University of Houston
Department of Computer Science

In Partial Fulfillment of the Requirements for the Degree of
Master of Science
Mohammad Najam Murtuza

Will defend his thesis

Three-Dimensional Biometrics:
Using data from the nose and/or the ear

Abstract

Recently, the University of Houston’s Computational Biomedicine Lab has developed the UR3D framework for 3D face recognition that is based on an annotated deformable face model. UR3D currently claims the best results for 3D face recognition on the largest 3D face dataset (FRGC v2). The power of the UR3D framework is that it can be adapted for any 3D object. In this thesis, we focus on the information from nose and ear data. First, we present a method for nose tip detection which achieves accuracy of 99.0%. Second, we apply the UR3D framework to nose and ear data. Nose-based identification achieves a 92% rank-one recognition rate at the FRGC v.2 dataset. Ear-based identification achieves a 96.7% rank-one recognition rate on the University of Houston ear dataset and 92.7% on the University of Notre Dame ear dataset. Using 3D information for biometric authentication/identification paves the way for accurate, robust, and field deployable biometric systems.

Date: Monday, June 26, 2006
Time: 11:00 AM
Place: 218D-PGH

Faculty, students, and the general public are invited.
Advisor: Prof Ioannis Kakadiaris