Institute of Statistics

Workgroup "Inverse problems"

• Dr. Tim Kutta

Description:

Inverse problems occur if the quantity of interest is not directly observable. An example is the convolution of an image in optical observations such as astronomical imaging by telescopes with the so-called point-spread-function due to the physical characteristics of the defraction of light at surfaces of mirrors and lenses in the telescope. This results in an observed image, which shows point-like objects as little "discs". Recovery of the true, unobservable image from the convolved image is now an inverse problem. A major problem in the estimation of the true image is that noise in the observed data can result in noise of the reconstruced image of, in general, unlimited magnitude. The workgroup on statistical inverse problems deals with solutions to this problem and, particularly, with statistical inference methods in such models.