Automatic Person Verification Using Speech and Face Information

Interest in biometric based identification and verification systems has increased considerably over the last decade. As an example, the shortcomings of security systems based on passwords can be addressed through the supplemental use of biometric systems based on speech signals, face images or fingerprints. Biometric recognition can also be applied to other areas, such as passport control (immigration checkpoints forensic work (to determine whether a biometric sample belongs to a suspect) and law enforcement applications (e.g. surveillance). While biometric systems based on face images and/or speech signals can be useful, their performance can degrade in the presence of challenging conditions. In face based systems this can be in the form of a change in the illumination direction and/or face pose variations. Multi-modal systems use more than one biometric at the same time. This is done for two main reasons — to achieve better robustness and to increase discrimination power. This thesis reviews relevant backgrounds in speech and face processing, as well as information fusion. It reports research aimed at increasing the robustness of single- and multi-modal biometric identity verification systems. In particular, it addresses the illumination and pose variation problems in face recognition, as well as the challenge of effectively fusing information from multiple modalities under non-ideal conditions.