Optimal Local Direction Binary pattern based palmprint recognition

This paper presents a novel palmprint recognition scheme which utilizes directional representation of Anisotropic Filters (AFs) optimally followed by Local Binary Pattern (LBP) called Optimal Local direction Binary patterns (OLdirBP). Optimal orientation image from different orientation can be excavated by opting orientation corresponding to maximum of summed value of convoluted responses between local directional patterns of size 3×3 and AF filtered responses, then Micro pattern representation is obtained by LBP descriptor. New OLdirBP scheme enhances robustness to illumination variations and extort discriminative information from palmprint along different orientations with low computational complexity. Experimental results demonstrates efficacy of the proposed scheme and achieves EER 0.907% and GAR 99.13 %.