Sparse Nonlinear Electromagnetic Imaging Accelerated With Projected Steepest Descent Algorithm
KAUST DepartmentComputer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division
Electrical Engineering Program
Online Publication Date2017-04-03
Print Publication Date2017-07
Permanent link to this recordhttp://hdl.handle.net/10754/623106
MetadataShow full item record
AbstractAn efficient electromagnetic inversion scheme for imaging sparse 3-D domains is proposed. The scheme achieves its efficiency and accuracy by integrating two concepts. First, the nonlinear optimization problem is constrained using L₀ or L₁-norm of the solution as the penalty term to alleviate the ill-posedness of the inverse problem. The resulting Tikhonov minimization problem is solved using nonlinear Landweber iterations (NLW). Second, the efficiency of the NLW is significantly increased using a steepest descent algorithm. The algorithm uses a projection operator to enforce the sparsity constraint by thresholding the solution at every iteration. Thresholding level and iteration step are selected carefully to increase the efficiency without sacrificing the convergence of the algorithm. Numerical results demonstrate the efficiency and accuracy of the proposed imaging scheme in reconstructing sparse 3-D dielectric profiles.
CitationDesmal A, Bagci H (2017) Sparse Nonlinear Electromagnetic Imaging Accelerated With Projected Steepest Descent Algorithm. IEEE Transactions on Geoscience and Remote Sensing: 1–13. Available: http://dx.doi.org/10.1109/TGRS.2017.2681184.