title:
Compressive strategies for ill-posed problems |
name:
Dirk Lorenz |
first name:
Gerd Teschke, Evelyn Herrholz
|
location/conference:
SPP-JT09
|
PRESENTATION-link:
http://dfg-spp1324.de/download/jt09/talks/lorenz_teschke_herrholz.pdf |
abstract:
In the short project presentation we discuss first ideas on a framework for solving inverse problems capitalizing compressive sampling strategies. Typically, in inverse problems the usual assumptions for compressive sampling are nor fulfilled and have to adapted. The actual approach relies on a specific combination of regularization and compression resulting in a particular design of the sampling functions. The applicability of this proceeding is verfied by first numerical experiments. In addition to this, we present verifiable conditions for exact recovery with orthogonal matching pursuit which also work for these problems and present an example from digital holography. |