Contracting authority: Unitatea Executivă pentru Finanțarea Învățământului Superior, a Cercetării, Dezvoltării și Inovării (UEFISCDI) - Executive Unit for the Financing of Higher Education, Research, Development and Innovation

Project Code: PN-III-P1-1.1-TE-2019-1111

Project type: Research projects to stimulate young independent teams (TE)

Project title: Database and fog removal techniques for scenes affected by dense fog

Project acronym: Data - Dehaze


Project start date: 01.11.2020

Project end date: 31.10.2022

Total value of the contract: 450.000 RON

Total duration of the project: 24 months


Project manager:

Conf. Dr. habil. Ing. Codruta Ancuti

University Politehnica Timisoara,

Faculty of Electronics, Telecommunications and Information Technologies, B312 

Blvd. Vasile Parvan 2, 300223, Timisoara, Romania

Tel:  (+40)-0256-403363

Fax: (+40)-0256-403295


Coordinating institution:

University Politehnica Timisoara, Faculty of Electronics, Telecommunications and Information Technologies, Measurements and Optical Electronics (MEO) Department



Project Executive Summary (English):

Outdoor images often suffer from poor visibility introduced by weather conditions, such as haze or fog. Haze is a common atmospheric phenomena produced by small floating particles that absorb and scatter the light from its propagation direction. This results in selective and significant attenuation of the light spectrum, and causes hazy scenes to be subject to a loss of contrast and sharpness for distant objects. The problem worsens in case of poor illumination conditions, as encountered during the night, where artificial lightning becomes non-uniform and biased in terms of spectral distribution. However, to estimate their key internal parameters (e.g. airlight and transmission model), most of those solutions assume homogeneous distribution of light and haze, which is rarely the case in practice (e.g. lighting is non-uniform in space and frequency during the night, attenuation caused by haze depends on the light frequency). Image dehazing thus remains a largely unsolved problem in case of dense and non-homogeneous haze scenes.

The main objective of this project is to design effective image dehazing techniques but also an image interpretation framework that are robust to haze, including the challenging cases where the sources of light and impairment are non-uniformly distributed over the scene. On the basis of this image data-base, our project aims at implementing dehazing methods that are suited to dense and non-homogeneous hazy scenes.

This implies the following three objectives:

(O1) build up the first (world-wide) image dataset including pairs of hazy and haze-free scenes, for which hazy scenes include real, dense, and non-homogeneous haze; O1 was addressed in stage I and stage II.

(O2) develop and train original and effective deep dehazing neural networks to derive the dehazed images from hazy inputs. O2 was addressed in stage II and stage III.

(O3) train deep image interpretation models that are suited to images captured in adverse conditions. O1 was addressed in stage II and stage III.



Stage I: Construction of the database with realistic images affected by fog with non-homogeneous distribution and evaluation of the optical model.

Summary of the results for Stage I

Stage II:  Construction of the database with realistic images affected by dense fog with uneven distribution and evaluation of the optical model. Development of the dehazing method with deep-learning.

Summary of the results for Stage II

 Stage III:  Interpretation of recorded images with deep learning. Dissemination of results. Development of a dehazing method with deep-learning, using the base built of images affected by haze

Rezumat Etapa III si rezultate obtinute



1. C. O. Ancuti, C. Ancuti, F.-A. Vasluianu, R. Timofte, M. Fu, H. Liu et al., „NonHomogeneous Dehazing Challenge Report”, NTIRE 2021, IEEE CVPR 2021, US

2. A. Kis, H. Balta, C. O. Ancuti, „The Impact of Haze Non-Homogeneity on the Recent Image Dehazing Methods”, IEEE ELMAR, Zadar, Croatia, septembrie, 2021

3. M. Sbert, C. Ancuti, C. O. Ancuti, J Poch, S Chen, M Vila, „Histogram Ordering”, IEEE Access, 2021

4. Q Hao, Q Zhao, M Sbert, Q Feng, C. Ancuti, M Feixas, M Vila, J Zhang, „Information-Theoretic Channel for Multi-exposure Image Fusion”, The Computer Journal 2021

5. A Kis, CO Ancuti, Night-time image dehazing using deep hierarchical network trained on day-time hazy images, 2022 International Symposium ELMAR, 199-202