HKBU  |  SCI  |  BUniPort  |  Library  |  Alumni  |  Job Vacancies  |  Intranet  |  Sitemap        
Undergraduate Admissions
Taught Postgraduate Admissions
Research Postgraduate Admissions
Job Vacancies
News & Achievements
Research Highlights
Contact & Direction

Department of Computer Science Seminar
2008 Series

High Resolution Acquisition, Tracking and Animation of Dynamic Facial Expressions

Dr. Yang Wang
Robotics Institute of Carnegie Mellon University

Date: January 30, 2008 (Wednesday)
Time: 2:30 - 3:30 pm
Venue: SCT909, Cha Chi Ming Science Tower, Ho Sin Hang Campus

Generating realistic facial animations is important in many computer graphics settings such as games and movies. Recent advent of new technologies allows us to capture massive amounts of high resolution, high frame rate geometry and appearance data. In order to use such data for the temporal analysis and synthesis of subtle dynamic motion, such as facial expressions, an efficient object registration and tracking algorithm is needed. This problem remains challenging for non-rigid objects (e.g., faces) and is particularly difficult in the presence of unseen appearance variation. In the first part of my talk, I will present our new approach for 2D face image alignment and tracking based on support vector machines (SVMs) and constrained local models (CLMs). A major advantage of our approach over conventional methods such as active appearance models (AAMs) is the ability to generalize well to unseen subjects and offer greater invariance to illumination changes. In the second part of my talk, I will present my work on 3D facial motion analysis. To this end, we develop a real-time structured-light ranging system to capture moving objects without the use of markers. Further, we propose an efficient algorithm to estimate non-rigid 3D motion based on multi-resolution 3D deformable models. A unifying framework is developed to connect 2D and 3D based registration approaches using Riemannian geometry. Due to the strong implicit and explicit smoothness constraints imposed by the algorithm and the high-resolution data, we can establish dense inter-frame correspondences even for subtle facial expressions. Finally, I will demonstrate several graphics applications where we use the high quality motion data to transfer facial expressions from a source face to a target face as well as to synthesize completely novel expressions.

Dr. Yang Wang is a postdoctoral fellow in the Robotics Institute of Carnegie Mellon University, working with Professors Takeo Kanade and Jeffrey F. Cohn and Dr. Simon Lucey. He received his Ph.D. degree from Stony Brook University in 2006 and his master and bachelor degrees from Tsinghua University in 2000 and 1998, respectively. He specializes in non-rigid motion tracking, facial expression analysis and synthesis, and illumination modeling. He is a member of ACM and IEEE.

********* ALL INTERESTED ARE WELCOME ***********
(For enquiry, please contact Computer Science Department at 3411 2385)
Copyright © 2021. All rights reserved.Privacy Policy
Department of Computer Science, Hong Kong Baptist University
Hong Kong Baptist University