Self-supervised denoising for massive noisy images
This addresses the challenge of denoising in fields like materials science and astronomy where clean data is scarce, offering a practical solution.
The authors tackled the problem of signal reconstruction from massive noisy images without needing clean samples, signal priors, or noise model calibration, achieving effective performance across applications like atomic and astronomy imaging.
We propose an effective deep learning model for signal reconstruction, which requires no signal prior, no noise model calibration, and no clean samples. This model only assumes that the noise is independent of the measurement and that the true signals share the same structured information. We demonstrate its performance on a variety of real-world applications, from sub-Ångström resolution atomic images to sub-arcsecond resolution astronomy images.