Please use this identifier to cite or link to this item: http://hdl.handle.net/10603/219506
Title: SPEAKER IDENTIFICATION BASED ON BIOMETRIC FEATURES USING SOFT COMPUTING TECHNIQUES
Researcher: Dekate S K
Guide(s): Zadgaonkar A S
Keywords: ANFIS
Artificial Neural Network,
Face Recognition
Features Extraction
Multi view
PSO-NN
University: Dr. C.V. Raman University
Completed Date: 2015
Abstract: Biometric finds wide application in the field of recognizing to identifying or recognize newlinethe person by their physical or behavioral characteristic. These characteristic may be face, newlinefinger, retina, gait, speech etc. It is more secure than password because it cannot be newlineshared, copied or lost. It is associated with the biological features of the person itself. newlineThe present work uses facial biometrics to recognize the people. As compared to other newlinebiometric; like finger and palm, face has distinct advantage of being a non contact newlineprocess. Face recognition use the spatial geometric or distinct features of face. But it is newlinenot always efficient to use only front view of face because of non-cooperative behaviors. newlineSo this work used up, front and down view in the face based recognition process. For newlineeach view some important special geometric features like right eye height, right eye newlinewidth, right eye area, left eye height, left eye width, left eye area, mouth height, mouth newlinewidth, nose width, face height, face width, face area, center of mass are extracted. Data newlineset are created for each view separately and the soft computing models like ANN, PSONN newlineand ANFIS are used to train and test the model. newlineIn the neural network based recognition process the optimum efficient model has been newlinedesigned by changing parameters like number of neurons in hidden layer to create the newlinevariation of models. The neural network model is having one input, one output and 10 newlineneurons in the hidden layer, training function is Levenberg-Marquardt, learning mu rate newlineis .0001, and performance function is mean square error with random data division. This newlinework checks the accuracy of individual face view and combined face view and the result newlineshowed that combined view gives the good results as compared to individual and the newlineaccuracy of the result is 97.2%.
Pagination: 
URI: http://hdl.handle.net/10603/219506
Appears in Departments:Department of Electronic Engineering

Files in This Item:
File Description SizeFormat 
acknowledgement.pdfAttached File4.71 kBAdobe PDFView/Open
appendix.pdf3.04 kBAdobe PDFView/Open
certificate.pdf1.03 MBAdobe PDFView/Open
chapter1.pdf270.66 kBAdobe PDFView/Open
chapter2.pdf307.07 kBAdobe PDFView/Open
chapter3.pdf934.21 kBAdobe PDFView/Open
chapter4.pdf4.15 MBAdobe PDFView/Open
chapter5.pdf1.05 MBAdobe PDFView/Open
chapter6.pdf1.89 MBAdobe PDFView/Open
chapter7.pdf90.44 kBAdobe PDFView/Open
list of figure.pdf72.23 kBAdobe PDFView/Open
list of publication.pdf1.33 MBAdobe PDFView/Open
list of table.pdf5.59 kBAdobe PDFView/Open
refrence.pdf341.68 kBAdobe PDFView/Open
table of content.pdf65.09 kBAdobe PDFView/Open
title.pdf15.05 kBAdobe PDFView/Open


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