Search

The highly random manner in which veins spread along a finger, their immunity to counterfeiting, active liveness, and user friendliness make finger veins the best choice for a biometric identification system (BIS). In this paper, veins of six fingers of two hands of a person are used to develop a secure, reliable, and robust multimodal BIS (MBIS). The main structure of the proposed MBIS is based on the effective combination of rank- and decision-level fusion. In the training step, the power (weight) of each single modality is estimated by extracting the information that lies in the cumulative match characteristic (CMC) curve. The testing step consists of two main parts. In the first part, the region of the finger vein is extracted by using a simple method, and then the binarized statistical image features (BSIFs) algorithm is used to extract feature vectors. In the second part, final decision for the test input probe is made by generating ' top rank-decision matrix', which fuses the information of each biometric identifier in the hybrid rank-decision level. The obtained results show that proposed method is more reliable and accurate than other fusion techniques at the post-classification fusion level. © 2017 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc. [ABSTRACT FROM AUTHOR]