Our limiting factor in astrophotography is definitely noise.It can only be reduced by stacking the traces or filtering during processing.There is two ways to reduce random noise level; a vertical one: stacking several pictures of the same object. And spatial ones, which are different kind of clever smoothing. I think that machine learning can open a new kind of noise reduction.
The trend to denoise images with machine learning is to train GAN with a lot of examples. It will be very efficient for smartphone pictures, but very dangerous for scientific data as it can introduce structures from another dataset. To improve my astronomy pictures quality, I’m using machine learning that only use corrupted data, i.e. without apriori information.
I used a deep convolutional network called DeepPior to denoise my astronomy data. This algorithm doesn’t search for the answer in the image space like a trained GAN should do, it searches for the answer in the space of neural network’s parameters. It’s like a brain that trains itself to recreate that image without noise.
I shot M51 galaxy with a 0.4m Ritchey Chretien telescope in New Mexico. On the top left of the attached figure, this is a raw shot of 20 minutes. On the bottom right, a restoration using deep prior. This is way above a conventional denoise result. Of course it's below a stack; this is just a test to show the technology. I highly recommend to use it post stack.
More of my work here: https://www.instagram.com/ben_b_allen/
Edited by BenAllen, 15 April 2019 - 06:33 PM.