Study of Different Fusion Techniques for Multimodal Biometric Authentication

This article gathers the studies and experiments carried out by the authors orientated to the combination of three unimodal systems into a single multimodal system. The fusion strategy used has been the so-called "score fusion" using different algorithms, among them "neural networks" and "support vector machines", etc. All of the experiments have been carried out from the same group of data obtained as a result of the execution of the processes of three unimodal verification systems on three independent biometrical databases. Results show the great improvement that fusion can generate when unimodal results are not good enough.