Optical Segmentation of Devanagari Script:A Scientific Analysis

In the subject Artificial Intelligence, OCR is of keen interest for the computer scientists and researchers. The reason being that the OCR helps in digitizing the printed or handwritten document into computer readable format. These digitizing of document have various futuristic use such as record of old document could be kept in track, multiple data storage of handwritten or printed document, etc. A large number of articles have been published in this area in various journals. A few works have been done in Devanagari Script. The Devanagari Script makes a full set on many other scripts like Hindi, Konkani, Marathi, Nepali, Sanskrit, Bodo, Dogri and Maithili. Challenging rate of OCR in Devanagari Script have not been solving in a better rate. Hence as effort have been generated to achieve a better rate of Segmentation in this Script. Objectives of this study are to find out the accuracy level of word and characters segmentation of Devanagari Script and then to analyze the Challenges faced during the process of word and character segmentation. The finding shows that the accuracy rate of character Segmentation is 89% and word segmentation is 91.70%. Keywords— Shirorekha; Devanagari; Printed; OCR; Conjunct