ANN-Assisted CoSaMP Algorithm for Linear Electromagnetic Imaging of Spatially Sparse Domains

Greedy pursuit algorithms (GPAs) are widely used to reconstruct sparse signals. Even though many electromagnetic (EM) inverse scattering problems solved on investigation domains, GPAs have rarely been for this purpose. This is because (i) they require a priori knowledge of the sparsity level in domain, which often not available EM imaging applications, and (ii) matrix does satisfy restricted isometric property. In work, these challenges respectively addressed by using an artificial neural network (ANN) estimate level, adding Tikhonov regularization term diagonal elements matrix. These enhancements permit compressive sampling matching (CoSaMP) algorithm be efficiently solve two-dimensional problem, linearized Born approximation, spatially domains. Numerical results, demonstrate efficiency applicability proposed ANN-enhanced CoSaMP algorithm, provided.