Mobile Wrist Vein Authentication Using SIFT Features

Mobile devices store sensitive and private data which has to be secured. To protect this data most of these devices implement authentication mechanisms like PIN, password, or unlock pattern. However, these approaches can be problematic in terms of usability and security. Users do not want to remember multiple and difficult authentication secrets, hence they tend to use easy and short secrets which are vulnerable to be captured and replayed by attackers. In recent years biometrics have become important for authentication on modern mobile devices. Thereby, different biometrics do not have to be remembered by users and are differently hard to observe by attackers. For example, veins used in vein pattern authentication remain hidden when not using specialized hardware. In this work we present a low cost mobile wrist vein authentication system based on Scale-Invariant Feature Transform (SIFT). We implement a low cost vein capturing sensor using near-infrared (NIR) illumination and a filter modified camera. For authentication we present an image preprocessing methodology and an image matching algorithm based on SIFT features. In parallel, using the proposed sensor we build up a self-recorded wrist vein database which contains 120 wrist vein images. Furthermore, we develop six different authentication decision models using 1 or 4 samples for enrollment and authentication. Then, we evaluate their performance using the self-recorded database and compare them with other existing vein authentication works. Concluding, our results indicate that the presented system using the proposed capturing sensor and SIFT features algorithm using 4 samples for enrollment is a viable approach for mobile wrist vein authentication.