedure is given in detail. Part II continues with the development of recognition algorithms and the application results.