| 摘要: |
| 提出了一种基于改进U-Net（M-Net）模型的电磁逆散射算法.M-Net模型主要由多尺度输入层、U型卷积神经网络（CNN）、多尺度均值输出层组成.将散射场数据作为网络输入，能够在保证计算精度与计算效率的同时，减少人工计算工作量.以二维电介质为重构目标的仿真实验表明：与U-Net模型对比，应用M-Net模型求解电磁逆散射问题较为高效，输出结果误差更小. |
| 关键词: 电磁逆散射 卷积神经网络(CNN) U-Net模型 M-Net模型 |
| DOI：10.3969/J.ISSN.1000-5137.2023.02.010 |
| 分类号:TP391.4 |
| 基金项目: |
|
| Two-dimensional dielectric object reconstruction based on improved U-Net model |
|
JIN Ming, YANG Chunxia
|
|
College of Information, Mechanical and Electrical Engineering, Shanghai Normal University, Shanghai 201418, China
|
| Abstract: |
| An electromagnetic inverse scattering algorithm was proposed based on the improved U-Net(M-Net) model in this paper. M-Net model was mainly composed of multi-scale input layer, U-type convolutional neural network (CNN) and multi-scale mean output layer. By using the scattering field data as the network input, it could ensure the calculation accuracy and efficiency, and reduce the workload of manual calculation. The simulation experiments with the two-dimensional dielectric as the reconstruction target showed that the M-Net model had high efficiency in solving the electromagnetic inverse scattering problem and resulted in less error comparing with the U-Net model. |
| Key words: electromagnetic inverse scattering convolutional neural network (CNN) U-Net model M-Net model |
