| Author: | Rein-Lien Hsu |
| Advisor: | Dr. Anil K. Jain |
| Email: | hsureinl@cse.msu.edu; http://www.cse.msu.edu/ hsureinl |
Face recognition has been an active area in computer vision and biometrics mainly because it is a passive, non-intrusive technology. The challenge in face recognition has been the high degree = of variability on faces. Over the past 30 years face recognition systems have been devoted to overcoming varying faces and have started to gain its use in security monitoring systems. Besides, face images under investigation have shifted from a large = database of still images to video sequences. In this poster workshop, I present a model-based 3D face recognition system for locating, tracking, and verifying (and furthermore, = recognizing) human faces from videos. This system consists of two stages: learning and recognition. In the learning stage, a 3D face model of an individual is constructed from a generic face model. In the recognition stage, this system will verify a frame/sequence of face images, based on the learned 3D model and the estimates of pose and illustration in each test frame. The key strategy in this system--face modeling--plays a crucial role in recently emerging applications, such as video surveillance and human tracking as well as animation.
Return to Workshop Listing