A comparative evalution of feature level based fusion schemes for multimodal biometric authentication

This paper proposes a novel fusion technique using iris-online signature biometrics at feature level space. The biometric features are extracted from the pre-processed images of iris and the dynamics of signatures. We propose different fusion schemes at feature level. In order to reduce the complexity of the fusion scheme, we adopt a binary particle swarm optimization (BPSO) procedure which allows the number of features to be significantly reduced while highlighting the difference between classes. This paper examines how the accuracy will be improved as several biometric data are integrated in an identification system. Results show a significant improvement in performance when classification performed at feature fusion level.