Multimodal Biometric System Integration of Face, Fingerprint and Signature

In this work, multimodal biometric system has been implemented by integrating signature, fingerprint and face templateswith very good accuracy rate. This paper analyses the performance of the proposed system under feature level fusion by integrating the extracted features from the biometric samples, score level fusion by fusing the match scores generated by the individual samples and decision level fusion with OR /AND logic. The signatures are acquired using WACOM digital pen tablet, fingerprints are collected using SecuGen Hamster fingerprint reader and webcam is used for acquiring the facial images. After pre processing of these samples, feature vectors from these templates are extracted using discrete cosine transform (DCT). These features are then examined first for their uni modal performances and then their combined performances under the three fusion schemes. Our experimental results on the image data set from 60 users confirm the usage of the proposed system in low security authentication as well as in high security authentication applications. On an average, the system needs around 0.08 seconds to authenticate a person under fusion schemes.