基于边缘检测暗通道先验的全变差图像去雾算法

A total variation image dehazing algorithm based on dark channel prior with edge indicator

为了恢复有雾图像中更多如边缘、结构等有价值的图像细节信息,文中引入带边缘检测的暗通道先验来估计初始场景深度,提出基于边缘增强的全变差图像去雾模型,证明该模型极小化问题解的存在性和唯一性。进一步,结合原始⁃对偶方法,设计该模型的快速数值求解算法,并给出了该算法的收敛性结果。最后,数值实验结果验证所提算法的可行性与有效性。

In order to recover more valuable image details such as edges and structures in hazy images,this paper introduces a dark channel prior with the edge indicator function to estimate the initial scene depth,and proposes a total variation image dehazing model based on edge enhancement.Then,the existence and uniqueness of the solutions to the minimization problem of this model are proved.Further,we design a primal⁃dual algorithm to solve the model and provide its convergence result.Finally,numerical results verify the feasibility and effectiveness of the proposed algorithm.