Fairness in Single Image Dehazing: Sustainable and Ethical AI

Akshay Juneja, Vijay Kumar, Sunil Kumar Singla

Abstract: Due to the presence of several air pollutants, drivers’ visibility is considerably reduced. The performance of object identification systems, surveillance, and autonomous vehicle is significantly affected. In recent years, various image dehazing algorithms have been developed for visibility restoration applications. A fast and robust algorithm called bounding function for gray-world kernel prior is proposed for image dehazing to generate transmission maps and minimize the distortion in dehazed images. It outperforms the existing filter-based and prior-based techniques in terms of Blind/Referenceless Image Spatial Quality Evaluator (BRISQUE) and Natural Image Quality Evaluator (NIQE). It generates minimum Fog Aware Density Evaluator (FADE) value and least computation time, resulting in fast dehazing. This article presents the importance of sustainable artificial intelligence (AI) to highlight the importance of resource conservation and hardware utilization. Furthermore, the concept of ethical AI comprising biasness and transparency in the dehazed images is introduced.

External IDs:dblp:journals/itpro/JunejaKS25