The latest challenges and opportunities in the current single image dehazing algorithms

Haze is one of the most common atmospheric phenomena. In hazy weather, smoke, dust and other dry particles obscure the clarity of the scenery objects severely degrading the quality of the outdoor images taken by camera. Moreover, most outdoor vision applications such as video-surveillance systems and traffic monitoring systems fail to work normally in this condition. Therefore, improving the technique of image dehazing has important and realistic significance. Although significant progress has been made during recent years, there are still some problems in image haze removal. In this paper, we will discuss three grand challenges of single image dehazing: 1) Find a way to solve the ambiguity problem between image color and depth in dehazing; 2) Create a more robust model for dehazing in inhomogeneous atmosphere; 3) Develop a reliable image quality assessment mechanism to evaluate the quality of dehazing algorithms. At the end of this paper, we summarize the discussions on these challenges.