New M.E. Thesis Submitted from CSE Student

OFF LINE SIGNATURE VERIFICATION ANALYSIS USING RBP AND RBF CLASSIFIERS By Punit Soni,cse

Abstract
A signature is a hand written (and some time stylized) depiction of some ones name, nickname or even a simple “X” that a person writes on documents as a proof of identity and intent. It has been a distinguishing feature for a person’s identification through ages. Signature is a part of behavioural biometrics. Signature is used everyday to authorize the transfer of the funds of millions of people. Bank checks, credit cards and legal documents all require our signatures. Forgeries in such transaction cost million of rupees each year. Here forgery meant copying, falsifying, or altering any kind of written or printed matters for purpose of defrauding others.Signature verification is the process carried out to determine whether a given signature is genuine or forged. Handwriting comes in many different forms and there is great deal of variability even signature of people that use same language. Some people simply write their name, while others may have signature that are only vaguely related to their name and , some signature may be quite complex while others are simple and appear as if they may be forged easily. In this thesis work the signatures are taken from publicly available “Grupo de Procesado Digital de Senales” (GPDS) signature database. This database is used in various standard IEEE papers. First of all the signatures are converted into PBM format to get less information to process. Then feature extraction is done. Four different features (Centroid, Tri surface feature, Length, Six fold surface features) are used to prepare the data files. Two different files are used one to train the network and another to test the network. Finally two neural network Resilient Back propagation and Radial basis function network (RBP and RBF) are used as classifiers. RBF provides better verification accuracy than RBP.




Leisure Readings :