Personal Verification for Mobile Devices
Haibo Li, Department of Applied Physics and Electronics

The fast growing public ownership and use of mobile devices, e.g. mobile phones and PDA's, creates new and exciting opportunities for convenient and secure commercial and service providing transactions. The newer generations of these devices incorporate digital cameras, microphones and signature pads, raising the possibility of using biometric-based authentication to protect mobile transactions against fraud and repudiation. The goal of this project is to develop reliable personal verification techniques for mobile devices to perform on-the-spot verification. Automatic personal verification has attracted increased interest recently. A popular approach is to employ computer vision techniques to measure personally physiological signatures, like face, iris, hand, fingerprint etc. It is very challenging to perform personal verification based on the features captured by the video camera embedded in mobile devices. This is mainly due to a wide variety of imaging problems, e.g. lighting, shadows, scale, and translation plague the attempt for unconstrained feature identification. In addition to the imaging problems, it is inherently difficult to collect consistent features from the body. Taking human face as example, face is arguably the most changing part of the body due to e.g. facial expressions, cosmetics, facial hair and hair styling. Moreover, a mobile device normally has very limited processing power and memory size, which greatly limits the possibility of adopting advanced computer vision techniques. In this project the research focus will be on the fundamental issues: 1. Choose the most effective biometric signature for mobile personal verification. 2. Study the theoretical limits of verification performance for the chosen biometric signature. 3. Establish statistical models and theories for system performance comparisons and performance predictions " Methods for summarizing large amounts of data into a small set of numbers reflecting the statistical essence of the data " Rank order statistics versus trade-off errors (receiver operating curves) " Design and acquisition of biometric databases (sampling design) A genuine interest in biometrics is required. The graduate student should preferably have a good background in image processing, pattern recognition, computer vision and security system.