2000 character limit reached
Ground Truth Free Denoising by Optimal Transport (2007.01575v1)
Published 3 Jul 2020 in cs.CV, cs.NE, math.FA, and math.OC
Abstract: 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.