A Regularization Parameter in Discrete Ill-Posed Problems
SIAM Journal on Scientific Computing
Rank-deficient and discrete ill-posed problems: numerical aspects of linear inversion
Rank-deficient and discrete ill-posed problems: numerical aspects of linear inversion
Hi-index | 7.29 |
Kirsch's factorization method is a fast inversion technique for visualizing the profile of a scatterer from measurements of the far-field pattern. We present a Tikhonov parameter choice approach based on a maximum product criterion (MPC) which provides a regularization parameter located in the concave part of the L-curve on a log-log scale. The performance of the method is evaluated by comparing our reconstructions with those obtained via the L-curve, Morozov's discrepancy principle and the SVD-tail. Numerical results that illustrate the effectiveness of the MPC in reconstruction problems involving both simulated and real data are reported and analyzed.