Decision Level Fusion Based Multimodal Biometric System

Systems that use unimodal biometrics tend to have less accuracy, variations that are due to intra class, restricted degree of freedom, non-universality, error rates that are not acceptable, etc. Based on these reasons, in the near past, most of the researchers have been concentrating on multimodal biometrics. Multimodal biometrics integrates various types of biometrics which outperform unimodal biometric systems. In this paper, it is proposed to combine Iris and Fingerprint biometric systems and try to exploit the use of such combination. Different features viz., texture, local binary pattern, are extracted from Iris and Fingerprint images of each class and combined to form the final feature vector. Various fusion techniques exist and in this work we concentrate on fusing the classifier results and then make an efficient decision out of that fusion. Results show that the proposed technique is outperformed existing techniques and an accuracy of 99%