Researchers Developing Single-Pixel Camera

Assassin bug writes "According to the BBC, researchers in the US are developing a single-pixel camera to capture high-quality images without the 'expense' of traditional digital photography. The idea behind such a device is that traditional digital photography is wasteful. Most of the information taken in by the camera is thrown away in the compression process. From the article: 'The digital micromirror device, as it is known, consists of a million or more tiny mirrors each the size of a bacterium. "From that mirror array, we then focus the light through a second lens on to one single photo-detector - a single pixel." As the light passes through the device, the millions of tiny mirrors are turned on and off at random in rapid succession. Complex mathematics then interprets the signals assembling a high resolution image from the thousands of sequential single-pixel snapshots. '"

Comments

RAW format anyone?

[Anonymous Coward]

> Most of the information taken in by the camera is thrown away in the compression process.

Doesn't the RAW format take care of this?

Re:RAW format anyone?

[John Meacham]

The problem is not getting at that extra information, like you say, we can already do that with RAW. the problem is that a lot of resources (such as CCD area) go into capturing this extra information which is then simply discarded. By taking a random sampling of pixels, one gets exactly as much information is needed to construct the compressed version of the image without waste. plus, with only a single CCD, you can make it incredibly sensitive, to the point where it can count single photons. Heck, you could probably have some fun with wavelengths. different wavelengths get diffracted slightly differently, if you could take advantage of that to redirect photons of different wavelengths at the sensor. you could have a camera that takes _full spectrum_ pictures. not just at the single pretty but not very informative red green and blue lines. (tetrachromats rejoice!). Full spectrum sampling in a small package would be really cool, I mean, that is tricorder technology. This is very neat research.

it's called "Compressed Sensing"

[toby]

And this story hit the UK Guardian on 9 Nov 2006. (via CS maven my slice of pizza.)

Ah, more moving parts. THAT's helpful.

[ScentCone]

Is it just me, or does the concept seem inherently more complex and fragile than a multi-pixel sensor with light cast on it?

And how can this possibly deal with the equivalent of a range of shutter speeds in front of a standard sensor? Perhaps it's a matter of how many times the pixel is exposed to the same part of the lens' projection in repeated scans... but that just seems clunky, and that much harder/slower to re-assemble into a stored image.

And it doesn't stop the megapixel chest thumping - it just starts up megamirror arguments, instead.

Seems like it would have one huge drawback

[Solandri]

Low light sensitivity. Digital cameras gain light sensitivity by acting as light buckets. Moreso for CMOS sensors than CCD, but the important thing is that all those sensor pixels are collecting light for their individual pixel simultaneously - in parallel. With a single pixel sensor, this light collection would have to happen in series to achieve the same light sensitivity. If your shutter speed in low light is 1/25 sec with a 5MP traditional digital camera, in order for a single-pixel camera to take the same picture it would need an exposure time of (1/25 sec/pixel)*(5M pixels)*(10% assumed algorithmic efficiency) = 20,000 sec = 5 hours 33 min 20 sec.

Of course since you're doing all this with mirrors, you could set up a megapixel array and have different mirrors shine at different pixels simultaneously (just like a DLP). But that seems to defeat the purpose of the whole rig.

HDR!

[ACMENEWSLLC]

What is cool about this is that it could allow HDR(http://en.wikipedia.org/wiki/High_dynamic_rang e_imaging) in the camera itself.

While you eye can see many different luminosities of light, a camera has limited contrast. Since it is taking not a single picture, but millions of them in an instant - it could also adjust contrast dynamically.

That would be cool.

Re:Still patented too

[Rei]

At first, I thought this was going to be similar to the method of generating hires images from a small number of sensors utilized by jumping spiders. Basically, they vibrate their retinas, recording datapoints from the in-between locations to get in-between pixels.