Please use this identifier to cite or link to this item: http://hdl.handle.net/10603/40736
Title: Optimized Complex Extreme Learning Machine for Classification of 11th Century Handwritten Tamil Scripts
Researcher: Sridevi N
Guide(s): Subashini P
Keywords: Handwritten Characters
11th century Tamil Scripts
ELM, DE-CELM
Upload Date: 9-May-2015
University: Avinashilingam Deemed University For Women
Completed Date: 01/04/2013
Abstract: Tamil is one of the ancient languages which contain information about ancient history of India medicinal notes astrology and so on which have been written in ancient Tamil scripts which is different from current Tamil scripts If these inscriptions were digitized the contents available in them can be used by various categories of people with ease and comfort Hence this research work is carried out in order to classify the handwritten ancient Tamil scripts that facilitate the development of handwritten character recognition The main objective of this research work to find an optimal solution for the classification of handwritten Tamil scripts In order to achieve this the various tasks in image processing are carried out First a binarization algorithm is proposed using Otsu and Particle Swarm Optimization technique to convert the input samples into a binary image As the document image analysis methods are very sensitive to rotation the document skew should be corrected For skew detection and correction modified projection profile method is used This method corrects the document skew with minimum estimation error Using Particle swarm optimization technique the text lines are segmented from the document and to segment the characters a method combining connected components and nearest neighborhood method is used Then a new set of feature vectors is formed by extracting the features from the character segments using Zernike moments and regional features The features are fed to Extreme Learning Machine ELM and Complex Extreme Learning Machine CELM for newlineclassification ELM and CELM takes higher number of hidden neurons in order to classify the 11th century scripts Hence in order to obtain the highest accuracy for the classification with minimum number of hidden neurons a new method is proposed using Differential Evolution algorithm in the CELM The proposed method is tested on 11th century handwritten Tamil scripts It is observed that the proposed method achieves a higher classification rate when compared with other methods
Pagination: 
URI: http://hdl.handle.net/10603/40736
Appears in Departments:Department of Computer Science

Files in This Item:
File Description SizeFormat 
nsridevi_chapter1.pdfAttached File482.26 kBAdobe PDFView/Open
nsridevi_chapter2.pdf296.78 kBAdobe PDFView/Open
nsridevi_chapter3.pdf434.88 kBAdobe PDFView/Open
nsridevi_chapter4.pdf615.48 kBAdobe PDFView/Open
nsridevi_chapter5.pdf757.67 kBAdobe PDFView/Open
nsridevi_chapter6.pdf260.43 kBAdobe PDFView/Open
nsridevi_chapter7.pdf2.73 MBAdobe PDFView/Open
nsridevi_intro.pdf240.63 kBAdobe PDFView/Open


Items in Shodhganga are protected by copyright, with all rights reserved, unless otherwise indicated.