Hybridization of discrete binary particle swarm optimization and invariant moments for dorsal hand vein feature selection

Nowadays, hand vein recognition is amongst the most recent biometric technologies used for the identification/authentication of individuals. Indeed, hand veins biometric are robust and steady human authentication unlike to other biometric technologies as fingerprint, face, signature and voice. In the present work, the proposed system consists of image preprocessing, feature extraction and identification. This paper outlines a novel approach for identification, based on seven Hu´s invariant moments that are extracted from the vein images as feature representation, due to its invariant features on image translation, scaling and rotation. However, they are sensitive to noise. Therefore, discrete binary particle swarm optimization (PSO) is applied in solving a problem of optimization; for selecting optimal features of Hu´s invariant moments that minimize false accept rate (FAR) and false reject rate (FRR).The experimental results carried out on 102 users show that the discrete binary PSO-Invariant Moments improve the performance of our biometric system with FAR= 0% and FRR=0%, with fewer number of features and threshold of 72%.