A MODIFIED GAUSS-NEWTON ALGORITHM FOR FAST MICROWAVE IMAGING USING NEAR-FIELD PROBES

作者：Islam Md Asiful; Kiourti Asimina; Volakis John L

来源：Microwave and Optical Technology Letters, 2017, 59(6): 1394-1400.

DOI：10.1002/mop.30540

• Gauss-Newton algorithm
• industrial flow monitoring
• inverse problems
• microwave imaging

In this work, effort has been made to reduce the reconstruction time of microwave tomographic images using the conventional Gauss-Newton (GN) algorithm. As the forward scattering solution takes the most percentage of the total time, a modification to the GN algorithm by employing a Taylor series approximation of the forward problem is proposed. This helps to avoid the time consuming forward scattering solution required in each iteration, as such, the image reconstruction time is only limited by the time of the algebraic calculations involving matrices and vectors. Moreover, unlike most other works, in this paper, the imaging domain along with the actual transceivers are modeled by a full-wave finite element method based commercial software package which reduces the so called "model error." For validation of this method, a set of transceivers consisting of 8 dipole antennas operating at 1.2 GHz was fabricated on a flexible substrate and placed conformally around a cylindrical imaging domain. Simulation and measurements showed good agreement and fast image reconstruction. Overall, the proposed modified GN algorithm offers a fast imaging method towards developing high-speed conformal microwave imagers for continuous monitoring.

• 出版日期2017-6