Single-image dehazing is an extensively studied field and an ill-posed problem faced by vision-based systems in an outdoor environment. This paper proposes a dark channel network to estimate the transmission map of an input hazy scene for single-image dehazing. The architecture constitutes two major components—feature extraction layer and convolutional neural network layer. The former extracts the haze relevant features, while latter convolve these features with filter kernels to estimate the true scene transmission. Finally, the estimated transmission map is used to obtain the dehazed image using atmospheric scattering model. The experiments have been performed on synthetic hazy images and benchmark hazy dataset available in the literature. The performance of the proposed architecture outperforms the existing models in terms of standard quantitative metrics—mean square error, structural similarity index, and peak signal-to-noise ratio.

单图像除雾是一个广泛研究的领域，也是室外环境中基于视觉的系统面临的不适问题。本文提出了一种暗通道网络来估计用于单图像去雾的输入朦胧场景的传输图。该架构由两个主要组件组成-特征提取层和卷积神经网络层。前者提取雾度相关特征，而后者将这些特征与滤波器内核进行卷积以估计真实的场景传输。最后，利用大气散射模型将估计的透射图用于获得去雾图像。实验是在文献中可获得的合成模糊图像和基准模糊数据集上进行的。