Noise2Atom: Unsupervised Denoising for Scanning Transmission Electron Microscopy Images

Noise2Atom

Requirements:

• Python 3.8.5
• Tensorflow 1.14 important
• opencv 4.4.0
• python-imageio 2.8.0
• python-numpy 1.19.1
• python-tifffile 2020.7.24
• python-pathos 0.2.3

We use a telegram bot to monitor the real time training process. The private key and private chat id in file code/message.py should be updated before training.

Denoising on your own dataset

• Simulating Gaussian-like atomic images by using routine implemented in file code/simulate_physical_model.py.
• Config then execute the training routine implemented in file code/train.py

Cite us

@article{wang_noise2atom_2020,
	title = {{Noise2Atom}: unsupervised denoising for scanning transmission electron microscopy images},
	volume = {50},
	copyright = {All rights reserved},
	issn = {2287-4445},
	shorttitle = {{Noise2Atom}},
	url = {https://doi.org/10.1186/s42649-020-00041-8},
	doi = {10.1186/s42649-020-00041-8},
	language = {en},
	number = {1},
	urldate = {2020-10-23},
	journal = {Applied Microscopy},
	author = {Wang, Feng and Henninen, Trond R. and Keller, Debora and Erni, Rolf},
	month = oct,
	year = {2020},
	pages = {23}
}

License

GNU AGPLv3