Atlanta
(2003-10-28)
Emory Scientists Seek Perfect Pictures
A computer simulation shows “before” and “after” pictures of a satellite in space, using image restoration technology.
Image Courtesy: Emory University
Taking pictures has gained popularity, because it's become easier to become a good photographer.
Digital cameras and home printers make it difficult for the amateur photographer to take blurry or out-of-focus shots.
However, there are scientists all over the world who would like to see even more improvements in the field of imaging technology. Those who work with long-range telescopes or radiological machines see blurred pictures regularly.
Understandably, the people who have to interpret those pictures want clearer images.
Researchers at Emory University's School of Mathematics and Computer Science are working on solutions for scientists in the fields of radiology, astronomy and even crowd surveillance.
Emory's Dr. James Nagy says taking the type of clear pictures you expect to see out of a good camera is almost impossible for astronomers - at least right now - because a variety of elements can interfere.
“An astronomer using a ground-based telescope may take a picture of things in the sky, but the earth's atmosphere may actually cause those images to be distorted by some blurring,”, Dr. Nagy explained.
To correct that problem in everything from telescopes to microscopes Dr. Nagy and graduate students Lisa Perrone and Katrina Palmer spend hours writing novel computer programs.
When perfected, someone such as David Tyler might use them in his research for the Air Force.
Dr. Tyler is a research associate professor of optical sciences. He flies between the University of Arizona and the Maui Scientific Research Center in Hawaii to study images from deep space. He and scientists like him depend on the programs like the ones Nagy, Perrone and Palmer write to make sense out of the pictures retrieved from space.
Dr. Tyler says this kind of research could have procedure-changing consequences for researchers all over the world.
“A good example of that is, if you had a camera hung on a wall at a bus terminal and you have a crowd of people in the bus terminal and these people are all at different distances away from the camera. Some of them are going to be in focus and some are going to be out of focus,” Dr. Tyler said during a telephone interview.
“If we can apply some of the techniques my colleagues are working on, we can get all of these people - regardless of how far or near they are to the camera. And this certainly has some national security implications.”
Of course, what makes this type of security measure plausible isn't a high-tech camera, but the process that tells it what to do. That's why the work done by Dr. Nagy and graduate students Perorne and Palmer can become so potentially important.
“Our research is not to use these imaging devices, but to write computer programs to do the image enhancement techniques on the computer,” Dr. Nagy said.
“The process of doing that requires solving tens of thousand maybe millions of mathematical equations.”
Those equations are called algorithms. They're simply instructions for the computer. But in the case of making blurring picture crisp there's a problem. Noise.
To understand the concept of noise in a picture, tune an old television set to a channel it doesn't receive. That crackling sound and image is noise. When it interferes with an astronomy picture, noise can distort the image of a satellite to the point that no one can identify it.
Emory graduate student Lisa Perrone describes it as the most exasperating part of her research.
“It's the noise that causes the trouble,” Perrone said. “Just random radiation. Random light rays coming in from other sources. Not predictable, just out there.”
Scientist David Tyler of Arizona says mathematicians call this field of study inverse problem-solving. It's been around for nearly 30 years.
But only now have high performance computers and powerful imaging devices allowed theories to transfer from paper to real applications in the fields of physics, chemistry, biology and medicine.
“It requires a lot of creativity to try and undo this problem and backtrack our way through the distortion and blurring and noise contribution to get what's really there,” Tyler said.
“In the case of the medical image, people's lives may depend on this.”
Tyler says the reasons for distortion and noise in medical images is a little different, compared to what obscures pictures from space.
“If I have an object orbiting the earth it is well above, free and clear, of the distorting medium, which is the atmosphere,” Tyler said.
“The telescope is immersed in the atmosphere. When you're trying to image things inside the human body, then the object itself is inside the distorting medium and the detection device is usually free of that distorting medium.”
Emory graduate students Perrone and Palmer take two different approaches to eliminating noise and blurring from all types of high-tech images.
Perrone is building computer instructions that will allow an image to go from fuzzy to focused in one step. Palmer's research centers on a more repetitive process.
“So we kind of make a guess at what the image should look like,” Palmer said.
“We look at some mathematical things about it. If it isn't not good enough we do the process again, then compare it.”
Both students are on the cutting-edge of a field that could possibly allow them to set precedents for inverse problem-solving - or, at the very least, develop an algorithm process that will make even consumer cameras take better pictures.
© Copyright 2010, WABE
