Multimodal Biometric Authentication System For Enhanced Security

Abstract

The multimodal biometric recognition system is one of the recent trends as it paves newlinethe way for automatic recognition of humans through their biometrics. Usage of newlineseveral combination of biometrics, such as finger vein, knuckle, palm print, and newlinehuman cue, solve the security-related issues prevailing in various environments. Even newlinethough numerous research works related to the biometric recognition have addressed newlinevarious challenges, handling of the score level fusion for integrating various newlinebiometrics is more essential. This work profoundly contributes two biometric newlinerecognition algorithms, namely Fuzzy Brain Storm Optimization (FBSO) and Fuzzy newlineLeast Brain Storm Optimization (FLBSO) by using the biometrics, such as hand vein, newlinepalm vein, and the finger vein. Both, the proposed schemes perform the biometric newlinerecognition by utilizing the vein based patterns. The proposed FBSO algorithm is newlinenewly formed by integrating the fuzzy theory with the Brainstorm Optimization newlineAlgorithm (BSO). Here, optimum fusion score for integrating the features of various newlinebiometrics is identified through the prediction based distance measure. The second newlinealgorithm, FLBSO uses the least mean square (LMS) algorithm along with the FBSO newlinefor calculating the optimal fusion score. Also, FLBSO algorithm employs the newlineEntropy-based Euclidean Distance (EED) for identifying the fusion score. The newlineproposed FBSO and the FLBSO algorithms are evaluated with the use of the standard newlinedatabases, and evaluated based on the metrics, accuracy, False Acceptance Rate newline(FAR), and False Rejection Rate (FRR). Analyzing based on accuracy metric, the newlineproposed FBSO algorithm achieved accuracy value of 0.8130, while the proposed newlineFLBSO algorithm obtained a relatively higher accuracy value of 0.8990. Considering newlinethe performance of the proposed FBSO and the FLBSO algorithms based on the FAR newlinemetric, the analysis shows, the FBSO has the FAR value of 0.1870, but the better newlineperformance is achieved by FLBSO with low FAR value of 0.1010. Further, analysis newlinebased on FRR shows that the proposed FBSO and the FLBSO algorithms have newlineachieved improved performance with the values of 0.1009 and 0.0088, respectively. newline