基于深层小波网络的电容层析成像图像重建  

Image Reconstruction of Electrical Capacitance Tomography Based on Deep Wavelet Networks

作　　者：张立峰[1] 钱立凤 华回春 刘帅 ZHANG Lifeng;QIAN Lifeng;HUA Huichun;LIU Shuai(Department of Automation,North China Electric Power University,Baoding,Hebei 071003,China;Department of Mathematics and Physics,North China Electric Power University,Baoding,Hebei 071003,China)

机构地区：[1]华北电力大学自动化系,河北保定071003 [2]华北电力大学数理系,河北保定071003

出　　处：《计量学报》2024年第9期1353-1359,共7页Acta Metrologica Sinica

基　　金：国家自然科学基金(61973115)。

摘　　要：提出了一种基于深层小波网络的电容层析成像(ECT)图像重建算法,采用Landweber算法生成初始重建图像作为网络输入;以U-Net深度卷积神经网络模型为骨干模型,通过在上、下采样层引入小波变换提取不同层次的特征,以及采用跳跃连接方式搭建高频特征传递通道,保留更多的细节信息,充分利用特征图中的全局和局部信息特征。仿真及静态实验结果均表明,基于该算法的图像重建精度更高,仿真及静态实验重建图像的平均相对图像误差分别为0.1918及0.6570,平均相关系数分别为0.9685及0.8169。The image reconstruction algorithm of electrical capacitance tomography(ECT)based on deep wavelet network is presented.The Landweber algorithm is used to generate the initial reconstructed image as the input of the network.Taking the U-Net deep convolutional neural network model as the backbone model,the wavelet transform is introduced into the upper and lower sampling layers to extract the features of different levels and the high-frequency feature transfer channel is built through a skip connection to retain more detailed information and make full use of global and local information features in the feature map.Both simulation and experimental results show that the proposed image reconstruction algorithm has higher image reconstruction accuracy.The average relative image errors of simulated and static experimental reconstructed images were 0.1918 and 0.6570,respectively,with average correlation coefficients of 0.9685 and 0.8169.

关 键 词：多相流测量 电容层析成像 图像重建 深度学习 小波变换 

分 类 号：TB937[一般工业技术—计量学] TB973[机械工程—测试计量技术及仪器]