A Comparison of 3D Biometric Modalities

Recently, researchers have focused on the use of threedimensional data as a source of distinguishing features for personal identification. In this paper, the recognition performance of three three-dimensional biometric modalities (face, ear, and finger surface) is compared. In addition, we combine the modalities in an attempt to improve overall recognition performance. The Iterative Closest Point (ICP) algorithm and root mean square (RMS) registration error are used to measure match quality in each case. The experimental results using a dataset of multi-modal biometric samples collected from a group of 85 individuals are presented.