Enhanced Recognition of Ancient Character in Medicinal Manuscripts of Tamil Palm Leaves Using Machine Learning Techniques

Abstract

Optical character recognition is considered a mission  for Tamil text because of huge  number of difficult structures and scripts that differ from characters to font types. Due to the difficulties of identifying Tamil characters machine learning is a challenging task. .Ancient letters still pose a massive obstacle to converting since the technology of image recognition has almost mastered when the Tamil character text is scanned. This paper proposes a custom made Convolutional model to identify Tamil ancient manuscripts wriiten on palm leaves. A particular type of multilayer model is designed to boost the accuracy of recognition. It will distinguish patterns with extraordinary variations and intensity in geometrical shifts and distortions.  Finally Softmax classifier is used for character identification..The system proposed solves the problem by translating all manuscripts of palm leaves into digital script formats. The dataset includes individual manuscript characters from palm leaves. The model proposed enhances ancient Tamil character recognition accuracy in manuscripts of Palm Leaves .