基于多尺度自适应网络的ECT图像重建  

Image Reconstruction of ECT Based on Multi-scale Adaptive Network

作　　者：张立峰[1] 常恩健 ZHANG Lifeng;CHANG Enjian(Department of Automation,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

出　　处：《计量学报》2024年第8期1139-1146,共8页Acta Metrologica Sinica

基　　金：国家自然科学基金(61973115)。

摘　　要：为求解电容层析成像(ECT)中的非线性病态反问题,提出了一种多尺度自适应网络(MSANet)模型,该模型实现了更精细维度上多尺度特征的融合,且模型参数量相对较小。通过在单个残差块内构建树状结构组成多尺度特征融合模块,MSANet实现了更好的鲁棒性和更低的计算参数量;采用加入自适应空间阈值模块方式,进一步提高了图像的重建精度。实验结果表明:与线性反投影(LBP)算法、Landweber迭代算法及常用深度学习方法相比,该方法平均相对误差最小且平均相关系数最大,分别为0.181及0.967。In order to solve the nonlinear ill-posed inverse problem of electrical capacitance tomography(ECT),a multi-scale adaptation network(MSANet)model is proposed,which achieves the fusion of multi-scale features in a more fine-grained dimension and has a relatively small number of model parameters.By constructing a tree structure within a single residual block to form a multi-scale feature fusion module,MSANet achieves more robustness and lower computational parameters.Furthermore,by adding an adaptive spatial threshold module,the reconstruction accuracy of the images is further improved.Compared with linear back projection(LBP)algorithm,Landweber iterative algorithm,and commonly used deep learning methods,this method has the smallest average relative error and the largest average correlation coefficient,with 0.181 and 0.967,respectively.

关 键 词：多相流测量 机器视觉 电容层析成像 图像重建 深度学习 多尺度 自适应网络 

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]