Journal of Algorithms and Computation
Open Access

Superpixel-Guided Dark Channel Prior for Efficient Single Image Dehazing

Document Type : Research Paper

Authors

Hossein Noori 1
Mohammad Hossein Gholizadeh 2
Gholamreza Memarzadeh 2

1 Vali-e-Asr university of rafsanjan

2 Department of electrical engineering

https://doi.org/10.22059/jac.2025.400037.1239
Abstract
Outdoor cameras play a vital role in security and social governance systems. However, bad weather
can significantly reduce image quality. This can hinder the effectiveness of these systems. This study proposes a new method to remove fog from images captured by outdoor cameras. Unlike traditional methods that analyze small square areas of the image, our approach works with superpixels, which are smarter groupings of pixels. The algorithm first calculates the ”dark channel” for each superpixel. The proposed method then merges superpixels with close dark channel values. To prevent unwanted halos around objects after removing the fog, the algorithm applies ”guided filtering” to the merged dark channel. The effectiveness of this new approach is compared to existing methods using various
tests that measure image quality. The results show that the proposed method outperforms existing algorithms. This allows for clearer images and improved performance of security and social governance systems. Details and numerical rsults might be seen at https://my.uupload.ir/dl/v9EnmnBr.

Keywords

Defogging/dehazing
superpixel segmentation
visibility enhancement
dark channel improvement
region aggregating
Journal of Algorithms and Computation
Volume 57, Issue 2
December 2025
Pages 170-186

Home

Guide for Authors

Submit Manuscript

Reviewers

Contact Us