Operator Splitting Techniques in Image Processing
Nonlinear Aspects of Seismic Imaging Recently, operator splitting methods have been successfully applied to various image processing tasks like the denoising and deblurring of images also in the presence of non-additive noise, inpainting, sparse recovery problems and multi-task learning. Splitting methods allow us to decompose the original problem into subproblems which are easier to solve. The talk reviews and relates various of these optimization methods from the point of view of averaged operators, Bregman proximal point methods and primal-dual Lagrangian approaches. Attention is also paid to multistep methods. Then, various examples are presented how splitting algorithms can be successfully applied to image recovery problems.
Gabriele Steidl, Universität Mannheim, Germany