Saturation Based Iterative Approach for Single Image Dehazing
Zongwei Lu, Bangyuan Long, Shiqi Yang
Abstract
In this letter, we explore the dark channel prior from a new perspective. By re-formulating the dark channel, it is found that the dark channel correlates closely with the saturation and brightness. A novel way to estimate the transmission without the need to compute the dark channel is then presented and it can prevent the transmission from being under-estimated. In order to discourage the scene radiance from becoming over-saturation and remove the haze effectively, an iterative approach with a tolerable bound on the saturation is proposed. Qualitative and quantitative experimental results demonstrate that the proposed algorithm can effectively restore the scene radiance with a comparable or better visual quality than many other algorithms. Furthermore, it runs faster than most state-of-the-arts.