Making Every Pixel Count


Professor Lawrence Carin of Duke University speaks as part of the ECE Distinguished Lecture Series.
Professor Lawrence Carin of Duke University speaks as part of the ECE Distinguished Lecture Series.

When it comes to image processing, electrical engineers are working to find ways to collect data in a manner that uses the least amount of energy and battery life. Improving the practice could enhance everything from security monitoring to MRI imaging.

Using compressive sensing, an information processing method that emerged about five years ago, a signal can be sampled and simultaneously compressed. This technique requires less energy and obtains enough data to still create an accurate replication.

Professor Lawrence Carin of Duke University and his research team make up some of the minds studying compressive sensing and how it can be used to improve image processing.

“The idea behind compressive sensing is that maybe we don’t need to measure as much data as we normally do,” said Carin.

Carin recently discussed their research at Boston University as part of the Department of Electrical & Computer Engineering Distinguished Lecture Series.

In contrast with compressive sensing, the Nyquist theory, a standard approach commonly taught to electrical engineering undergraduates, assumes that much more data is required in order to create a perfect reconstruction.

In collaboration with Professor David Brady of Duke, Carin designed a compressive hyperspectral camera to help test the effectiveness of compressive sensing. They hope that they will be able to demonstrate that high-quality images can be yielded with substantially less data than a regular camera.

“We’re redesigning a camera from the very beginning,” said Carin. “We’re hoping to show that downsizing data sampling won’t affect the quality of the image.”

Carin said that while the Nyquist theory still plays an important part in information processing, compressive sensing may be the solution to problems such as trying to get a MRI of a young child who can’t lie still for very long.

“Compressive sensing won’t replace the Nyquist theory, but it will enable us to collect data we couldn’t otherwise,” he said.

Carin’s talk was the first in the three-part Spring 2012 Distinguished Lecture Series. The next talk features Professor Mau-Chung Frank Chang of University of California, Los Angeles, who will speak on the topic, “Millimeter-Wave and Terahertz System-on-a-Chip for Radio, Radar and Imaging Systems.” Hear him on March 7, 2012, at 4 p.m. in PHO 211.

-Rachel Harrington (