Wavelet Total Variation Fusion and Fuzzy Logic Decision-Based Haze Density and Sky Region-Aware De-Hazing.

In this study, we present a fuzzy logic-based classifier with a decision-based partial differential equation (PDE) for single image de-hazing. This enhanced partitioning is achieved via adaptive segmentation and morphological region-growing processes. Also, the first stage fusion of enhanced inverted log illumination and log reflectance components leads to minimization or elimination of edge artifacts. Additionally, improved scene discrimination between sky and non-sky regions due to the fuzzy logic decision-based approach is achieved. This results in effective de-hazing without sky region over-enhancement. Furthermore, a modified wavelet multi-level decomposition-based de-hazing algorithm is presented. It improves details and local contrast in de-hazed images, especially for images with thick haze. Ultimately, these algorithms are fused to obtain the best and most consistent results for all images. Experimental comparisons with the state-of-the-art algorithms show improved sky discrimination and improved clarity. [ABSTRACT FROM AUTHOR]