Pattern Recognition Framework For Transliterating Modi Script For Visually Impaired People

Abstract

The many families in the state of Maharashtra and surrounding state of it newlineare the majorly know about the Ancient MODI script. The MODI script is not newlineeasy to read for the unknown people, because the writing style is cursive and not newlinein readable form. Also, the very few persons know the knowledge of the Ancient newlineMODI script. According to the literature available these ancient MODI script newlinedocuments are available in the form of property documents, ancienter newlinedocuments, historical document, etc. These documents are access by the Visually newlineimpaired people and sight people too. Understanding the Ancient MODI script newlinehandwritten character and transliteration is the major challenge in the field newlinepattern recognition. newlineIn the Present research work the transliteration framework of MODI script newlineto Braille script was designed. The major purpose to design this framework is to newlinehelp the visually impaired people majorly. The application for transliterating the newlineMODI script to Braille script is designing with the help of MATLAB. Through newlinethis research experimental work, the different challenges related with ancient newlineMODI script transliteration was identified. Secondly the evaluation of OCR newline(Optical Character Recognition) related to ancient MODI script was newlineaccomplished. With the help of transliteration framework, the mapping of MODI newlinescript and Braille script was finished appropriately. The present application not newlineonly showcase the induvial braille cell of respective MODI script handwritten newlinecharacter but generate the Audio form output to the end user.The full proof design of present experimental work was evaluated by newlineusing MATLAB programming platform. The overall application was designed newlinein the pretrained conventional neural network i.e., AlexNet. The handwritten newlinecharacter recognition application implementation is judge based on the number newlineof characters recognized correctly i.e., accuracy ratio. The final accuracy of newlinegenerated by machine using K-fold method was 94.18 %.