Development of Novel Techniques to Recognize Tamil Language Characters

Abstract

To rebuild the history and culture of ancient Tamil civilizations, Inscriptions plays a primary role. In order to disclose the details from the ancient Tamil civilizations, Ancient Tamil character has to be indentified and it is the significant task. Modern Tamil language s scripts have been developed over centuries, and finally it was converted to the original format. The current approaches tries to identify the ancient Tamil characters by utilizing the SIFT features and presented a fresh and effective method according to the bag-of key points representation. These fixed-length feature vectors are utilized to categorize the characters. The proposed methods accomplishes a highest accuracy of about 84 % by utilizing SIFT features. Nevertheless, it doesn t effectively identify the warped patterns or structures comprehended by elegant features of the Tamil scripts. And also this approach has maximum computational complexity and low efficiency. newline