Deep Learning Based Single Image Dehazing
Deep Learning Based Single Image Dehazing
复制标题
DOI：
10.1109/cvprw.2018.00162
复制
发表时间：
2018-06
期刊：
影响因子：
--
通讯作者：
Patricia L. Suárez;A. Sappa;B. Vintimilla;R. Hammoud
中科院分区：
文献类型：
--
作者：
Patricia L. Suárez;A. Sappa;B. Vintimilla;R. Hammoud
本文提出了一种使用堆叠条件生成对抗网络 (GAN) 消除 RGB 图像中雾霾退化的新方法。它采用三元组 GAN 来独立消除每个颜色通道上的雾气。提出了应用于条件概率模型的多重损失函数方案。所提出的 GAN 架构学习去除雾霾，使用带有雾霾的图像作为条件入口，从中获得清晰的图像。这样的公式确保了快速的模型训练收敛和同质的模型泛化。实验表明，所提出的方法可以生成高质量的清晰图像。
This paper proposes a novel approach to remove haze degradations in RGB images using a stacked conditional Generative Adversarial Network (GAN). It employs a triplet of GAN to remove the haze on each color channel independently. A multiple loss functions scheme, applied over a conditional probabilistic model, is proposed. The proposed GAN architecture learns to remove the haze, using as conditioned entrance, the images with haze from which the clear images will be obtained. Such formulation ensures a fast model training convergence and a homogeneous model generalization. Experiments showed that the proposed method generates high-quality clear images.