Performance enhancement of multibiometri authentication system using score level fusion

Abstract

Biometric systems are the most secured system to identify the newlineperson based on their physical and behavioral components. Nowadays newlinemany bio metric systems are used for authentication purpose. Since newlinePhysical components never change at life time (finger print, iris, etc). newlineBiometric systems are more secured than any other traditional security newlinesystems. A biometric system is most securable component rather than other newlinesecurity models. Because every human is having unique characteristics. newlineEvery biometric feature is having their own limitations and advantages. newlineNormally all the biometric systems are used for authentication or newlineverification process. Significantly if it is not possible to conclude that one newlinebio metric system is best among all the available systems. Because it newlinedepends upon the application, where it is being used. newlineGenerally, Unimodel systems are based only one entity of human. newlineSo, it causes problems like attacks, noisy data, unaccepted error etc. To newlineovercome the above problem only, the multi biometric system are newlineintroduced. It is good option that two physical traits of human can be newlinecombined together in multi biometric system. Evolutionary algorithms play newlinemain role in multibiometric systems, because it gives optimal solution newlineamong large population and final optimal solutions are achieved through newlinethe updating and searching of the past features of ACO, PSO, and Genetic newlineAlgorithms. Normalization of score is difficult in fusion process when we newlineare combining different type of biometric modalities. Even after newlineauthentication process, the output score obtained by each individual model newlineare need not be same in nature and may not be in same numerical range. newlineNormalization technique must convert the output score in common range newlinethat can be used for further decision process. newline