Ellen C. Hildreth
Panagiotis Metaxas
Randy Shull
Go to the Computer Science Department Home Page
My research interests lie in the study of human and computer vision, using an interdisciplinary approach that integrates computational studies of visual processing with experimental observations from perceptual psychology and the neurosciences. My past research has addressed the problems of edge detection, the measurement of motion, detection of motion boundaries and the recovery of three-dimensional structure from motion. My current interests focus on the integration of the recovery of structure from motion with the reconstruction of a complete 3-D surface representation, the recovery of the 3-D motion of an observer from the changing image, and the integration of binocular stereopsis with the analysis of visual motion.
In my current research, I have developed a computational model for recovering the direction of heading of an observer who is moving relative to a scene that may contain self-moving objects. The model uses velocity differences computed in image regions of high depth variation to locate the focus of expansion of the image motion field, which indicates the observer's heading direction. The model exhibits behavior similar to that observed in experimental studies of the ability of human observers to judge heading direction.
With regard to the recovery of 3-D structure from motion, perceptual studies suggest that the interpolation of a complete 3-D surface representa- tion may play an important role in the underlying mechanisms. In a collaborative effort with R. Andersen, H. Ando and S. Treue at MIT, we have developed a model that combines an algorithm for recovering 3-D structure from the motion of image features over an extended time, with a process that interpolates a smooth surface over these features. The model allows multiple surfaces to be represented in a given viewing direction, in order to interpret transparent surfaces. The model exhibits interesting behaviors similar to that observed for the human visual system by Ramachandran, Andersen and others.
Return to Computer Science Faculty List
Parallel Computation:
Despite the always-growing human "appetite" for more computational power, our ability to create faster machines containing only one processor is limited by many, difficult to overcome factors (such as the speed of light). The obvious (and only) alternative to increase the power of our computers is to create machines containing many (hundreds or even thousands) processors.
Such machines (called parallel machines) can use more than one processor to solve a problem: Each processor solves small parts of the problem and communicates with the rest to produce the global solution.
Parallel Computation is the emerging area that explores solutions in that direction. In this young and promising field of computer science the theoretical and implementation issues offering good opportunities for student projects are unlimited.
I am interested in every aspect of parallel computation, but more specifically in the design and implementation of parallel algorithms.
Multimedia and Algorithm Visualization:
If "a picture is worth a thousand words" (as they usually say), an animation
is probably worth a million.
Algorithm animation for education (one of the visualization areas) tries to achieve a better and faster understanding of the algorithm being described, through a carefully chosen and aesthetically pleasing representation. This representation involves abstract mathematical objects, data structures, and the algorithmic steps being executed. The use of sound, video, hypertext and graphics in the animation (called multimedia) greatly helps in this process. The computer screen becomes an interactive video. It is an area between science and art!
Student Projects:
There are many exciting projects in the above described areas. In parallel computation typical projects involve implementation of efficient algorithms on machines containing up to 65,000 processors! Working with multimedia, you can create an animation of your favorite algorithm using user-friendly tools on the Macintosh or on an X-window environment. And, of course, you can always combine the areas by animating parallel algorithms.
Return to Computer Science Faculty List
Randy Shull - Design of Algorithms, combinational optization
Return to Computer Science Faculty List