Relevance of Color spaces and Color channels in performing Image dehazing

Abstract

In recent years, image dehazing has became a widely popular research topic because of its societal importance. Proper dehazing of images is very necessary to restore the visibility of images. In order to perform accurate dehazing of images it is very essential to identify the color channels of hazy images in different color spaces which are getting more influenced by the scattering and attenuation of scene light due to the presence of haze particles in the atmosphere in comparison to the other less influenced color channels so that the effect of airlight on the selected (more influenced) color channels can be removed and visibility of degraded hazy images can be fully restored without even processing the least influenced color channels. In this paper, we have performed a comprehensive analyses by comparing the histograms and the intensities of the color channels of different hazy (both nighttime and daytime) images and haze free images of the same scene to identify which of the color channels in RGB, HSV and L*a*b color spaces are getting more influenced by the presence of haze in the atmosphere in both daytime and nighttime hazy images as the characteristics of nighttime hazy images vastly differs from those of daytime hazy images.