Department of Image Processing

In the talk we will consider a linear inverse problem that is contaminated with mixed Poisson-Gaussian noise, and when there are in addition outliers in the measured data. The Poisson-Gaussian noise leads to a weighted minimization problem, with solution-dependent weights. To address outliers, we propose a modification of the data fidelity function using the idea of robust regression. Convexity and regularization parameter selection schemes will be discussed as well as other computational aspects.