OPTICAL CHARACTER AND WORD RECOGNITION OF ARABIC PRINTED TEXT

the project objective to:Prepare a general study about the OCR field, Convey the standard methods used in OCR systems and design others that suit the Arabic language and Design an Arabic OCR system that will be able to recognize an Arabic printed text.The Optical Character Recognition (OCR) is defined as the process of automatic conversion of scanned images of machine printed or handwritten documents into a format that a computer can process. Many OCR systems are well developed and implemented for different languages such as English and Japanese. However, Arabic suffers from the lack of research. This thesis represents a design of an Arabic OCR system. It is designed to recognize a range of sizes between 18 to 72 points of Times New Roman font. Basically, a set of unique and robust features are selected to define each Arabic letter. These features are extracted from templates images and saved for a later comparison. A proposed algorithm is used to segment the Arabic text into letters. These letters are then passed to a feature extraction stage which extracts those pre-defined features from the letter. The extracted features and the saved ones are compared together to recognize the letters. A recognized letter is written in a text file