Optical Character Recognition for Printed Sanskrit Text using OPEN-CV

Abstract

India has various treasures of knowledge which are inscribed in various materials such as stones, tree barks and leaves. The life time for these materials is limited and as the generations pass by, these materials are decaying and the valuable information is getting vanished. This paper uses the manuscripts which are digitized in order to pass the knowledge to future generations. The recognition of characters is helpful to extract the information from these digitized manuscripts using Optical Character Recognition (OCR) technique (printed text). In this paper OCR is implemented for recognizing printed Sanskrit characters in Devanagari script by using convolution neural network principle. The result obtained have justified that the proposed system projects good recognition rate for the input dataset. These results are evaluated using well known evaluation metrics namely Precision, Recall.

 Keywords:Optical Character Recognition (OCR), Character Recognition, Devanagari Script, Precision, Recall, Convolution Neural Network.