Images taken in foggy or hazy weather circumstances can be acutely affected by scattering of atmospheric particles, which reduces the contrast, modifies the colour, and makes the object features diffcult to identify by human vision and by some outdoor computer vision systems. Hence, image dehazing is a vital problem and has been widely researched in the domain of computer vision. The role of image dehazing is to lessen the impact of weather elements to enhance the visual effects of the image and give assistance to post processing of the image. This seminar reviews the main methods of image dehazing that have been developed over the past decade and explores a technique using Ranking CNN to automatically learn haze relevant features which can be used to dehaze input hazy image.
By training RankingCNN in a well-designed manner, powerful haze-relevant features can be automatically learned from massive hazy image patches. Based on these features, haze can be effectively removed by using a haze density prediction model trained through the random forest regression.