Big telescopes plus cheap image-processing outdo Hubble

(((This is a big deal. Imagine this technology embedded in vidcams and binoculars.)))

(((Another swell find from WIRED SCIENCE blog.)))

Link: Press Releases August 2007.

" A team of astronomers have taken pictures of the stars that are sharper than anything produced by the Hubble telescope, at 50 thousandths of the cost.
The researchers, from the University of Cambridge and the
California Institute of Technology (Caltech), used a technique called “Lucky
Imaging” to take the most detailed pictures of stars and nebulae ever produced –
using a camera based on the ground.

Images from ground-based telescopes are usually blurred by the Earth’s atmosphere - the same effect that makes the stars appear to twinkle when we look at them with the naked eye.

The Cambridge/Caltech team, however, surpassed the quality of images taken from space by using a high-speed camera to take numerous images of the same stars at a rate of 20 frames per second. Because of fluctuations in the atmosphere, some of these were less smeared than others. The team then used computer software to choose the best images, and these were combined to create pictures far sharper than anything that has been taken from space.

Dr Craig Mackay, from the Institute of Astronomy at the
University of Cambridge, who led the research, said: “To produce images sharper than Hubble from the ground is a remarkable achievement by anyone’s standards. (((Boy, I'll say.)))

“These are the sharpest images ever taken either from the ground or from space and yet we are essentially using ‘Blue Peter’ technology.
Amateur Lucky Imaging is popular because the technique is so cheap and effective. The low cost means that we could apply the process to telescopes all over the world.”

http://www.ast.cam.ac.uk/~optics/Lucky_Web_Site/index.htm

"Why Has Lucky Imaging Not Been Possible before?

"The principal difficulty with this method is the performance of the CCD cameras that are used at telescopes. CCD detectors are now very close to being theoretically perfect. They have nearly 100% quantum efficiency, superb imaging and cosmetic quality, are available in large areas (the biggest manufactured and sold commercially so far has 110 million pixels) and a very robust electrically and mechanically.

Astronomers generally read out their cameras slowly (typically 30-500 kilohertz pixel rate) in order to minimise the readout noise, the noise that is added to every pixel, irrespective of the light level within it, because of the amplifier on the CCD output. If the CCD is read out quickly enough to be useful for Lucky Imaging (5-35MHz pixel rate) the readout noise is very much higher, typically 100 electrons per pixel per frame read. As an example, with Lucky Imaging we may select 4,000 images out of a total run of 40,000. If the readout noise was as high as 100 electrons per pixel per frame we would find that our summed image has a noise floor that is 64 times higher (square root of 4,000) than the single frame read noise. This would be 6400 electrons per pixel in background noise alone. This will have a dramatic effect on our overall sensitivity. A star image is detected over perhaps 10 pixels and therefore it would need to have something of the order of 300,000 electrons (detected photons) to get significantly above this background noise level.

What has happened recently is that an entirely new output structure (known as low light level CCDs or L3CCDs) has been developed for CCDs by E2V Technologies (Chelmsford, UK). Similar technology electron-multiplying CCDs (EMCCDs) have been developed by Texas Instruments (Japan). Both work by extending the output register with an additional section that is clocked with much higher voltages than usual so as to give a noiseless electron multiplication stage before the output amplifier. This signal amplification stage effectively reduces the readout noise of the on-chip amplifier by the gain factor of this multiplication register which may be set as high as many thousands. A single frame read out at high speed has essentially no readout noise with this technology. It is possible to build cameras that can see each and every individual photon detected by the CCD chip.

"The effect of this is that our limiting sensitivity is dramatically increased to perhaps only 30-100 detected photons from the reference star per frame, a factor of over 3000 on the above example. Lucky Imaging is therefore only possible because of this new technology which we have used in our experiments and observations. For more details of how EMCCDs work in click here."

((("Extraordinary claims require extraordinary evidence," and here is some of it:)))

(((Before:)))

(((After:)))