A finger vein authentication system based on pyramid histograms and binary pattern of phase congruency

Finger vein recognition is a novel biometric method which is getting more attention recently because of its high security and convenience. However, there are still many interference while recognizing, such as uneven illumination caused by non-uniform near-infrared light source or image dislocation caused by finger position change. Therefore, this paper proposes a finger vein feature extraction strategy integrating binary pattern of phase congruency (BPPC) and pyramid of histograms of orientation gradients (PHOG). We first calculate the energy information of each part in different regions and represents it with binary pattern, and then analyze the shape and texture of finger vein with image pyramid of histogram, and finally fuses them to obtain better recognition efficiency. Our method considers the multi-scale features of finger vein images so that some local disturbance like overexposure of light source or insufficient illumination can be ignored, and our method is suitable for most environments and finger vein collection equipment. We experimented on both public database and database built by our lab with recognition rate more than 98%, which confirmed that our method is effective and has good practical significance.