Certain investigations on classification of indo aryan and tamil text from historic manuscripts

Abstract

Text segmentation is one of most important step in document newlineimage processing and is classified into two major areas viz hand written and newlineprinted text segmentation This thesis is focused on printed text segmentation newlineand classification The text segmentation of printed characters has the newlinefollowing issues like touching of characters slant characters mixed font style newlinedifferent font sizes which reduces the efficiency of segmentation algorithms newlineThe need for better character segmentation and recognition algorithms for newlineIndian Languages has been a demand for more than a decade Wide research newlinehas been carried out and appreciable results have been obtained for languages newlineviz English French and European Languages Experiments have also been newlinecarried out for languages with compounded and complex characters like newlineArabic and appreciable results have been obtained newlineIndia is a land of multi-script country with eighteen different scripts newlineauthorized by The Government of India In the land of diversity minor newlinelanguages are available with its own scripts are available One such language newlineis Saurashtra which belongs to the Indo Aryan sect The language with its newlinecomplex compounded character set has been very minimally researched for newlinecharacter segmentation and classification newlineThis thesis introduces a novel text segmentation and classification newlinealgorithm which can segment and classify the texts written in complex newlinedesigned scripts of Indo-Aryan origin with special focus on Saurashtra newline newline