Contactless Palmprint Recognition Using Discriminant Multi-directional Feature Codes

Contactless palmprint recognition attracts researchers due to challenges stemming from pose and illumination, impacting palm line layouts and visibility. This paper introduces a robust technique, Discriminant Multi-Directional Feature Codes (DMDFC), for contactless palmprint recognition. DMDFC features compute the Multi-Directional Feature Pattern (MDFP), which includes information about the clearest and second clearest ridges, along with the count of ridges passing through the examined point. Moreover, DMDFC segregates ridge and wrinkle features using the MDFP of the Negative Ridge Direction Image (NRDI) and the MDFP of the Positive Ridge Direction Image (PRDI). These MDFP maps undergo \(\text {(2D)}^{2}\) LDA for dimensionality reduction, resulting in highly discriminative features. Experimental evaluations using public datasets from PolyU and IITD demonstrate the superior accuracy of the proposed approach, indicating substantial progress in contactless palmprint recognition across diverse applications.