An adaptive image dehazing algorithm based on dark channel prior

Traditional dehazing algorithm based on dark channel prior may suffer weak robustness against the variation of hazy weather and may fail in bright regions. To resolve these issues, this paper proposes an improved adaptive dehazing algorithm based on dark channel prior. Our method can adaptively calculate dehazing parameter, such as the degree of haze removal. Here the dehazing parameters are local, rather than global variables. We compute the local dehazing parameter automatically according to haze distribution, which makes our method being able to handle different dehazing degrees under various weather conditions, and makes haze removal more robust. We also propose a new method to optimize the rough transmission paramters, which can help to remove the distortion in bright regions. Experiments confirm the advantages of our method, such as robustness against different scenes, high color fidelity of the restored images and greatly enhanced details of the hazy regions.