Direct Template-Free Encryption Key Generation from Palm-Veins

Abstract

In this paper, an investigation into the feasibility of applying the palm-vein biometric modality within a template-free key generation framework was conducted. In partnership with iCognize GmbH, experiments were conducted on both the raw images and the encoded biometric templates. Although feature analysis suggested difficulties may ensue due to the significant overlapping between genuine and impostor distributions, it was found that the framework still performed well for dataset of 20 subjects. The experiments resulted in both the key reproducibility and key uniqueness rates achieving 100% with considerable effective key lengths. These results are highly encouraging and demonstrate that stable keys can be achieved from poorly separable features.