Fusion of multimodal biometric authentication using gradient pyramid, PCA and DWT

Authentication and identification is the most challenging task in our
daily life. Biometric system provides an automatic identification of an
individual using his/her behavioural or physiological traits. In this work,
multimodal biometric traits namely fingerprint and iris, have been used. These
traits were pre-processed using Wiener filter and applying some morphological
operations. The pre-processed biometric traits were segmented and fused using
three algorithms namely discrete wavelet transform (DWT), principal
component analysis (PCA) and gradient pyramid (GP). The fused biometric
traits using GP provides a better result without losing the meaningful
information. The feature extraction and classification were carried out using
grey scale co-occurrence matrices (GLCM) and support vector machine
(SVM). Authentication using fused biometric traits gives accuracy as 83.75,
whereas the accuracy using fingerprint 73.75% and iris was 78.48%.