一种基于改进U-Net模型的电磁层析成像算法An electromagnetic tomography algorithm based on improved U-Net model
李秀艳,李明廷,李晓捷,王琦,张荣华,汪剑鸣
摘要(Abstract)：
为解决电磁层析成像（electromagnetic tomography,EMT）传统成像算法由于逆问题的不适定性和病态性导致重建图像质量差的问题，提出了一种基于改进U-Net深度网络模型的新型电磁层析成像方法。首先，以UNet深度网络模型为基础，加入残差模块使网络提取更多特征信息并避免网络训练时梯度消失的问题；其次在此结构上引入注意力机制来提升重要特征信息，抑制无用的特征信息，加强对缺陷边缘和形状特征的权重分配。通过仿真和金属缺陷检测实验评估了本文所提出算法的性能，并与线性反投影算法和共轭梯度算法进行了对比。仿真实验和金属缺陷检测实验结果表明：本文提出的算法在精确率、召回率和F1-Score分别达到88.41%、90.38%和89.38%，重建图像对于缺陷位置和形状的预测更为准确。
关键词(KeyWords)： 电磁层析成像;深度学习;图像重建算法;U-Net网络
基金项目(Foundation): 国家自然科学基金资助项目（61872269,61601324,61903273）;; 天津市自然科学基金资助项目（18JCYBJC85300）;; 天津科技计划项目（19PTZWHZ00020）
作者(Author): 李秀艳,李明廷,李晓捷,王琦,张荣华,汪剑鸣
参考文献(References)：
- [1]霍继伟，刘泽，王成飞，等.基于幅值特征提取的混频电磁钢轨探伤方法[J].电子测量技术，2020,43(13):105-110.HUO J W,LIU Z,WANG C F,et al. Multi-frequency electromagnetic railway inspection based on amplitude feature extraction algorithm[J]. Electronic Measurement Technology,2020,43(13):105-110(in Chinese).
- [2] RAMLI A S. Feasibility study of planar-array electromagnetic inductance tomography(EMT)[J]. Contribution in Book/report/proceedings,1999(1):14-17.
- [3]刘泽，薛芳其，杨国银，等.电磁层析成像灵敏度矩阵实验测试方法[J].仪器仪表学报，2013,34(9):1982-1988.LIU Z,XUE F Q,YANG G Y,et al. Experimental measurement method of sensitivity matrix for electromagnetic tomography[J]. Chinese Journal of Scientific Instrument,2013,34(9):1982-1988(in Chinese).
- [4] YIN W L,PEYTON A J. A planar EMT system for the detection of faults on thin metallic plates[J]. Measurement Science and Technology,2006,17(8):2130-2135..
- [5]王琦，崔莉莎，汪剑鸣，等.基于电磁层析成像的金属缺陷稀疏成像方法[J].仪器仪表学报，2017,38(9):2291-2298.WANG Q,CUI L S,WANG J M,et al. Defects detection based on electromagnetic tomography for sparse imaging method[J].Chinese Journal of Scientific Instrument,2017,38(9):2291-2298(in Chinese).
- [6] YANG W Q,PENG L. Image reconstruction algorithms for electrical capacitance tomography[J]. Measurement Science and Technology,2002,14(1):14.
- [7]王莉莉，陈宇，陈德运，等.基于改进信赖域的电容层析成像图像重建算法[J].仪器仪表学报，2010,31(5):1076-1081.WANG L L,CHEN Y,CHEN D Y,et al. Improved trust region based image reconstruction algorithm for electrical capacitance tomography system[J]. Chinese Journal of Scientific Instrument,2010,31(5):1076-1081(in Chinese).
- [8]刘向龙.电磁层析成像关键问题及其在高铁车轮探伤中的应用研究[D].北京：北京交通大学，2019.LIU X L. Research on key issues of electromagnetic tomography and its application in high-speed rail wheel defect detection[D]. Beijing:Beijing Jiaotong University,2019(in Chinese).
- [9] XIAO J,LIU Z,ZHAO P F,et al. Deep learning image reconstruction simulation for electromagnetic tomography[J]. IEEE Sensors Journal,2018,18(8):3290-3298.
- [10]徐明达.基于自编码器神经网络的EMT图像重建算法研究[D].沈阳：辽宁大学，2020.XU M D. Research on EMT image reconstruction algorithm based on auto encoder neural network[D]. Shenyang:Liaoning University,2020(in Chinese).
- [11] LIU Z Q,CAO Y W,WANG Y Z,et al. Computer visionbased concrete crack detection using U-net fully convolutional networks[J]. Automation in Construction,2019,104:129-139.
- [12] JOHNSON J,KARPATHY A,LI F F. DenseCap:Fully convolutional localization networks for dense captioning[C]//2016IEEE Conference on Computer Vision and Pattern Recognition(CVPR). Las Vegas,NV,USA:IEEE,2016:4565-4574.
- [13]贾树开.深度学习在图像分割中的应用：基于深度学习的甲状腺结节超声图像分割[D].成都：电子科技大学，2020.JIA S K. The application of deep learning in image segmentation-segmentation of ultrasound images of thyroid nodules based on deep learning[D]. Chengdu:University of Electronic Science and Technology of China,2020(in Chinese).
- [14]殷晓航，王永才，李德英.基于U-Net结构改进的医学影像分割技术综述[J].软件学报，2021,32(2)::519-550.YIN X H,WANG Y C,LI D Y. Suvery of medical image segmentation technology basedon U-Net structure improvement[J].Journal of Software,2021,32(2):519-550(in Chinese).
- [15] HAMILTON S J,HAUPTMANN A. Deep D-bar:Real-time electrical impedance tomography imaging with deep neural networks[J]. IEEE Transactions on Medical Imaging,2018,37(10):2367-2377.
- [16]叶明，李晓丞，刘凯，等.一种基于U2-Net模型的电阻抗成像方法[J].仪器仪表学报，2021,42(2):235-243.YE M,LI X C,LIU K,et al. Image reconstruction method for electrical impedance tomography using U2-Net[J]. Chinese Journal of Scientific Instrument,2021,42(2):235-243(in Chinese).
- [17] HE K,ZHANG X,REN S,et al. Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition(CVPR),Las Vegas,NV,USA:IEEE,2016:770-778.
- [18] RONNEBERGER O,FISCHER P,BROX T. U-Net:convolutional networks for biomedical image segmentation[C]//International Conference on Medical Image Computing and Computer-Assisted Intervention. Springer International Publishing.Munich,Germany:Springer,2015:234-241.
- [19] HU J,SHEN L,SUN G. Squeeze-and-excitation networks[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City,UT,USA:IEEE,2018:7132-7141.
- [20] HU J,SHEN L,SUN G,et al. Squeeze-and-excitation networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2020,42(8):2011-2023.