Evaluation of image dehazing techniques based on a realistic benchmark

Image dehazing is a complex problem that aims to restore the Visibility of hazy scenes. In recent years, there has been a remarkable progress of single image dehazing techniques, but an important problem is the lack of realistic dehazing benchmarks. In this paper we introduce a new realistic dehazing dataset that consists of 22 pairs of hazy images and haze-free images (ground truth). The sets of images are recorded indoor and are characterized by dense and homogeneous hazy images that have been recorded by introducing real haze, generated by professional haze machines. The scenes have been captured under the same illumination conditions for both hazy and haze-free images. In addition, we perform a comprehensive evaluation of the state-of-the-art single image dehazing techniques providing also a quantitative validation of the existing dehaing methods using the PSNR, SSIM and CIEDE2000 objective measures.