A residual fully convolutional network (Res-FCN) for electromagnetic inversion of high contrast scatterers at an arbitrary frequency within a wide frequency band
Hao-Jie Hu, Jiawen Li, Li-Ye Xiao, Yu Cheng, Qing Huo Liu
{"title":"A residual fully convolutional network (Res-FCN) for electromagnetic inversion of high contrast scatterers at an arbitrary frequency within a wide frequency band","authors":"Hao-Jie Hu, Jiawen Li, Li-Ye Xiao, Yu Cheng, Qing Huo Liu","doi":"10.1088/1361-6420/ad4171","DOIUrl":null,"url":null,"abstract":"\n Many successful machine learning methods have been developed for electromagnetic inverse scattering problems. However, so far, their inversion has been performed only at the specifically trained frequencies. To make the machine learning based inversion method more generalizable for realistic engineering applications, this work proposes a residual fully convolutional network (Res-FCN) to perform EM inversion of high contrast scatterers at an arbitrary frequency within a wide frequency band. The proposed Res-FCN combines the advantages of the Res-Net and the fully convolutional network (FCN). Res-FCN consists of an encoder and a decoder: the encoder is employed to extract high-dimensional features from the measured scattered field through the residual frameworks, while the decoder is employed to map from the high-dimensional features extracted by the encoder to the electrical parameter distribution in the inversion region by the up-sample layer and the residual frameworks. Four numerical examples verify that the proposed Res-FCN can achieve good performance in the 2-D EM inversion problem for high contrast scatterers with anti-noise ability at an arbitrary frequency point within a wide frequency band.","PeriodicalId":50275,"journal":{"name":"Inverse Problems","volume":null,"pages":null},"PeriodicalIF":2.0000,"publicationDate":"2024-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Inverse Problems","FirstCategoryId":"100","ListUrlMain":"https://doi.org/10.1088/1361-6420/ad4171","RegionNum":2,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"MATHEMATICS, APPLIED","Score":null,"Total":0}
引用次数: 0
Abstract
Many successful machine learning methods have been developed for electromagnetic inverse scattering problems. However, so far, their inversion has been performed only at the specifically trained frequencies. To make the machine learning based inversion method more generalizable for realistic engineering applications, this work proposes a residual fully convolutional network (Res-FCN) to perform EM inversion of high contrast scatterers at an arbitrary frequency within a wide frequency band. The proposed Res-FCN combines the advantages of the Res-Net and the fully convolutional network (FCN). Res-FCN consists of an encoder and a decoder: the encoder is employed to extract high-dimensional features from the measured scattered field through the residual frameworks, while the decoder is employed to map from the high-dimensional features extracted by the encoder to the electrical parameter distribution in the inversion region by the up-sample layer and the residual frameworks. Four numerical examples verify that the proposed Res-FCN can achieve good performance in the 2-D EM inversion problem for high contrast scatterers with anti-noise ability at an arbitrary frequency point within a wide frequency band.
期刊介绍:
An interdisciplinary journal combining mathematical and experimental papers on inverse problems with theoretical, numerical and practical approaches to their solution.
As well as applied mathematicians, physical scientists and engineers, the readership includes those working in geophysics, radar, optics, biology, acoustics, communication theory, signal processing and imaging, among others.
The emphasis is on publishing original contributions to methods of solving mathematical, physical and applied problems. To be publishable in this journal, papers must meet the highest standards of scientific quality, contain significant and original new science and should present substantial advancement in the field. Due to the broad scope of the journal, we require that authors provide sufficient introductory material to appeal to the wide readership and that articles which are not explicitly applied include a discussion of possible applications.