CNN-Bidirectional LSTM Based Optical Character Recognition of Sanskrit Manuscripts : A Comprehensive Systematic Literature Review

Abstract

Optical character recognition (OCR) is a technology that allows you to convert different types of documents or images into searchable, editable, and analyzable data. The current work on Sanskrit Character Recognition from Images of Text Documents is one of the most difficult due to similarities in the forms of unique letters, script complexity, non-forte in the representation, and a vast number of symbols. The Devanagari script is used to write the Sanskrit language. There are a variety of approaches for recognizing characters in a scanned image. The present research initiatives highlight the importance and needs of efforts in recognition of printed and handwritten documents written in Sanskrit language. This paper is aims at reviewing the state of various scripts in use including those from medieval to present era and explores the prospective of digital recognition of printed texts and thereby pointing towards futuristic trends in developing restoration software for Sanskrit scripts. Challenge is due to the number of languages and their diverse scripts. The scarcity of digitized linguistic resources makes the task a tougher one. The paper also highlights on the characteristics and challenges of recognition of scripts of Sanskrit origin. Largely the digital recognition is limited to simple numerals and isolated characters. In addition, this review article serves the purpose an optical character recognition (OCR) system that enables to analyse the word recognition and translate various types of Sanskrit documents or images into text using deep learning architectures which include Convolutional Neural Network (CNN) and Bidirectional long-short term memory (Bidirectional LSTM).