Monday,
May 4, 10:00 a.m. PSA 206
Wei Shen
School of Math&Stat Sciences
A Fast Reconstruction Algorithm with TV-L1 Regularization and Applications in MRI
Abstract
The recently developed mathematical theory of compressive sensing shows that it
is possible to reconstruct certain images with sparse representation from
relatively few linear measurements by minimizing a L1-norm related non-smooth
convex problem. In this work, we study an approach which minimizes an objective
function which includes the sum of a quadratic error term combined with the
L1-norm and total variation (TV) regularization terms. We propose an
algorithm based on the nonlinear conjugate gradient method with preconditioning
to solve this TV-L1 regularized problem; we test our algorithm and compare
its numerical performance with several recently proposed algorithms to show
that our scheme performs efficiently and stable. We analyze the convergence
properties of our algorithm, and in the last part of this paper, we will go
over some applications in reconstructing magnetic resonance images by using
our algorithm.
For further information please contact:
mittelmann@asu.edu
