Stochastic Frequency Masking to Improve Super-Resolution and Denoising Networks

Super-resolution and denoising are ill-posed yet fundamental image restoration tasks. In blind settings, the degradation kernel or the noise level are unknown. This makes restoration even more challenging, notably for learning-based methods, as they tend to overfit to the degradation seen during training. We present an analysis, in the frequency domain, of degradation-kernel overfitting in super-resolution and introduce a conditional learning perspective that extends to both super-resolution and denoising. Building on our formulation, we propose a stochastic frequency masking of images used in training to regularize the networks and address the overfitting problem. Our technique improves state-of-the-art methods on blind super-resolution with different synthetic kernels, real super-resolution, blind Gaussian denoising, and real-image denoising.


Published in:
[Proceedings of ECCV '20]
Presented at:
16th European Conference on Computer Vision - ECCV '20, [Online event], August 23-28, 2020
Year:
Jul 22 2020
Keywords:
Note:
Code: https://github.com/majedelhelou/SFM
Additional link:
Laboratories:


Note: The status of this file is: Anyone


 Record created 2020-07-22, last modified 2020-10-29

Fulltext:
Download fulltext
PDF

Rate this document:

Rate this document:
1
2
3
 
(Not yet reviewed)