Single image dehazing and denoising with variational method
Faming Fang, Li Fang, Xiaomei Yang, Chaomin Shen, Guixu Zhang · 2010
In this paper, we propose an unified variational approach for image dehazing and denoising from a single input image. Total variation regularization terms are used in the energy functional. Also, we use the negative gradient descent method to solve the corresponding Euler-Lagrange equations. To obtain good initial values, we improve the estimation of transmission map with the windows adaptive method based on the dark channel prior which can overcomes the block effects. The numerical results demonstrate that our algorithm is effective and promising.