@inproceedings{4971f6c0f8a44db1a3fb78d4e896520f,
title = "Synthetic Images as a Regularity Prior for Image Restoration Neural Networks",
abstract = "Deep neural networks have recently surpassed other image restoration methods which rely on hand-crafted priors. However, such networks usually require large databases and need to be retrained for each new modality. In this paper, we show that we can reach near-optimal performances by training them on a synthetic dataset made of realizations of a dead leaves model, both for image denoising and super-resolution. The simplicity of this model makes it possible to create large databases with only a few parameters. We also show that training a network with a mix of natural and synthetic images does not affect results on natural images while improving the results on dead leaves images, which are classically used for evaluating the preservation of textures. We thoroughly describe the image model and its implementation, before giving experimental results.",
keywords = "Deep learning, Image restoration, Natural image models",
author = "Rapha{\"e}l Achddou and Yann Gousseau and Sa{\"i}d Ladjal",
note = "Publisher Copyright: {\textcopyright} 2021, Springer Nature Switzerland AG.; 8th International Conference on Scale Space and Variational Methods in Computer Vision, SSVM 2021 ; Conference date: 16-05-2021 Through 20-05-2021",
year = "2021",
month = jan,
day = "1",
doi = "10.1007/978-3-030-75549-2\_27",
language = "English",
isbn = "9783030755485",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Science and Business Media Deutschland GmbH",
pages = "333--345",
editor = "Abderrahim Elmoataz and Jalal Fadili and Yvain Qu{\'e}au and Julien Rabin and Lo{\"i}c Simon",
booktitle = "Scale Space and Variational Methods in Computer Vision - 8th International Conference, SSVM 2021, Proceedings",
}