Details

Autor(en) / Beteiligte

• Rao, T Srinivasa
• Reddy, E Srinivasa

Titel

Multimodal Optimal Score Level Fusion Method In Multimodal Biometric System

Ist Teil von

• Webology, 2022-01, Vol.19 (2), p.8565-8586

Ort / Verlag

Tehran: Dr. Alireza Noruzi, University of Tehran, Department of Library and Information Science

Erscheinungsjahr

2022

Link zum Volltext

Quelle

EZB Free E-Journals

Beschreibungen/Notizen

• An invariable mix of Biometric recognition as well as authentication process is used in almost every application for prevention of unauthorized access to computing resources. Though several research initiatives have been undertaken in recent years for making biometric algorithms more accurate for various kinds of features, they have largely ignored the critical aspects like reliability and robustness. Typically, information from multiple modalities is fused in a Multimodal biometric system to counter the shortcomings an individual classifier. This work primarily seeks to make optimum integration of biometric traits viz., iris, fingerprint and finger vein that have great complementary relation among them. It is basically done through optimization of performance in every single classifier through a novel Search Optimization Algorithm with backtracking function. Moreover, it tries to resolve conflicting beliefs affecting individual classifiers through use of proportional conflict redistribution rules, before obtaining a concurrent solution. Optimal behavior under a dynamic environment is shown in this system, either by boosting or, even by suppressing concurrent classifiers, thereby ensuring that classifiers do not have any more issues between them. The suggested multimodal system is developed by applying multimodal datasets generated using benchmarked digital imagery. Our system performs better averaging precision at 98.43% having 1.57% EER, thereby outperforming the existing techniques and promising to develop an advanced multimodal biometric system in subsequent efforts.