Joint self-supervised blind denoising and noise estimation

Jean Ollion (1) , Charles Ollion (2) , Elisabeth Gassiat (3, 4) , Luc Lehéricy (5) , Sylvain Le Corff (6, 7, 8)

1 SABILab - Scientific Analysis of Bio-Images Laboratory
2 CMAP - Centre de Mathématiques Appliquées - Ecole Polytechnique
3 LMO - Laboratoire de Mathématiques d'Orsay
4 Université Paris-Saclay
5 LJAD - Laboratoire Jean Alexandre Dieudonné
6 IP Paris - Institut Polytechnique de Paris
7 TSP - CITI - Communications, Images et Traitement de l'Information
8 TIPIC-SAMOVAR - Traitement de l'Information Pour Images et Communications

We propose a novel self-supervised image blind denoising approach in which two neural networks jointly predict the clean signal and infer the noise distribution. Assuming that the noisy observations are independent conditionally to the signal, the networks can be jointly trained without clean training data. Therefore, our approach is particularly relevant for biomedical image denoising where the noise is difficult to model precisely and clean training data are usually unavailable. Our method significantly outperforms current state-of-the-art self-supervised blind denoising algorithms, on six publicly available biomedical image datasets. We also show empirically with synthetic noisy data that our model captures the noise distribution efficiently. Finally, the described framework is simple, lightweight and computationally efficient, making it useful in practical cases.

Domaines: Machine Learning [stat.ML]; Applications [stat.AP]; Apprentissage [cs.LG]

Jean Ollion, Charles Ollion, Elisabeth Gassiat, Luc Lehéricy, Sylvain Le Corff. Joint self-supervised blind denoising and noise estimation. 2021. ⟨hal-03140686⟩