MICRO ENGINEERING A MACRO IMAGE
LSU engineers devising new hardware and software for image capture and processing

Materials smaller than the width of a human hair can be used to modify images at a scale of one five-thousandth of an inch to create an improved, high-definition reproduction. At least, that is the goal of a pair of LSU engineers who are paving the way to the next generation of imaging systems for use in digital cameras, surveillance equipment, medical scanners, and satellites.

Jin-Woo Choi and Bahadir Gunturk, both assistant professors in LSU’s Department of Electrical & Computer Engineering, are harnessing the powers of hardware and software to create an imaging system that can take multiple images of the same scene and combine the best elements of those images to create a high-definition image. Inside the camera, resolution and dynamic range—the range between the lightest and darkest colors in the image—could be dramatically improved over today’s standards.

“Most consumers today use computer software programs to adjust their photos to get more detail out of them,” says Gunturk. “Our new system will hopefully eliminate some of that work to create a vastly improved image.”

To begin, Choi is leading the hardware development aspect of the project. To allow the camera to take four different images of the same scene, the image sensor must move. So, Choi is developing micro-actuators that can slightly and accurately move the sensor horizontally and vertically along the image plane. Micro-actuators are made of special materials that are typically less than the width of a human hair. They expand and contract when an electrical voltage is applied to them, and Choi is aiming to create new micro-actuators with that property. The sensor could then take different shots of the same scene.

Gunturk is responsible for creating the high-definition copy from these images using a concoction of mathematics and image-acquisition modeling. To accomplish that, he is writing various algorithms—a series of well-defined mathematical instructions for the camera to follow in constructing the final image, much like a cook following a recipe to create a culinary masterpiece.

The smallest unit of any image is a pixel, which represents a single color and can vary in size depending on the image resolution. The more pixels per inch in an image, the more colors and detail the image can have. Each pixel would be converted into a signal for the camera to read and, based on that signal, decide whether or not to use that pixel in the final image it constructs. Ultimately, the camera will combine all of the common elements for each of the images taken and use the algorithm programming to fill in any missing data.

Their research has tremendous possibilities for future application. Outside of digital photography, it could be used to improve the power of surveillance equipment and allow better zoom on license plates and faces of criminals, for example. In medicine, it could allow for more precise identification and location of tumors. To Choi and Gunturk, these micro-actuators and tiny pixels can make a big difference.

ON THE WEB:
LSU Department of Electrical & Computer Engineering

from Winter 2006 Issue