DocumentCode
1203315
Title
An inexact-Newton method for short-range microwave imaging within the second-order Born approximation
Author
Estatico, Claudio ; Pastorino, Matteo ; Randazzo, Andrea
Author_Institution
Dept. of Math., Univ. of Genoa, Genova, Italy
Volume
43
Issue
11
fYear
2005
Firstpage
2593
Lastpage
2605
Abstract
A new approach to noninvasive inspection of dielectric targets at microwave frequencies is proposed. Cylindrical dielectric objects are reconstructed under the second-order Born approximation. A multi-illumination configuration is considered. The continuous model is discretized by the moment method and an efficient inexact-Newton method is applied. The dielectric profile is iteratively reconstructed starting from the measured scattered data, which are related to the unknown target through the inverse scattering equations written in a variational setting. Several numerical results are reported, which are aimed at assessing the capabilities of the approach in dealing with the nonlinear ill-posed inverse problem associated to the short-range microwave imaging. Single, multilayer, and separate cylinders are reconstructed in noiseless and noisy environments.
Keywords
Newton method; dielectric measurement; geophysical techniques; inverse problems; microwave measurement; remote sensing; cylindrical dielectric objects; dielectric measurement; inexact-Newton method; inverse scattering; iterative regularization; multiillumination configuration; nonlinear ill-posed inverse problem; second-order Born approximation; short-range microwave imaging; Approximation methods; Dielectrics; Image reconstruction; Inspection; Inverse problems; Microwave frequencies; Microwave imaging; Microwave theory and techniques; Scattering; Working environment noise; Dielectrics; Newton methods; imaging; inverse scattering; iterative regularization;
fLanguage
English
Journal_Title
Geoscience and Remote Sensing, IEEE Transactions on
Publisher
ieee
ISSN
0196-2892
Type
jour
DOI
10.1109/TGRS.2005.856631
Filename
1522620
Link To Document