A breakthrough for ground-based astrophotography
A new algorithm named ImageMM is showing promise in transforming ground-based telescope images by removing atmospheric blur, bringing them closer to the quality long associated with space telescopes. Developed by Johns Hopkins mathematician Yashil Sukurdeep, the method has completed successful tests on Hawaii’s Mauna Kea with the eight-meter Subaru Telescope and is now poised for application to data from the Vera C. Rubin Observatory in Chile when it begins science operations later this year.
“We dubbed our algorithm ‘Image MM’ because, at its core, it relies on the Majorization–Minimization, or MM, method — an elegant mathematical technique that we've adapted in a new way for exploring the cosmos,” Sukurdeep said. The project reflects a growing effort to push the limits of ground-based astronomy and narrow the gap with space-based instruments like the Hubble and the James Webb Space Telescope.
How ImageMM works
Earth's atmosphere distorts incoming light as it travels to ground-based telescopes, a phenomenon astronomers call “seeing.” The result is subtle blurring that prevents even the best telescopes from reaching their theoretical resolution, known as the Dawes limit. Traditional approaches, including adaptive optics, use artificial guide stars and minute optical adjustments to counteract turbulence. ImageMM takes a different tack by modeling how light passes through the turbulent atmosphere and then reconstructing a sharpened image from a series of imperfect observations.
“Think of the atmosphere as a restless sheer curtain, constantly shifting and shimmering, so the scene behind it always looks blurred,” Sukurdeep explained. “Our algorithm learns to see past that curtain, reconstructing the still, sharp image hidden behind it.” The framework does not merely denoise; it jointly recovers missing pixels and refines the underlying signal by exploiting the physics of light transport through the atmosphere.
In practice, ImageMM fits a forward model of how a celestial object becomes a blurred image on the detector and then iteratively improves the estimate of the true scene. The Majorization–Minimization approach provides a mathematically principled way to converge toward the best possible reconstruction, even when data are noisy or partially missing. Sukurdeep and colleagues report that the Subaru tests yielded images with sharper detail and finer structures than prior ground-based results from the same telescope.
Why this matters for Rubin Observatory
The Rubin Observatory is designed to survey a huge swath of the sky, producing wide-field images that map the distribution of dark matter through weak gravitational lensing. While Rubin’s field of view is expansive, the extra clarity that ImageMM affords could translate into more precise measurements of how light from distant galaxies has been subtly warped by mass along the line of sight.
“Rubin has a 3.5-degree field of view, which is large compared with many space missions, and tighter image quality across those degrees can significantly improve lensing measurements,” said Tamás Budavári of Johns Hopkins University. By sharpening Rubin’s galaxies, observers can better quantify tiny distortions that reveal the distribution of dark matter, a central goal of contemporary cosmology.
Of course, space-based telescopes will continue to lead in ultimate resolution, and their images cover smaller patches of sky. Yet the Rubin Observatory's wide field is precisely where ImageMM could offer a substantial advantage, delivering sharper, more reliable data across a broad swath of the cosmos. Sukurdeep notes that while there may never be a perfect ground truth for all-sky observations, ImageMM represents a near-ideal improvement for ground-based imaging given current technology.
Looking ahead: challenges and promise
The transition from Subaru to Rubin will involve extensive validation, calibration, and joint optimization with Rubin's data processing pipelines. The aim is to maximize the accuracy of weak-lensing measurements without introducing biases that could mislead cosmological inferences. Researchers acknowledge that atmospheric conditions vary and that no single method can be a panacea for all images. The strength of ImageMM lies in its principled, model-based framework that can adapt to a range of observing conditions and telescope configurations.
“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,” Budavári emphasized. While space telescopes like Hubble or Webb will still display superior resolution in many cases, ImageMM adds a critical enhancement to Rubin's broader, deeper survey, potentially bringing some ground-based results very close to space-like fidelity and enabling more precise cosmological measurements in the near term.
As the Rubin Observatory prepares to commence science operations, the astronomy community will be watching closely to see how ImageMM integrates with ongoing surveys and how much it can sharpen the data used to map the unseen structure of the universe. Sukurdeep remains optimistic: “We'll never have ground truth, but we think this is as close as it currently gets to perfect [for ground-based telescopes].”
