一种增强型稠密连接网络模型应用于ECT图像重建

摘要
图/表
参考文献(20)
相关文章 (2)

全文: PDF (1781 KB) HTML (1 KB)
输出: BibTeX | EndNote (RIS)

摘要针对传统神经网络在电容层析成像图像重建中，存在难以较好地实现电容特征张量的底层位置特征和顶层语义特征融合的问题，提出了一种增强型稠密连接网络模型(EDC-net)。首先，训练全连接神经网络获得初步介电常数分布，利用全连接神经网络的输出特征图作为补偿型U-net网络的输入。其次，搭建补偿型U-net网络，在编码和解码器间添加类DenseNet结构的稠密跳跃连接机制，以保留大量的底层位置特征信息，减少模型多个输出节点的特征损失；同时利用多尺度密集空洞卷积模块替代补偿型U-net中普通卷积，扩大模型的感受野并丰富多尺度信息。最后采用高效通道注意力机制模块实现子解码器节点的输出特征跨通道交互，增强模型对重要信息的关注度，提高模型的非线性拟合能力；实验结果表明，与Landweber迭代算法、U-net算法对比，基于EDC-net算法的重建图像分辨率高，成像边缘清晰，且更具鲁棒性。

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	马敏
	孙妮

关键词 ：电容层析成像, 图像重建, 稠密跳跃连接, 特征跨通道交互

Abstract：In order to solve the problem that traditional neural networks cannot integrate the bottom position features and the top semantic features of the capacitive feature tensor well in ECT image reconstruction, an enhanced dense connection network model is proposed. First, the initial dielectric constant distribution is obtained by training the Fully Connected Neural network, and the output characteristic map of the fully connected neural network is used as the input of the compensated U-net network. Secondly, a compensated U-net network is built, and a DenseNet-like dense jump connection mechanism is added between the encoder and decoder to retain a large amount of underlying location feature information and reduce the feature loss of multiple output nodes of the model. At the same time, the multi-scale dense cavity convolutional module is used to replace the ordinary convolution in the compensated U-net to enlarge the receptive field of the model and enrich the multi-scale information. Finally, an efficient channel attention mechanism module is used to realize the cross-channel interaction of the output features of sub-decoder nodes, which enhances the model's attention to important information and improves the nonlinear fitting ability of the model. The experimental results show that the reconstructed images based on this algorithm have higher resolution, clearer imaging edges, and more robustness than the Landweber iterative algorithm and the U-net algorithm.

Key words： Electrical capacitance tomography Image reconstruction Dense jump connections Feature cross-channel interactions

收稿日期: 2023-07-12 出版日期: 2024-05-28

引用本文:

马敏，孙妮. 一种增强型稠密连接网络模型应用于ECT图像重建[J]. 振动与冲击, 2024, 43(10): 82-88.
MA Min，SUN Ni. Enhanced dense connectivity network model applied to ECT image reconstruction. JOURNAL OF VIBRATION AND SHOCK, 2024, 43(10): 82-88.

链接本文:

http://jvs.sjtu.edu.cn/CN/ 或 http://jvs.sjtu.edu.cn/CN/Y2024/V43/I10/82

[1] 王化祥. 电学层析成像技术[J]. 自动化仪表，2017，38(05):1-6. WANG Hua-xiang. Electrical Tomography Technology [J]. Process Automation Instrumentation, 2017, 38(05): 1-6. [2] Xia Z H, Cui Z Q, Chen Y X, et al. Generative adversarial networks for dual-modality electrical tomography in multi-phase flow measurement[J]. Measurement, 2020, 173(24):108608. [3] 温银堂，赵丽梅，张玉燕，等. 基于ECT的复合材料构件胶层缺陷检测[J]. 仪器仪表学报，2015，36(08):1783-1791. WEN Yin-tang, WEN Li-mei, ZHANG Yu-yan, et al. Defect detection of the adhesive layer of composite component based on the ECT [J]. Technology Chinese Journal of Scientific Instrument, 2015, 36(08):1783-1791. [4] 马敏，于洁. 基于改进联合稀疏电容层析成像算法滑油监测研究[J]. 推进技术，2022，43(04):345-351. MA Min, YU Jie. Lubricating Oil Monitoring Based on Improved Joint Sparse Electrical Capacitance Tomography Algorithm [J]. Journal of Propulsion Technology, 2022, 43(04):345-351. [5] 段晨龙，刘锡波，周晨阳，等. 基于电容层析成像技术(ECT)对干法重介流化床中分离机制的研究[J]. 煤炭学报，2022，47(02):945-957. DUAN Chen-long, LIU Xi-bo, ZHOU Chen-yang, et al. Research on the separation mechanism of dry dense medium separation fluidized bed based or electrical capacitance tomography[J]. Journal of Coal Science, 2022, 47(02):945-957. [6] Wang C, Guo Q, Wang H X, et al. ECT image reconstruction based on alternating direction approximate Newton algorithm[J]. IEEE Transactions on Instrumentation and Measurement, 2020, 69(7): 4873–4886. [7] Cui L Q, Tian P, Wang C, et al. Iterative weighted fidelity and hybrid regularization-based algorithm for ECT[J]. Measurement Science and Technology, 2022, 33(6) :065404. [8] 郭志恒，律德财，邵富群，等. 基于特征值谱区域优化的Tikonov正则化实现ECT图像重建[J]. 中国电机工程学报，2021，41(19):6641-6649. GUO Zhi-heng, LV De-cai, SHAO Fu-qun, et al.Realization of ECT Image Reconstruction by Tikonov Regularization Technology Based on Eigenspectrum Region Optimization[J]. Proceedings of the CSEE, 2021, 41(19):6641-6649. [9] Tian Z N, Gao X X, Qiu L M, et al. Experimental imaging and algorithm optimization based on deep neural network for electrical capacitance tomography for LN2-VN2 flow[J]. Cryogenics, 2022, 127:103568. [10] Yang Y J, Peng L H. Data Pattern With ECT Sensor and Its Impact on Image Reconstruction[J]. IEEE Sensors Journal, 2013, 13(5):1582-1593. [11] 马敏，李继伟. 基于SENet双路径多尺度特征融合的ECT图像重建[J]. 振动与冲击，2023，42(07):180-186. MA Min, LI Ji-wei. ECT Image Reconstruction based on SENet dual path multi-scale feature fusion[J]. Journal of Vibration and Shock, 2023, 42(07):180-186. [12] Zhang L F, Zhang M H. Image reconstruction method for electrical capacitance tomography using adaptive simulated annealing algorithm[J]. Review of Scientific Instruments, 2021, 92(10):105009. [13] Ye J M, Mao M X, Wang H G, et al. Image Reconstruction for ECT Based on Extended Sensitivity Matrix [J]. IEEE Sensors Journal, 2016, 16(8):2466-2476. [14] Yang X M, Zhao C J, Chen B, et al. Big Data driven U-Net based Electrical Capacitance Image Reconstruction Algorithm [C]//2019 IEEE International Conference on Imaging Systems and Techniques (IST). Abu Dhabi, United Arab Emirates: IST. 2019. 1-6. [15] 任子茂，路慧敏，冯丽雅，等. 基于全连接神经网络的自然光谱复现方法研究[J]. 光学学报，2023，43(10):277-285. REN Zi-mao, LU Hui-min, FENG Li-ya, et al. A Method for Natural Spectral Reproduction Based on Fully Connected Neural Network[J]. Acta Optica Sinica, 2023, 43(10):277-285. [16] 邸拴虎，杨文瀚，廖苗，等. 基于RA-Unet的CT图像肝脏肿瘤分割[J]. 仪器仪表学报，2022，43(08):65-72. DI Shuang-hu, YANG Wen-han, LIAO Miao, et al. Liver tumor segmentation from CT images based on RA-Unet[J]. Chinese Journal of Scientific Instrument, 2022, 43(08):65-72. [17] 吕少云，李佳田，阿晓荟，等. Res＿ASPP＿UNet++：结合分离卷积与空洞金字塔的遥感影像建筑物提取网络[J]. 遥感学报，2023，27(02):502-519. LV Shao-yun, LI Jia-tian, A Xiao-hui, et al. Res_ASPP_UNet++: Building an extraction network from remote sensing imagery combining depthwise separable convolution with atrous spatial pyramid pooling[J]. National Remote Sensing Bulletin, 2023, 27(02):502-519. [18] Li F, Tan C, Dong F, et al. V-Net Deep Imaging Method for Electrical Resistance Tomography[J]. IEEE Sensors Journal, 2020, 20 (12):6460-6469. [19] Gao X X, Tian Z N, Qiu L M, et al. A hybrid deep learning model for ECT image reconstruction of cryogenic fluids[J]. Flow Measurement and Instrumentation, 2022, 87:102228. [20] Wang Q L, Wu B G, Zhu P F, et al. ECA-Net: Efficient Channel Attention for Deep Convolutional Neural Networks[C] //2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Seattle, WA, USA: CVF. 2020.11531-11539.