An Algorithm for Feature Level Fusion in Multimodal Biometric System

The increasing demand for high secure and reliable authentication schemes, led to improvement in unimodal biometric system and hence multimodal biometric system has emerged as a mean of more secure and reliable authentication scheme. Multimodal biometric fusion is very promising process to enhances the strengths and reduce the weaknesses of the individual measurements. In multimodal biometrics, the classification results obtained from each independent biometric channel is fused to obtain composite classification is known as biometric fusion. The appropriate fusion scheme combines the processed information, which can be further used for authentication. In this paper, I propose an ensemble algorithm for feature level fusion in multimodal biometric system. The proposed feature level fusion algorithm is accomplished by normalizing feature point sets extracted independently from two modalities of user and making the two point sets compatible for concatenation and then performing feature selection on the concatenated vector.