Ma aims to produce high-resolution images with low-quality instruments
CSL researcher Yi Ma is studying how to get the most out of very little.
The Knowledge Enhanced Compressive Measurement (KECoM) program is a three year grant of $293,960 sponsored by DARPA, the Defense Advanced Research Projects Agency.
The goal is to get high quality information without a high quality instrument -- such as a high-resolution photo taken by a traditional camera.
“If you want to take a photo you use a high-resolution camera for better quality,” said Ma, an associate professor of electrical and computer engineering. “The key thing is, in reality, you may not have the luxury of so many pixels. Sensors are expensive and the traditional camera is cheaper. How can you still obtain a high resolution photo and take as few measurements as possible?”
Through the KECoM project, Ma hopes to develop a sensor that will combine prior task knowledge along with context clues in order to take fewer measurements and that are still accurate. This would achieve a high performance, low-cost defense mechanism. Compressive measurements are also key to reconstructing images using far fewer measurements.
“You want to use less physical measurements to recover signals 100 times more accurate,” Ma said.
Ma’s primary responsibility is to develop algorithms for compressed sensing and to test them.
“It does take a long time to achieve these goals so this is very ambitious, but I hope we will be there by the three years end,” Ma said.
Since the project is multi-faceted, Ma will be collaborating with HRL Laboratories, the University of California, Los Angeles, Duke University and the University of California, Merced.
“This is certainly a technological breakthrough for defense,” Ma said.