Physics-Driven Neural Network for Solving Electromagnetic Inverse Scattering Problems

TL;DR

This paper introduces a physics-driven neural network approach for electromagnetic inverse scattering problems that improves reconstruction accuracy and stability without relying heavily on training data, by iteratively updating solutions based on physical constraints.

Contribution

The paper presents a novel physics-driven neural network method that avoids data dependence and enhances stability and accuracy in electromagnetic inverse scattering reconstructions.

Findings

High reconstruction accuracy demonstrated in numerical tests.

Strong stability even with complex lossy scatterers.

Efficient imaging through subregion identification.

Abstract

In recent years, deep learning-based methods have been proposed for solving inverse scattering problems (ISPs), but most of them heavily rely on data and suffer from limited generalization capabilities. In this paper, a new solving scheme is proposed where the solution is iteratively updated following the updating of the physics-driven neural network (PDNN), the hyperparameters of which are optimized by minimizing the loss function which incorporates the constraints from the collected scattered fields and the prior information about scatterers. Unlike data-driven neural network solvers, PDNN is trained only requiring the input of collected scattered fields and the computation of scattered fields corresponding to predicted solutions, thus avoids the generalization problem. Moreover, to accelerate the imaging efficiency, the subregion enclosing the scatterers is identified. Numerical and experimental results demonstrate that the proposed scheme has high reconstruction accuracy and strong stability, even when dealing with composite lossy scatterers.