ASTERIS Denoising Algorithm Boosts JWST Detection Limits
ASTERIS, a novel denoising algorithm based on self-supervised transformers, enhances the detection limits in astronomical imaging by 1.0 magnitude at 90% completeness and purity. This algorithm is specifically designed for spatiotemporal denoising across various exposures, maintaining both photometric accuracy and the point spread function. Validation through data from the James Webb Space Telescope and Subaru telescope uncovered previously unseen low-surface-brightness galaxy structures and gravitationally-lensed arcs. When applied to deep images from JWST, ASTERIS successfully identified three times the number of galaxy candidates with a redshift greater than 9, showcasing their rest-frame ultraviolet luminosities.
Key facts
- ASTERIS is a self-supervised transformer-based denoising algorithm
- Improves detection limits by 1.0 magnitude at 90% completeness and purity
- Preserves point spread function and photometric accuracy
- Validated with data from James Webb Space Telescope and Subaru telescope
- Identified previously undetectable low-surface-brightness galaxy structures
- Identified gravitationally-lensed arcs
- Applied to deep JWST images, identified three times more redshift > 9 galaxy candidates
- Paper available on arXiv: 2602.17205
Entities
Institutions
- arXiv
- James Webb Space Telescope
- Subaru telescope