Photo-realistic dehazing via contextual generative adversarial networks

摘要

Single image dehazing is a challenging task due to its ambiguous nature. In this paper we present a new model based on generative adversarial networks (GANs) for single image dehazing, called as dehazing GAN. In contrast to estimating the transmission map and the atmospheric light separately as most existing deep learning methods, dehazing GAN restores the corresponding hazy-free image directly from a hazy image via a generative adversarial network. Extensive experimental results on both synthetic dataset and real-world images show our model outperforms the state-of-the-art algorithms.