Enhancing multimodal biometric frameworks using fuzzy fusion at matching score level
No Thumbnail Available
Date
2021
Journal Title
Journal ISSN
Volume Title
Publisher
University of Oum El Bouaghi
Abstract
Multimodal biometric systems aim to achieve the best performance in persons recognition, by combining more than one biometric modality so that the combination is done using the fusion principle. In fact, although different levels of fusion are available, score-level fusion is the most commonly used. Indeed, in this process and before combining, the heterogeneous scores provided by the different biometric subsystems must be normalized. In this paper and in order to cancel the scores normalization step, a fuzzy fusion strategy has been proposed. Our experimental results, using a PolyU finger-knuckle-print database, show that the proposed fusion scheme improves the results compared to the Z-score normalization method followed
by the min and max classical combining arithmetic rules.
Description
Keywords
Biometrics; Multimodal biometrics; Data fusion; Fusion at score level. Fuzzy logic