An Analytical Study on integration of Multibiometric Traits at Matching Score Level using Transformation Techniques

Biometric is one of those egressing technologies which are exploited for identifying a person on the basis of physiological and behavioral characteristic. However, unimodal biometric system faces the problem of lack of individuality, spoof attacks, non-universality, degree of freedom etc., which make these systems less precise and erroneous. In order to overcome these problems, multi biometric has become the favorite choice for verification of an individual to declare him as an imposte or a genuine. However, the fusion or integration of multiple biometric traits can be done at any one of the four module of a general multibiometric system. Further, achieving fusion at matching score level is more preferable due to the availability of sufficient amount of information present over there. In this paper we have presented a comparative study of normalization methodology which is basically used to convert the different feature vectors of individual traits in common domain in order to combine them as a single feature vector. Keywords— Biometric, Multibiometric, Normalization, Unimodal, Unsupervised Learning rules, Imposter, Genuine.