Fuzzy First Image Of A Supermassive Black Hole Gets A HIgh-Fidelity AI Makeover
A team of researchers has shown off a new machine-learning technique by enhancing the Event Horizon Telescope's images of a supermassive black hole at the center of a galaxy 55 million light-years from Earth. The technique, called PRIMO, enhances the fidelity and sharpness of radio interferometry images.
The Event Horizon Telescope (EHT), responsible for the original images, is an international collaboration capturing images of black holes using a virtual Earth-sized telescope. As incredible and iconic as the original images of the supermassive black hole at the center of Messier 87 were four years ago, they were still fuzzy in appearance. Researchers using PRIMO, which stands for principal-component interferometric modeling, have been able to give a bit of clarity to the original images.
Medeiros added the that width of the ring in the new image is smaller by around a factor of two, which she says will be "a powerful constraint" for theoretical models and tests of gravity.
By applying PRIMO to the original EHT images, computers scoured over 30,000 high-fidelity simulated images of gas accreting onto a black hole in order to find common patterns within the images. Those results were then "blended to provide a highly accurate representation of the EHT observations." The team then confirmed the newly rendered images were consistent with the EHT data and with theoretical expectations.
According to Medeiros, this is just the beginning for PRIMO. "If a picture is worth a thousand words, the data underlying that image have many many more stories to tell. PRIMO will continue to be a critical tool in extracting such insights."