Palm Vein Age and Gender Estimation Using Center Symmetric-Local Binary Pattern

Abstract

Many investigations on hand veins modality have been done in the literature for identification and recognition systems. However, researches on age and gender estimation by hand veins are very limited and very preliminary. Our contribution in this paper is to propose a system able to estimate the age and the gender of a person from its hand veins. Accordingly, we are interested in studying the discriminating features for the prediction of a person’s age and gender. In fact, hand vein images are very rich in orientation and contour characteristics and they are faced with poor quality and illumination variation. Hence, we investigate texture analysis invariant to illumination as well as venous pattern gradient information determination by Center Symmetric-Local Binary Patterns (CSLBP) descriptor. Since Region Of Interest (ROI) extraction is important in a biometric system, we aim to cover the whole informative region of hand veins by our dynamic ROI extraction method. Our experimental study is based on palm vein VERA database. As considered database has a class imbalance problem, we remediate this problem by using Weighted K-Nearest Neighbor (WKNN). The obtained performance metrics demonstrate the effectiveness of our proposed system for gender classification and age estimation respectively: 95.8% and 94.2% for F-measure, 95.9% and 94.4% for G-mean.