Skip to content

This source code is a MATLAB implementation of a haze removal algorithm that can deal with the post-dehazing false enlargement of white objects effectively. The work was published in MDPI Sensors journal under the title "Single-Image Visibility Restoration: A Machine Learning Approach and Its 4K-Capable Hardware Accelerator".

datngo93/HRwFE

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

10 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

HRwFE

Note

I am unable to public the source code of Adaptive Tone Remapping (ATR) due to copyright restrictions. Thus, the dehazed results at this time are different from those being presented in the published paper. I will try my utmost to circumvent this issue.

Summary

This source code is a MATLAB implementation of a haze removal algorithm that can deal with the post-dehazing false enlargement of white objects effectively. The following table summarized the main features of this algorithm.

Feature Description
Background noise removal The backgroud noise originated from the spike-like noise in the saturation channel. Thus, a simple low-pass filter removed the background noise effectively.
Color distortion correction The employed linear regression model for estimating the scene depth misinterpreted dark regions as close regions, giving rise to the color distortion. HRwFE resolved this problem by an efficient means of adaptive weighting.
Post-dehazing false enlargement correction The difference between the atmospheric light pixel and pixels surrounding white objects caused the post-dehazing false enlargement. HRwFE employed a compensation scheme that scales up the atmospheric light according to solve this problem.

Feel free to contact me (ngodat9093@gmail.com) if you are interested in the hardware implementation of this dehazing algorithm.

Run the code

Run the file "demonstration_without_ATR.m" in MATLAB R2019a to view the result. The source code is in the source_code folder.

Qualitative results

First Second Third

About

This source code is a MATLAB implementation of a haze removal algorithm that can deal with the post-dehazing false enlargement of white objects effectively. The work was published in MDPI Sensors journal under the title "Single-Image Visibility Restoration: A Machine Learning Approach and Its 4K-Capable Hardware Accelerator".

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages