Daytime and Nighttime Dehazing Benchmarking

Published: 1 May 2024| Version 1 | DOI: 10.17632/zmg8mrcm9w.1
Sriparna Banerjee,


This database contains a total of 330 images out of which 24 images are hazy and haze-free (Ground Truth) images of four scenes which are captured during both daytime as well as nighttime while others are output images which are obtained by applying several well-known dehazing methods on hazy images (both synthetically generated as well as real) present in this database. For a particular scene, in this database we have included four hazy images and two hazefree/ Ground Truth images like for image 1(also we can refer to as scene 1) we have included four images (image1_daytime_realhaze, image1_daytime_synthetichaze, image1_nighttime_realhaze and image1_nighttime_synthetichaze) and two haze-free images (image1_daytime_hazefree and image1_nighttime_hazefree) in this database. The images whose names include“_realhaze term” are captured in the presence of real atmospheric haze and the images whose names include _synthetichaze term are obtained by synthetically introducing haze in images using the “Dehaze” filter option present under the “Effects” section in the “Adobe Lightroom App”. The images whose names include“_hazefree term” are captured during clear weather conditions. Apart from hazy and haze-free images of each scene, in this database we have also included the outputs obtained by applying various well-known dehazing methods on hazy images. We have named those images as “imagename_type of haze (realhaze/synthetichaze)_first author surname_publication year”.



Computer Vision, Image Processing, Image Dehazing