The megapixel wars are so overrated–at least according to a team of researchers at Rice University. By focusing light on a single-pixel sensor, they promise reduced power consumption & consequently greater battery life in digital cameras. The digital micromirror device, says the BBC, "consists of a million or more tiny mirrors each the size of a bacterium." As the light passes through the device, the millions of tiny mirrors are turned on and off at random in rapid succession.
Photoshop engineer Zalman Stern points out the researchers’ info on compressive imaging & writes,
The design uses a micromirror array and a lens to perform a pseudorandom
weighting of the image. The result is
sampled using a single photo detector. The image presented to the
micromirror array is from a standard lens system of some sort.
The interesting part is the math underlying the reconstruction from the
samples. There is recent theory work that determines how good a
reconstruction you can get for a given amount of sampling reduction.
That is, one takes significantly fewer samples than the number of pixels
in the output image and gets a moderately acceptable rendition of the
original scene. One way they have of looking at this is that image
compression is done during sampling, rather than digitally afterwards.
The device is currently the size of a suitcase, so getting it into practical applications is likely to take some time. [Via Kevin Tieskoetter]
* For those interested in these things: Zalman was on the Photoshop team way back in the day (doing the port to PowerPC, as well as the ye olde GIF
89a Export plug-in). After that he left, joined Macromedia, then left and started a company, then found his way back to Macromedia, and thus to Adobe. (Kind of a nice pallindrome…) Now he’s contributing code to Camera Raw that’s "rockin’ like Dokken." He was a creator of Contribute, which I’m using to type this now. It’s a small world, after all.