A novel MF-LDTP approach for contactless palm vein recognition

In this paper, we propose a new method for contactless palm vein recognition with the help of texture based LDTP descriptor. Compared to existing methods, Local Directional Texture Pattern extracts both directional as well as contrast information. Improved foreground based LDTP descriptor is adopted in order to increase the efficiency. We first extract region of interest (ROI) from CASIA multispectral palmprint images. Then, extracted ROI is filtered and subtracted from background followed by binarization which helps to suppress noise and improve efficiency. The rigorous experimental results shows the superiority of our method compared to previous works.