Dual-Scale Single Image Dehazing Via Neural Augmentation

TL;DR

This paper introduces a novel single image dehazing method that combines model-based and data-driven techniques using dual-scale GANs to effectively remove haze from both real-world and synthetic images.

Contribution

It proposes a hybrid dehazing algorithm that refines transmission and atmospheric light estimates with dual-scale GANs, achieving fast convergence and improved haze removal.

Findings

Effective haze removal on real-world images

High PSNR and SSIM on synthetic images

Fast convergence of the neural augmentation approach

Abstract

Model-based single image dehazing algorithms restore haze-free images with sharp edges and rich details for real-world hazy images at the expense of low PSNR and SSIM values for synthetic hazy images. Data-driven ones restore haze-free images with high PSNR and SSIM values for synthetic hazy images but with low contrast, and even some remaining haze for real world hazy images. In this paper, a novel single image dehazing algorithm is introduced by combining model-based and data-driven approaches. Both transmission map and atmospheric light are first estimated by the model-based methods, and then refined by dual-scale generative adversarial networks (GANs) based approaches. The resultant algorithm forms a neural augmentation which converges very fast while the corresponding data-driven approach might not converge. Haze-free images are restored by using the estimated transmission map and atmospheric light as well as the Koschmiederlaw. Experimental results indicate that the proposed algorithm can remove haze well from real-world and synthetic hazy images.