ARCHIVES

The use of biometric identification methods is nowadays inevitable for the sake of securing information however, the existing approaches, which often rely solely on one trait, tend to be vulnerable to certain factors, such as environmental interferences or deliberate attempts at deception. To resolve this issue, this research paper proposes an approach that involves multi-modal biometric authentication through a parallel feature extraction scheme which is particularly intended for ear contour and palmprint identification. This approach is different from the current practices because there is a single algorithm for identifying both features; however, this system takes into consideration the distinctiveness of each feature. Meanwhile, the algorithm for palmprint identification identifies the distinctive lines present on the palm using the help of Gabor filters and Wavelet transformations prior to applying PCA. The independent generation of feature sets from both the systems is then combined at the level of feature sets. The classification process uses the Mahalanobis distance to measure variance. Whereas the individual systems gave 93.3% and 91.8% accuracy levels, their combination gave an impressive 99.9% peak accuracy level. The use of this method results in an efficient, compact, and accurate system without the use of deep neural networks.