Work

Optimising hazy image clarity through blue channel balancing in lab color space

Year: 2026

Type: article

Abstract: The presence of haze in the atmosphere results in a bluish and yellowish tint in the affected area. This research assessed the histogram of hazy images and found that the blue channel was overly dominant, particularly in images exhibiting non-homogeneous haze. This pronounced blue channel is associated with the blue or yellow appearance of the haze, especially in the sky regions of the images. The blue channel in the haze image causes a loss of details and visual distraction. To address this issue, this research proposed a blue channel balancing method within the LAB colour space. This adjustment reduces the excessive blue levels, thereby enhancing image clarity by mitigating the hazy effect. Qualitative results indicate that balancing the blue channel significantly improves the dehazing process, making it more effective in removing haze, even in conditions where the haze is quite dense. However, it is important to note that this method may result in a slight darkening of the image and the appearance of twinkles (i.e., sparkles or localised brightening in certain areas of the image). The blue channel balancing method has proven particularly effective under medium and low haze conditions. Similarly, this method effectively reduces haze in images that are heavily affected by haze.

Source: Discover Artificial Intelligence

DOI: 10.1007/s44163-025-00798-5

Authors: Ibrahim Salim Sulaiman Al Farsi, Mohd Shafry Mohd Rahim, Devi Willieam Anggara, Falah Y. H. Ahmed, Asniyani Nur Haidar Abdullah

Institutions: University of Technology Malaysia, Sohar University, Bandung Institute of Technology, Universiti Teknikal Malaysia Melaka

Language: English

Topic: Image Enhancement Techniques

Subfield: Computer Vision and Pattern Recognition

Field: Computer Science

Domain: Physical Sciences

Open Access status: gold

Estimated APC paid: $1,590

Funder: Ministry of Higher Education, Research and Innovation