A Novel Rank-Level Fusion for Multispectral Palmprint Identification System

Multispectral imaging has been effectively used in the fields of computer vision and human verification to analyze information captured from several bands of the electromagnetic spectrum recently. In this paper, a novel Gabor-based palm print identification algorithm on multispectral images is proposed. Various features are extracted from convolving the Gabor kernels with each image sample. Logistic regression method is used to fuse obtained ranks of each spectral band authentication. The proposed algorithm is evaluated on PolyU database which contains palm print images from 400 individuals from each band. The obtained results show robustness of our algorithm in comparison with other works that presented in the literature.