Abstract

Optical Character Recognition (OCR) of handwritten text is still a challenging research area due to different writing styles of different individuals. Character recognition helps in extracting important and required text from a document. The difficulty in handwritten character recognition is due to different individual writing styles and diacritics in some languages like Yoruba which lead to tonal difference in their text making handwritten recognition system a complex system. To overcome this challenge, researchers are carrying out researches to discover better ways of developing automated character recognition systems with excellent recognition accuracy. This study aims to review the work on computer vision regarding character recognition in the last seventeen years. Over 52 publications on OCR of handwritten documents are extracted from science direct, google and scopus databases. Review of Machine Learning techniques for handwritten character recognition is presented. Machine learning methods used for classification in OCR used to achieve better results are also presented. The Literature review presented in this study reveals importance of OCR in areas such as automated library and office documents, bank check processing, banking services, postal services, and health service and in museum for National archives for digital searchable storage of historical text and shows that perfect OCR system is a difficult and challenging research area in terms of handwritten character recognition. This study would provide knowledge and creation of perfect character recognition systems and guide researchers and expert in future research