Waterloo IVC Dehazed Image Database

Kede Ma , Wentao Liu and Zhou Wang


Introduction

Single image dehazing algorithms are expected to restore a clear and natural image from a hazy one, and make the image applicable for further processing. Currently, a lot of dehazing algorithms were proposed in literature, but little effort on evaluating performances of these algorithms has been made. This dataset includes the results of 8 state-of-the-art dehazing algorithms applied on hazy images of various contents, as well as subjective quality scores assigned to both hazy and dehazed images.

We are making the IQA database available to the research community free of charge. If you use this database in your research, we kindly ask that you reference our paper listed below:


Database Specification

The dataset consists of 25 hazy images covering diverse outdoor scenes and indoor static objects. 22 images of outdoor scenes are captured in the real world and are degraded by haze to different extents, while the hazes of the other 3 indoor images are simulated homogenously. Then 8 dehazing algorithms proposed between 2009 and 2014 are selected to produce 8 different dehazed images for each of the 25 hazy images to form 25 image sets, each of which includes 9 images of the same content (1 hazy, 8 dehazed). In the subjective test, 24 subjects were showed 9 images of one image set at the same time on a display, and were asked to give scores according to perceptual qualities of these images. The score range is from 1 to 10, where 10 indicates the best quality while 1 means the worst. The dataset provided here includes all the 225 images, and all subjective quality scores as well as the Mean Opinion Score (MOS) for each image.


Download

Dehazing Subjective Database


Copyright

Copyright © 2015 The University of Waterloo
All rights reserved.

Permission is hereby granted, without written agreement and without license or royalty fees, to use, copy, modify, and distribute this database (the images, the results and the source files) and its documentation for any purpose, provided that the copyright notice in its entirity appear in all copies of this database, and the original source of this database, Image and Vision Computing Laboratory (IVC, https://ivc.uwaterloo.ca/) at the University of Waterloo (UW, http://www.uwaterloo.ca), is acknowledged in any publication that reports research using this database.

The following papers are to be cited in the bibliography whenever the database is used as:

IN NO EVENT SHALL THE UNIVERSITY OF WATERLOO BE LIABLE TO ANY PARTY FOR DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OF THIS DATABASE AND ITS DOCUMENTATION, EVEN IF THE UNIVERSITY OF WATERLOO HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

THE UNIVERSITY OF WATERLOO SPECIFICALLY DISCLAIMS ANY WARRANTIES,INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OFMERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE DATABASEPROVIDED HEREUNDER IS ON AN "AS IS" BASIS, AND THE UNIVERSITY OFWATERLOO HAS NO OBLIGATION TO PROVIDE MAINTENANCE, SUPPORT, UPDATES,ENHANCEMENTS, OR MODIFICATIONS.


Copyright

Copyright © 2015 The University of Waterloo
All rights reserved.

Permission is hereby granted, without written agreement and without license or royalty fees, to use, copy, modify, and distribute this database (the images, the results and the source files) and its documentation for any purpose, provided that the copyright notice in its entirity appear in all copies of this database, and the original source of this database, Image and Vision Computing Laboratory (IVC, https://ivc.uwaterloo.ca/) at the University of Waterloo (UW, http://www.uwaterloo.ca), is acknowledged in any publication that reports research using this database.