Relative Statistical Ensembles of Curved Gabor Wavelets for Palmprint Verification

Abstract

Palmlines are the inherent features of the palm-print which demonstrate curvature in itself. The straight Gabor wavelets have been used to get statistical and binary representation of the palmprint. But, the employment of the straight Gabor wavelets may not be suitable for representing the curvature of the palmlines. This motivated us to investigate the curved Gabor wavelets for feature extraction. Therefore, this work employs the relative statistical ensembles “Relative statistical variances” of the transformed palmprint images using curved Gabor wavelets for palmprint representation. These variances are computed at different layers which are generated through partitioning the transformed palmprint image plane. The experimental results confirm that the proposed relative statistical ensembles has good discrimination capability.