Comparison of multidirectional representations for multispectral palmprint recognition

Abstract

Palmprint is emerging as a new multi-modal biometric for human recognition. Multispectral palmprint images captured in the visible and infrared spectrum not only contain the superficial structure of a palm, but also the underlying structure of veins; making them a highly discriminating person identifier. This study comparatively analyzes multidirectional representations for multispectral palmprint recognition which show promising results. Comprehensive experiments for both identification and verification scenarios are performed on three public datasets. The accuracies of state-of-the-art clearly indicate the viability of multidirectional coding methods for multispectral palmprint recognition.