arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2007.01575cs.CVcs.NEmath.FAmath.OC
收藏

基于最优传输的真值自由去噪

Ground Truth Free Denoising by Optimal Transport

University of Bremen(不来梅大学);University of Cambridge(剑桥大学)

Sören Dittmer, Carola-Bibiane Schönlieb, Peter Maass

摘要:

我们为任意类型的数据提出了一种学习的无监督去噪方法,我们在图像和一维信号上进行了探索。训练完全基于噪声数据的样本和噪声的示例,关键是不需要成对出现。我们只需要假设噪声是独立的和可加的(尽管我们描述了如何扩展)。该方法依赖于Wasserstein生成对抗网络设置,该设置利用两个批评者和一个生成器。

英文摘要:

We present a learned unsupervised denoising method for arbitrary types of data, which we explore on images and one-dimensional signals. The training is solely based on samples of noisy data and examples of noise, which -- critically -- do not need to come in pairs. We only need the assumption that the noise is independent and additive (although we describe how this can be extended). The method rests on a Wasserstein Generative Adversarial Network setting, which utilizes two critics and one generator.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑