Performance Analysis and Dynamic Evolution of Deep Convolutional Neural Network for Electromagnetic Inverse Scattering

The solution of electromagnetic (EM) inverse scattering problems is hindered by challenges, such as ill-posedness, nonlinearity, and high computational costs. Recently, deep learning was shown to be a promising tool in addressing these challenges. In particular, it is possible to establish a connection between a deep convolutional neural network (CNN) and iterative solution methods of EM inverse scattering. This led to the development of an efficient CNN-based solution to EM inverse problems, termed DeepNIS. It has been shown that DeepNIS can outperform conventional inverse scattering solution methods in terms of both the image quality and computational time. In this letter, we evaluate the DeepNIS performance as a function of the number of layers using structure similarity index measure and mean square error metrics. In addition, we probe the dynamic evolution behavior of DeepNIS by examining its near-isometry property and show that, after a proper training stage, the proposed CNN has near-optimal stability properties.