STABLE METHODS FOR RECONSTRUCTION OF NOISY IMAGES

We consider the numerical reconstruction of noisy images by using two regularization algorithms. The basis of algorithms is Tikhonov regularization with two special nondifferentiable stabilizers. To solve the problem of a nonsmooth minimizing the proximal method and the subgradient process are involved. The results of calculations on the supercomputer 'Uranus', are presented.

Keywords

numerical methods, algorithms, ill-posed problems, inverse problems, iterative regularization, nonsmooth optimization.