8:23 a.m., May 6, 2009----Imagine an imaging system that would allow a surgeon to see an organ from all angles before making the first cut with a scalpel, or a law enforcement agent to determine not only the speed of a car as it races through a red light but also the number on the license plate.

THIS STORY

E-mail

Print

Twitter

With support from a National Science Foundation Faculty Early Career Development Award, Jingyi Yu, assistant professor in the Department of Computer and Information Sciences at the University of Delaware, will study multi-perspective imaging, rendering, and projection -- techniques that have the potential to dramatically change how images are captured, delivered and viewed by the human eye.

The technology, which combines computer graphics, vision and photography, yields images similar to those created by artists in the cubist movement, where objects are depicted from multiple viewpoints to show the subject in a broader context.

The goal of the five-year, $400,000 research project is to develop a complete framework to characterize and design new multi-perspective cameras and displays and to use these systems for computer vision and graphics applications.

“The multi-perspective camera combines what is seen from several viewpoints into a single image and provides a potentially advantageous imaging system for understanding the structure of observed scenes,” Yu says. “But most existing computer vision and graphics algorithms are not directly applicable to multi-perspective cameras. This grant will enable us to work toward closing that gap.”

Yu explains that any of the three key components of a camera -- the flash, the shutter or the lens -- can be modified to enhance the images it records. This work focuses on the lens and aperture system.

“A typical pinhole camera works like the human eye,” Yu says, “capturing what's known as a perspective image. In contrast, a multi-perspective camera works more like a fly's vision system, which includes many small eyes that work together to yield a single multi-perspective image from a number of individual images.”

Yu's research will focus on not only development of the camera itself, but also on its use for 3D reconstruction and on generation of a multi-perspective projection system. The latter would enable omni-directional viewing of a scene.

“For example,” Yu says, “the Google Earth application now enables the user to view a street scene through a sequence of single-perspective images gathered by a conventional camera. With multi-perspective technology, we would be able to view that same scene in a panoramic way.” The projection technology would also enable individuals to create mini IMAX-type theaters in their homes.

Another potential application for Yu's work is the area of biometrics, where human physical and behavioral traits are used as a means of secure identification. Through the development of more sophisticated cameras as well as improved rendering techniques, Yu is working to generate facial images without the “noise” that impairs such images when traditional technology is used.

Yu views himself as straddling the line between the science and engineering of imaging, as his work involves both hardware development and the generation of algorithms to run the hardware and optimize the images it generates.

His new NSF project will include a strong educational component by involving undergraduate and graduate students in self-contained projects with both theory and application components, as well as attracting under-represented students through a pre-engineering program involving Delaware State University.

Yu joined the University of Delaware faculty in 2005 after earning his doctorate at the Massachusetts Institute of Technology. In 2007, he won a UD Exemplary Use of Technology in Teaching Award for using emerging interactive technologies such as TabletPCs and Wiimotes in his computer graphics course (CISC 440/640).

Article by Diane Kukich
Photo by Ambre Alexander