ProjectHTML

CURRENT RESEARCH PROJECTS

	2D->3D Face Model Construction We developed a linear algorithm that uniquely recovers the 3D non-rigid shapes and poses of a human face from a 2D monocular video. Details
	3D Head Motion Recovery in Real Time We developed a cylindrical model-based algorithm for recovering the full motion (3D rotations and 3D translations) of the head in real time. Details
	Face Model Building and Fitting Techniques for building and fitting 2D and 3D models of human faces and heads. Details
	Real-Time AAM Fitting Algorithms Fast algorithms for fitting Cootes and Taylor's "Active Appearance Models." Details
	Facial Expression Analysis Automatic facial expression encoding, extraction and recognition, and expression intensity estimation for the applications of MPEG4 application: teleconferencing, human-computer interaction/interface. Details
	Online Expression Control We are developing methods for real-time performance-driven animation and control of three-dimensional facial expressions using a single video camera. Details
	Human Identification at a Distance We are developing and evaluating human identification technologies as part of the Defense Advanced Research Projects Agency (DARPA) sponsored program in Human Identification at a Distance (HumanID). Details
	Face and Facial Feature Tracking Rigid Tracking of Faces and Non-Rigid Tracking of Facial Features. Details

PAST RESEARCH PROJECTS

	Feature Grouping and Object Detection We developed an algorithm that groups edge feature segments based on Local and Global Grouping Factors and extracts elongate curve-like objects from images, such as roads and bridges from satellite images and blood vessels from medical images. Larger figure
	Image Segmentation We developed a nonparametric region competition algorithm that combines scale-space clustering and region competition for image segmentation. We also presented a formal and general procedure to automatically localize the initial segmentation seeds for region competition. Larger figure
	Enhancement of Far Infrared Image Sequences According to the far infrared imaging process, an ideal far infrared image would be piecewise constant. We design an adaptive spatiotemporal homomorphic filter (ASTHF) that efficiently enhances far infrared image sequences by simultaneously smoothing the inner regions of the objects and preserving their boundaries sharp. Larger figure

SELECTED COURSE PROJECTS

	Single View Modeling The project was to create 3D texture-mapped models from a single image using the method described in "Single View Metrology," by A. Criminisi, I. Reid, and A. Zisserman, ICCV 1999. The tasks consisted of image acquisition, calculating vanishing points, choosing reference points, computing textures and 3-D positions, and creating a VRML model. Details
	Video Texture Synthesis Given a short sample video of certain texture, such as flame and ocean wave, we developed a method to synthesize new videos of the perceptually similar texture. This approach treats a video as a 3D volume image and extend the 2D texture synthesis method described in "Texture Synthesis by Non-parametric Sampling," by A. A. Efros and T. K. Leung, ICCV 1999, to work on the 3D volumn data. Details
	View Transformation The task of this project is to acquire a range scan of a face and to render it from different camera viewpoints, using the image warping and visibility technique described in "Head-Tracked Stereoscopic Display Using Image Warping," by L. McMillan and G. Bishop, Stereoscopic Displays and Virtual Reality Systems II, SPIE, 1995. Details