Quality-based Score Level Fusion for Continuous Authentication with Motion Sensor and Face

Quality-based Score Level Fusion for Continuous Authentication with Motion Sensor and Face

연구 분야: Analysis

학회: ICCSP 2020: Proceedings of the 2020 4th International Conference on Cryptography, Security and Privacy

In the continuous authentication, single biometric has their own advantages, disadvantages and the most suitable application scenarios, which are often interfered with by various factors. When biometric information cannot be observed by the system, to solve the disadvantage that single biometric cannot continuously authenticate users, we integrate multiple biometrics to improve the security and reliability of the authentication system. In this paper, we describe the quality-based score level fusion for multimodal biometric continuous authentication. In this scheme, motion sensor and face are selected as biometric features. Meanwhile, one-class classifiers are used to train data, and sequential fusion is adopted. Combined with the quality scores of data, the probability densities of the samples are calculated by the Gaussian kernel density estimation. At the same time, the authentication scheme is implemented on mobile phones, and the authentication experiments are carried out on datasets and mobile phones.