Fuzzy vault based multimodal biometric authentication system using fused feature vectors

Abstract

In the modern society, Internet has been emerged as one of most important and commonly used technologies for sharing information. In secure communication, accurate authentication plays a vital role. But traditional authentication systems have some drawbacks like stealing of password and vulnerable to attacks. So, it is essential to develop an authentication system by using biometrics to solve the problems of traditional system. Biometric is the branch of science that recognizes the person based on the stored features. A template summarizes the group of important features of biometric. But biometric template is subjected to many attacks, so providing security to the biometric template is an important issue. The main focus is to implement a new technique for securing biometric template using fuzzy vault scheme and analyze the security of the template against some attacks. The proposed methodologies used in this research work are: i) Fuzzy vault based Unimodal biometric authentication ii) Fuzzy vault based Multimodal biometric authentication Unimodal biometric system identifies a person based on single biometric source of information. Fuzzy vault is a powerful cryptosystem and it provides high security for biometric template. Fingerprint biometric system is one of the most popular and accurate biometric authentication techniques because fingerprint characteristics will not be changed in a person s life. So, fuzzy vault based fingerprint template protection method is implemented. The fingerprint template protection scheme is based on the minutiae point of the fingerprint image and it is extracted by using cross numbering. Extracted features are prealigned based on the directed reference point and featurepoints are quantized to resist against correlation attack. Efficiency of the fuzzy vault based fingerprint template protection scheme is analyzed by varying the degree of polynomial. Furthermore, Security of the fuzzy vault based fingerprint scheme is also tested against brute force attack and correlation attack.