A brand new algorithm able to reworking ground-based telescope photographs by eradicating the blurring impact of the environment to supply as excellent a picture as potential has efficiently accomplished assessments on the eight-meter Subaru Telescope on Hawaii’s Mauna Kea. The subsequent step is to use it to photographs from the Vera C Rubin Observatory when it begins science operations later this yr.
The revolutionary algorithm was developed by Johns Hopkins mathematician Yashil Sukurdeep.
“We dubbed our algorithm ‘Picture MM’ as a result of, at its core, it depends on the Majorization–Minimization, or MM, methodology — a chic mathematical method that we have tailored in a brand new manner for exploring the cosmos,” Sukurdeep stated in an announcement.
Floor-based telescopes have at all times been at an obstacle in comparison with space-based observatories such because the Hubble and James Webb Space Telescopes because light has to pass through Earth’s atmosphere to reach them. The atmosphere distorts light as the result of tiny but ever-present fluctuations in temperature, pressure, the amount of airborne dust and so on. These distortions, which astronomers refer to as “seeing,” are what make stars seem to twinkle.
Astronomers are therefore constantly on a quest to improve their ground-based images and get them as close as possible to a telescope’s theoretical maximum resolution, known as the Dawes limit. Adaptive optics are one popular technique, which involves shining a laser into the sky to create an artificial guide star and then performing minute adjustments to the shape of the telescope’s optics to match the distortions in the guide star and counteract the effects of the seeing.
“Astronomers already have very sophisticated tools to analyze imaging data from telescopes, but they don’t remove all the noise, don’t remove all the blur and they don’t deal very well with missing pixel values,” said Sukurdeep. “Our framework can recover a near-perfect image from a series of imperfect observations.”
ImageMM works by modeling how light from objects in the night sky travels through the distorting atmosphere, and then applying this model to images.
“Think of the atmosphere as a restless sheer curtain, constantly shifting and shimmering, so the scene behind it always looks blurred,” said Sukurdeep. “Our algorithm learns to see past that curtain, reconstructing the still, sharp image hidden behind it.”
So far, the ImageMM algorithm has been tested on the Subaru telescope, returning images sharper and more detailed than what was previously possible with the Japanese-owned observatory.
Now the intention is to use it on images from the Vera C. Rubin Observatory in Chile, particularly because one of Rubin’s science objectives is to map the distribution of dark matter in the universe by measuring how the mass of the dark matter subtly warps space, causing the images of galaxies to be weakly gravitationally lensed and therefore appear slightly deformed. The effect of weak gravitational lensing is not as dramatic as the strong lensing that produces wonderful arcs of light stretching around galaxy clusters and multiple images of background galaxies, which means careful observations must be taken in order to detect weak lensing. ImageMM can sharpen Rubin’s already impressive images of galaxies, making measurements of weak lensing more accurate.
“When it comes to billion-dollar ground-based observatories, gaining even just a small degree of depth and quality improvement from these observations can be huge,” said Tamás Budavári of Johns Hopkins University.
Although space telescopes will still produce better images, they tend to have narrow fields of view, whereas Rubin has a much wider field of view of 3.5 degrees, or about the angular diameter of seven full moons. Therefore, using ImageMM to sharpen Rubin’s images will give it a huge advantage even if the quality of Hubble’s and James Webb’s images are greater overall.
“We’ll never have ground truth, but we think this is as close as it currently gets to perfect [for ground-based telescopes],” said Sukurdeep.
A paper describing ImageMM and its test results was published on Sept. 29 in The Astronomical Journal.
