Vision-Based Facial Analysis for Human-Computer Intelligent Interface
The human face is a powerful mean of
communication. We use it constantly, naturally and effortlessly not only
to identify one from another, but also to transmit information about our
feelings. Such natural style of communication with machines is the
ultimate goal in Human-Computer Intelligent Interface.
The overall goal and motivation of this research line is to provide
the computers with the vision-based ability to detect, identify, and
understand faces and facial expressions so that a very natural
communication channel between humans and computers can be
established. Computer vision algorithms for facial analysis also find
application in surveillance systems, video games and entertainment,
virtual agents, video conferencing, etc.
The figure above illustrates the overview of the vision-based facial analysis system we have developed. Most parts of this system have been implemented in real-time and integrated to a multimedia, multi-modal, web-based Computer Travel Companion.
More specifically, our research on facial analysis includes algorithms for:
1. Information-Based Maximum Discrimination Learning.
2. Face Detection in Complex Background. ( Online Demo )
4. Real-Time Tracking of Faces and Facial Features of Multiple People. (MPEG videos)
5. Gesture Analysis Based on Head Position.
6. Feature-Based Head Pose Tracking.
7. Embeded Face and Facial Expression Recognition. (MPEG videos)