WLMS-based Transmission Refined Self-Adjusted No Reference Weather Independent Image Visibility Improvement

Abstract

Single image dehazing is an ill-posed, challenging issue out of several dehazing techniques due to its limited information. Transmission map is a crucial reference to retrieve the dehazed output. An attempt has been taken to remove artefacts in the depth map of the hazy image with Weighted Least Mean Square-based edge preserved smoothing technique which in turn estimate artifact-free transmission map. Original bright, clear, and natural image radiance is restored using refined transmission estimation and image recovery optical model developed and it has been showed to be effective in comparison with the state-of-the art techniques. Extensive experiments (quality and quantity) validate that the proposed technique is computationally less expensive, simple, and efficient. Computational complexity is linear or quasi-linear depending on the size of the image, O(n) or O(nlogn). In either case complexity is still lower than other state-of-the art techniques. The proposed algorithm is adaptable to remove different types of haze. Visible distance, as well as extinction coefficient of the medium of the hazy and recovered images, are also measured.

KEYWORDS:

• WLMS
• Dehazing
• Edge preserving Smoothing
• Computer vision
• Computational complexity

Acknowledegments

The authors would like to thank the reviewers for their valuable comments. FRIDA, K He, and MSCNN dataset are used. MATLAB is used as software. Apple MacBook Air is used as workstation.

Additional information

Notes on contributors

Sangita Roy

Sangita Roy is an assistant professor at ECE Department of Narula Institute of Technology under WBUT. She has a teaching experience of more than 24 years. She was in Bells Controls Limited (instrumentation industry) for two years and West Bengal State Centre, IEI (Kolkata) in administration for two years. She completed her diploma (ETCE), AMIE(ECE) and MTech (Comm. Egg.). Currently, perusing her PhD under Sheli Sinha Chaudhuri at ETCE Department of Jaduvpur University since 2012. She is a member of IEI, IETE, FOSET, ISOC, and IEEE ComSoc. She has published journals as well as conference papers. Corresponding author. Email: [email protected]

Sheli Sinha Chaudhuri

Sheli Sinha Chaudhuri is a professor and current head of the ETCE Department of Jaduvpur University. She completed her BTech, MTech and PhD at Jaduvpur University. She has a vast teaching experience of 19 years. She has large number of papers in international and national level journals as well as conferences. Currently, research scholars are pursuing PhD under her guidance. She is a member of IEEE and IEI. Email: [email protected]