Enhancing multimodal mobile-based biometric authentication scheme for e-business transactions.
Hlongwane, Tiyani Christopher
Hlongwane, Tiyani Christopher
Citations
Altmetric:
Abstract
The increased use of mobile devices globally, such as smartphones and personal computers, has changed the way people communicate, buy, and sell online, and exchange data. As the use of mobile devices is increasing in the electronic business, more confidential information is stored on these devices; this increases risks such as the loss of login information by unauthorized users of the mobile device. Most of the users of mobile e-commerce utilize the traditional means of authentication to gain entry into the e-commerce transaction system. Such is proven to be unreliable, as the login credentials can be lost or stolen. In this research study, the multimodal mobile-based biometric identification system is proposed; by doing so, the e-business transaction system will only grant access to the legitimate user, thereby ensuring authentic authentication process. In a mobile-based e-business system, the process of personal identification during a transaction involves utilizing multiple source of information or traits, which is known as multimodal authentication. The proposed multimodal mobile-based biometric system is based on fingerprints, physiological traits, and facial recognition mechanisms through OpenCV. To simulate the multimodal biometric system for fingerprint and facial recognition, the OpenCV library in Python was utilized, using SIFT and face_recognition approaches. The facial recognition simulation was performed from the self-made database that consisted of 100 facial images captured from distinct facial and environment conditions; while the fingerprint simulator used the SOCOFing dataset that consists of 6000 fingerprint images with various fingerprint alterations. Some 98% of the fingerprint recognition accuracy rate was achieved using the SOCOfing dataset; while 87% of the facial recognition accuracy was achieved, this led to a combined accuracy rate of 93%. The accuracy rate mentioned pertains to the authentication of multiple modes collectively. Therefore, the use of the multimodal biometric recognition system outperformed the traditional and unimodal biometric recognition and authentication systems that achieved a recognition accuracy rate of 80% to 85% in an uncontrolled environment, including various facial expressions, and fingerprints distortions
Description
Submitted in fulfilment of the requirements for the degree Masters of Computing: Information Technology (Communication Networks) in the Department of Information Technology Faculty of Information and Communication Technology Tshwane University of Technology.
Date
2023-02-01
Journal Title
Journal ISSN
Volume Title
Publisher
Tshwane University of Technology
Research Projects
Organizational Units
Journal Issue
Keywords
Mobile device, Biometrics, E-business,, Multimodal biometrics, OpenCV, Traditional identification mechanismmobile