ECT Image Reconstruction algorithm based on SENet dual-path multi-scale feature fusion

PDF(2420 KB)

Journal of Vibration and Shock ›› 2023, Vol. 42 ›› Issue (7) : 180-186.

MA Min， LI Jiwei

Author information +

History +

Abstract

To solve the problem that it is difficult for a single neural network to capture complex and deep capacitance vector features in the process of capacitance tomography image reconstruction, a electrical capacitance tomography image reconstruction algorithm (BSFF) based on squeeze-and-excitation networks(SENet) dual-path multi-scale feature fusion was proposed. Firstly, a multi-scale dense deep cavity convolution module was constructed to obtain a larger local receptive field and maintain a lower computational complexity, and multi-scale feature fusion was achieved to capture the multi-scale detail features of capacitance vector to enhance the representation ability of the model. Secondly, residual neural network is used to solve the degradation phenomenon in feature extraction of deep network, and SENet module is added to recalibrate the corresponding weight of the channel which the capacitance feature tensor belongs and calibrate the feature response. Finally, a two-channel multi-feature fusion hybrid model with bidirectional feature extraction capability is formed to better fit the nonlinear mapping relationship between capacitance tensor and dielectric constant. The Experimental results show that BSFF algorithm has higher image reconstruction quality and better robustness than Landweber iterative algorithm and CNN algorithm.

Key words

Electrical capacitance tomography / Image reconstruction / Multi-scale feature fusion / Residual neural network

Cite this article

EndNote

Ris (Procite)

Bibtex

Download Citations

MA Min， LI Jiwei. ECT Image Reconstruction algorithm based on SENet dual-path multi-scale feature fusion[J]. Journal of Vibration and Shock, 2023, 42(7): 180-186

References

[1] 王化祥. 电学层析成像技术[J]. 自动化仪表，2017, 38(5):1-6.
WANG Hua-xiang. Electrical Tomography Technology [J]. Process Automation Instrumentation, 2017, 38(5): 1-6.
[2] 马敏，王涛，王力. 基于深度学习的滑油监测方法研究[J]. 推进技术，2020, 41(5): 1159-1167.
MA Min, Wang Tao, Wang Li. Research on Monitoring Method of Lubricating Oil Based on Deep Learning [J]. Journal of Propulsion Technology, 2020, 41(5): 1159-1167.
[3] 汤政，雷刚，王天祥，等. 模型参数对卷积神经网络电容层析成像图像重建的影响[J]. 仪器仪表学报，2021, 42(10): 72-83.
TANG Zheng, LEI Gang, WANG Tian-xiang, et al. Effects of model parameters on image reconstruction of convolutional neural network electrical capacitance tomography [J]. Chinese Journal of Scientific Instrument, 2021, 42(10):72-83.
[4] LU Lian, TONG Guo-wei, GUO Ge,et al. Split Bregman iteration based reconstruction algorithm for electrical capacitance tomography [J]. Transactions of the Institute of Measurement and Control, 2019, 32(7): 2389-2399.
[5] XIE Meng-tao, YANG Hai-ma, ZHAO Kai,et al. Regularization parameter optimization based on the constraint of Landweber algorithm for electrical capacitance tomography [J]. Flow Measurement and Instrumentation, 2019, 69(c): 101620-101620.
[6] 马敏，孙颖，范广永. 基于深度信念网络的ECT图像重建算法[J]. 计量学报，2021, 42(04): 476-482.
MA Min, SUN Ying, FAN Guang-yong. ECT Image Reconstruction Algorithm Based on Depth Belief Network [J]. Acta Metrologica Sinica, 2021, 42(04): 476-482.
[7] Chu P, Lei J,. Regularization Reconstruction Method for Imaging Problems in Electrical Capacitance Tomography [J]. IOP Conference Series: Earth and Environmental Science, 2017, 94(1): 012127- 012133.
[8] Alfred J M. Comparative Performance Evaluation of Entropic Thresholding Algorithms Based on Shannon, Renyi and Tsallis Entropy Definitions for Electrical Capacitance Tomography Measurement Systems [J]. International Journal of Intelligent Systems and Applications, 2018, 10(4): 41-49.
[9] ZHENG J, peng L. An autoencoder-based image reconstruction for electrical capacitance tomography [J]. IEEE Sensors Journal, 2018, 18(13): 5464-5474.
[10] 付荣，张新宇，王子辰，等. 基于V-ResNet的电阻抗层析成像方法[J]. 仪器仪表学报，2021, 42(09):279-287
FU Rong, ZHANG Xin-yu, WANG Zi-chen. Electrical impedance tomography method based on V-ResNet [J]. Chinese Journal of Scientific Instrumen, 2021, 42(09): 279-287.
[11] 孙先亮，李健，韩哲哲，等. 基于数据驱动的卷积神经网络电容层析成像图像重建[J]. 化工学报，2020, 71(05):2004-2016.
SUN Xian-liang, LI Jian, HAN Zhe-zhe, et al. Data-driven image reconstruction of electrical capacitance tomography based on convolutional neural network [J]. CIESC Journal, 2020, 71(05): 2004-2016.
[12] 高飞，闫镔，陈健，等. 基于堆叠树形聚合结构空洞卷积的肝脏肿瘤分割[J]. 光学学报，2021, 41(18):81-92.
GA Fei, YAN Bin, CHEN Jian, et al. Liver Tumor Segmentation Based on Dilated Convolution of Stacked Tree Aggregation Structure [J]. Acta Optica Sinica, 2021, 41(18): 81-92.
[13] Lei J, Liu Q B, Y x,et al. Combination regularization reconstruction method for electrical capacitance tomography [J]. International Journal of Intelligent Systems and Applications, 2018, 59(3): 135-146.
[14] 邓飞跃，丁浩，郝如江. 基于多尺度特征融合残差神经网络的旋转机械故障诊断[J]. 振动与冲击，2021, 40(24):22-28+35.
DENG Fei-yue, DING Hao,HAO Ru-jiang. Fault diagnosis of rotating machinery based on residual neural network with multi-scale feature fusion [J]. Journal of Vibration and Shock, 2021, 40(24): 22-28+35.
[15] 刘盛，黄圣跃，程豪豪，等. 结合空间深度卷积和残差的大尺度点云场景分割[J]. 中国图象图形学报，2021, 26(12):2848-2859.
LIU Sheng, HUANG Sheng-yue, CHENG Hao-hao. A deep residual network with spatial depthwise convolution for large-scale point cloud semantic segmentation [J]. Journal of Image and Graphics, 2021, 26(12): 2848-2859.