A Study on Multifactor Biometrics Based on Convolutional Neural Network

With the recent advancements and rise in technological applications, information security and privacy are the most prioritized domains in which people are concerned about. Be it a governmental institution or a small business firm, all their relevant data is being stored online and exposed to cyberattack in the digital world. Authentication using biometrics has been considered as the most suitable method owing to its universality, uniqueness, acceptability, and performance. However, relying solely on a single mode authentication system (referred to as Unimodal Biometric System) may prove inadequate in ensuring privacy and security of digital data. Hence, it has become essential to use multiple biometric modalities in verification/authentication systems. Multimodal biometric or multifactor biometric system consists of more than one physical or behavioral trait as an additional layer of protection for authentication of confidential data. present paper provides a brief account of three multimodal biometric systems and a comparison based on their experimental results. Here, the integration of the biometric traits is accomplished using the Convolutional Neural Network (CNN) approach, where facial recognition has been fused with other biometric traits in order to achieve the most accurate match for its implementation in real world applications.