Thursday, September 8, 9:30 am in room MTH 3206, University of Maryland,
College Park
Preconditioned Iterative Regularization
Dr. J. Nagy, Dept. of Computer Science, UMCP
nagy@cs.umd.edu
A preconditioned iterative regularization scheme is examined for solving
large scale structured linear systems g = Hf + eta, arising from the
discretization of ill-posed inverse problems in the presence of noise.
The case where H is a block Toeplitz matrix with Toeplitz blocks is
considered. The preconditioned conjugate gradient method is applied,
where H is approximated by a block circulant matrix with circulant
blocks, resulting in a fast 2-D FFT-based iterative scheme.
It is shown how the iterations can be effectively and efficiently
regularized for solving ill-posed problems by using the spectral
decomposition of the preconditioner. An application to image
restoration is presented.
Sept. 8, Numerical Analysis Seminar, U. of MD, College Park
|