Single image haze removal algorithm using pixel-based airlight constraints
Zhenwei Gao, Yongqiang Bai · 2016
Fast single image dehazing has been a challenging problem in many fields, such as computer vision and real-time applications. The existing image dehazing algorithms cannot achieve a trade-off between the dehazing performance and the computational complexity. The proposed approach first applies the mean filter twice to estimate airlight, which include pixel-based dark channel and bright channel constraints. And then the relationship between channel values of the restored image and atmospheric light is qualitatively analyzed to give the optimum estimate of atmospheric light. Using the airlight and atmospheric light, we can easily restore the scene radiance via the atmospheric scattering model. Compared with others, the main advantage of the proposed approach is its high speed and significant visibility improvement even in the sky and white areas. This speed allows the enhanced haze image to be applied in real-time processing applications. A comparative study and quantitative evaluation are proposed with a few other state of the art algorithms which demonstrates that similar or better quality results are obtained.