Date of Award


Publication Type

Master Thesis

Degree Name



Computer Science

First Advisor

Yuan, Xiaobu (School of Computer Science)


Computer Science.



Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.


For its unique advantages of preventing the loss of user identification, biometrics authentication is being increasingly used on mobile devices to meet the demand of access control and electronic transactions. Biometric community has been working on different approaches to improve reliability of security systems, multimodal authentication has attracted a lot of attention for its advantages over uni-modal biometric matchers. Nevertheless, errors caused by noises existing in real-world circumstances have become a major fact that slows down its acceptance in mobile computing. Aimed at improving the reliability of biometric authentication, current practice uses score-level fusion to combine normalized outputs of multiple classifiers. By investigating the performance of different score-level fusion methods with normalization techniques in different noise conditions, this work develops an algorithm to analyze the individual biometric matching scores in different noise conditions and dynamically select the combinations of normalization and fusion methods that are adequate for different working environments.