The computer vision work is carried out primarily through the Computer Vision Laboratory at the Center for Automation Research. A brief description of some of the major projects follows.

Research and Development for Image Understanding Systems:

In this project, a system is being developed for monitoring aerial images. In particular, the system allows an Image Analyst to specify significant changes and to select appropriate image understanding algorithms for detecting these changes. The approach has been illustrated by developing a site model supported image monitoring algorithm which uses contextual information, camera and illuminant models to detect and count vehicles or construction activities at the given site.

Unmanned Ground Vehicle-Reconnaisance, Search and Target Acquisition:

This project is being conducted by a consortium led by the Univeristy of Maryland that includes the University Pennsylvania, the University of Rochester and the National Institute of Standards and Techonology. The goal of this research effort is to demonstrate automatic target recognition using sensors mounted on an unmanned vehicle navigating over rough terrain.

Watching People in Action

This research involves the development of computer vision algorithms for monitoring human activity in real time. A system has been developed for recognizing a wide variety of facial expressions and tested on a database of over 25,000 images. Current projects include studying the effects of facial expressions on mugshot recognition algorithms, and modeling and recovering human body actions (sitting, kneeling) from dynamic stereo.