HANDWRITING-BASED BIOMETRIC KEY GENERATION SYSTEM

Abstract

Authentication process into any system or network involves secret knowledge of password, personal possession of private tokens and biometrics. The common drawback of both, knowledge and possession are caused by the fact that the authentication object can be handed over, stolen or get lost, and so it can be possibly used by non-authorised persons. Biometrics is by nature unsystematic and distinctive to a person and consequently they offer a scope of rendering themselves as a good material for biometric key generation. The target of this research work is to present handwriting based key generation system that generates biometric keys from handwriting semantics using Discrete Wavelength Transform – Discrete Fourier Transform (DWT-DFT) extraction with Random Number mixing to strengthen the biometric keys. Evaluation of the generated keys was done with the Hamming Distance and Percentage comparison method of the biometric keys. The semantics used revealed that for names forgers made a huge success of average of 68.53% and 66.46% for phone numbers enrolled while calculating the hamming distance. Also, the consistency in writing tested in an intra-class mode presents an average of 84.06% match while phone numbers had 80.00%. Though, professional forgers may be able to reproduce the handwriting over a period of time, based on the randomness and distinctiveness of these biometric keys, it is concluded that key generated from handwriting biometrics can be used for security on systems and networks. Meanwhile, handwriting biometrics is one of the recognised behavioural biometrics that is available for biometric key generation; it can generate biometric keys for authorization on software and computer networks.