Chinese researchers have made a major breakthrough in the field of astronomical imaging with a new AI model designed to explore the deepest reaches of the universe.
The model presented was named ASTERIS (Astronomical Spatiotemporal Enhancement and Reconstruction for Image Synthesis) and was developed to analyze computational optics and AI algorithms.
Research results published in magazines sciencedetects extremely weak astronomical signals, helps identify galaxies more than 13 billion light-years away, and produces the deepest deep-space images ever created. Exploring these distant, faint objects is critical for researchers to understand the origin and evolution of the universe, as weak signals are often obscured by background sky noise and thermal radiation.
The model identified more than 160 candidates for high-redshift galaxies from the “dawn of the universe,” about 200 million to 500 million years after the Big Bang, significantly expanding the number of known discoveries from this era. The model is a “photometric adaptive screamer” that analyzes fluctuations in noise to distinguish it from faint, blurry signals from distant stars and galaxies.
In this regard, Professor Dai Qionghai of the Department of Automation at Tsinghua University said, “In astronomical observations, even faint celestial objects that are hidden by optical noise can be reproduced with high fidelity.”
This research study marks a significant milestone in deep space exploration, paving the way for the deployment of AI in the next generation of telescopes, and facilitating breakthroughs in our understanding of dark energy, dark matter, and the origins of the universe.
