Certain Habits

You know how in the movies and on television crime shows you’ll see a detective or intelligence officer in a lab looking at a blurry digital picture? And they’ll issue a command to “enhance it” or just “zoom in” or “resolve the image?”

That’s been science fiction. You can’t resolve detail that wasn’t captured in the first place. Zoom in closer and closer and you get … bigger pixels, not clearer pictures.

At least that’s what I’ve always thought. It turns out that a new field of study called “compressed sensing” may make that science fiction scientific reality.

Wired Magazine does a better job than I could of explaining the field. Here’s a taste to wet your appetite:

Compressed sensing was discovered by chance. In February 2004, Emmanuel Candès was messing around on his computer, looking at an image called the Shepp-Logan Phantom. The image — a standard picture used by computer scientists and engineers to test imaging algorithms — resembles a Close Encounters alien doing a quizzical eyebrow lift. Candès, then a professor at Caltech, now at Stanford, was experimenting with a badly corrupted version of the phantom meant to simulate the noisy, fuzzy images you get when an MRI isn’t given enough time to complete a scan. Candès thought a mathematical technique called l1 minimization might help clean up the streaks a bit. He pressed a key and the algorithm went to work.

Candès expected the phantom on his screen to get slightly cleaner. But then suddenly he saw it sharply defined and perfect in every detail — rendered, as though by magic, from the incomplete data. Weird, he thought. Impossible, in fact. “It was as if you gave me the first three digits of a 10-digit bank account number — and then I was able to guess the next seven,” he says. He tried rerunning the experiment on different kinds of phantom images; they resolved perfectly every time.