Cutting Through the Blur
To the naked eye, stars twinkle, they do so not because the stars are massively variable in their light output (although some are variable, for a different reason), but because the atmosphere between us and space acts like an image does through moving water, it distorts, it shimmers and keeps changing.
For centuries since optics were invented this has been a limiting factor, very distant objects just smudge about too much to be really able to see them clearly.
In the 1950s and later in Sci Fi books of the 70s, the idea of adaptive optics took place, a counter balance to the ever changing flow of air, to be able to smooth out the distortion. It wasn't until the 1990s that such systems really came online, and improved our image of the universe, however by then the Hubble Telescope had already got above the atmosphere to make even the small enhancements of the ground based brethren limited.
Now, with several space based telescopes, a new method has been designed, one that is proving to be far superior to the older adaptive methods.
A team of US/China based scientists are using AI and deep learning to compare Hubble and ground based telescope images, and teach the AI how to compensate for the atmospheric turbulence. Comparisons have shown a 38% improvement compared to 7% from existing AO based solutions.
Space based Telescopes are the future, but ground based telescopes have many other kinds of advantages, particularly in terms of maintenance, upgrade and access. More so, not every bit of science done requires the access of the sharpest possible image, one such project aims to simply record widest possible image of the night sky over a period of time, so rather than 100 telescopes all peering at pin-pricks of the sky, these ground based telescopes scan huge areas of the sky and record it, looking for any changes.
There's a good reason nobody walks around with binoculars stuck to their faces, you may see more in the visible area, but miss the obvious sat next door.
Source :
The software is an adaptation of the same technology you use to improve your selfies.














