Appearance based modeling and learning of the human face with application to biometrics

Xu, Xun

Permalink

https://hdl.handle.net/2142/16531 Copy

Description

Title

Appearance based modeling and learning of the human face with application to biometrics

Author(s)

Xu, Xun

Issue Date

2010-06-29T00:59:54Z

Director of Research (if dissertation) or Advisor (if thesis)

Huang, Thomas S.

Doctoral Committee Chair(s)

Huang, Thomas S.

Committee Member(s)

• Ahuja, Narendra
• Levinson, Stephen E.
• Liang, Zhi-Pei

Department of Study

Electrical & Computer Eng

Discipline

Electrical & Computer Engr

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• computer vision
• pattern recognition
• machine learning
• face recognition
• face alignment
• biometrics

Abstract

In this dissertation we study key problems in face processing, with a focus on the applications in biometrics, including both hard biometrics (i.e. conventional face recognition) and soft biometrics, where demographical attributes (e.g. gender and age) are recognized. We categorize face processing techniques into two classes: appearance-based approaches that model facial appearance holistically, and feature-based approaches, which rely on the localization of facial feature points. In this dissertation we argue that appearance-based approaches are well suited for various face processing tasks. A fully automatic face processing system consists of several major modules: detection, alignment and recognition. In this dissertation we study the modeling and learning issues in the face alignment and recognition stages, and propose a series of algorithms to tackle these problems, demonstrating how recent developments in machine learning and applied mathematics can be employed to create effective solutions in building fully automatic, appearance-based face processing systems. Although face detection is not covered, as it is a relatively well-solved problem, some algorithms proposed in this dissertation can potentially be applied to that problem as well.

Graduation Semester

2010-5

Permalink

http://hdl.handle.net/2142/16531

Copyright and License Information

Copyright 2010 Xun Xu

Owning Collections