Segmentation of Printed Meitei/Meetei Script Documents

There are three main Process in Optical Character Recognition (OCR) System – Pre Processing, Segmentation and Recognition. Segmentation process of characters is one of the most crucial step in the development of OCR system of any language. Perfect segmentation of individual characters will determine the accuracy of the OCR system. It is used to segment the lines, words and individual characters from the document image. Meitei/Meetei script is not much popular script in India, but this language is schedule Indian language of Tibeto-Burman origin, which is also a very highly agglutinative language. Characters Segmentation of the Meitei/Meetei script is a difficult task because of the overlapping adjacent characters. In this paper we proposed a methodology, individual text lines and words are segmented by using Projection Profile technique. And for the individual characters we proposed Connected Component Analysis method. Proposed method was tested and segmentation accuracy rate of 95.6% is achieved.