TY - RPRT
T1 - A Parameter Choice Method for Simultaneous Reconstruction and Segmentation
AU - Romanov, Mikhail
AU - Hansen, Per Christian
AU - Dahl, Anders Bjorholm
PY - 2015
Y1 - 2015
N2 - The problem of finding good regularization parameters for the reconstruction problems without knowledge of the ground truth is a non-trivial task. We overview the existing parameterchoice methods and present the modified L-curves approach for a good regularization parameters selection that is suited for our Simultaneous Reconstruction and Segmentation method. We verify the validity of this approach with numerical experiments based on reconstructions of artificial phantoms from noisy data, and the problems in our numerical experiments are underdetermined.
AB - The problem of finding good regularization parameters for the reconstruction problems without knowledge of the ground truth is a non-trivial task. We overview the existing parameterchoice methods and present the modified L-curves approach for a good regularization parameters selection that is suited for our Simultaneous Reconstruction and Segmentation method. We verify the validity of this approach with numerical experiments based on reconstructions of artificial phantoms from noisy data, and the problems in our numerical experiments are underdetermined.
M3 - Report
T3 - DTU Compute Technical Report-2015
BT - A Parameter Choice Method for Simultaneous Reconstruction and Segmentation
PB - Technical University of Denmark
CY - Kgs. Lyngby
ER -