Wavelet Feature Extraction and ANOVA based Multimodal Biometric Fusion System

We live in a world today in which technology moves at a very rapid pace. Many of these technological advances can be used to make our everyday life easier and safer. One such technology which is very popular these days and is in extensive use is biometrics system. A biometric system attempts to confirm an individual’s claimed identity by comparing a submitted sample to one or more previously enrolled templates. But there are few limitations associated with unimodal biometric system such as noise in sensed data, non-universality, spoof attacks etc. So it is possible to overcome these limitation which gives rise to multimodal biometric system. In this work, the biometrics traits under consideration are: human iris, fingerprint and face. The reviewed papers propose methods that employ PCA, fisher face projection, minutia extraction and LBP feature extraction on different biometric traits. However these techniques contribute to complexity and accuracy of the entire system has scope for improvement. With advances in digital image processing methods, pre-processing of each unimodal system sample and techniques such as wavelet based feature extraction and ANOVA deals with the existing system in a way making it more secure. It can be gauged by parameters such as FRR, EER, FAR. Thus, the complexity and accuracy gaps between the reviewed and proposed system becomes substantially smaller. This paper is about the effectiveness of multimodal biometric fusion system and presents tools employed to enhance it which is depicted by ROC curve.