A wide variety of applications require reliable verification schemes to confirm the identity of an individual. The emergency of biometrics helps to solve the problems that the traditional methods such as password and IC cards have faced. But there are many problems such as noisy data, non-universality, which may affect the performance of the biometrics system when using a single biometric feature. Multiple biometrics can help to solve several practical problems, which single biometrics recognition system have faced. Based on the analysis of recent research on the multi-modal fusion system, we investigate three different kinds of multi-modal fusion system in this thesis including the voiceprint and fingerprint fusion system, the voiceprint and face fusion system and the gait and face fusion system. The main contributions of this thesis are as follows: 1. Compare 13 different kinds of fusion method in the context of the voiceprint and fingerprint fusion system and the experimental results show that Support Vector Machine and the Dempster-Shafer method are superior to other methods. 2. Compare several different kinds of normalization methods and fusion methods based on the XM2VTS database and also the combination system of more than two classifiers is investigated based on the same database. 3. Introduce a hierarchical fusion system of gait and face and also we investigate the fusion of gait sequence with different view angle. To the best of our knowledge, this is the first study on the fusion of multi-view gait sequence.
修改评论