利用多准则Hopfield网络对ECT进行图像重建

针对电容层析成像技术中图像重建质量较差的问题,提出一种基于多准则Hopfield网络模型的电容层析成像的改进图像重建算法。首先分析了ECT图像重建和Hopfield网络的基本原理,然后根据ECT图像重建的特点确定了4种准则函数:图像熵、测量电容和估计电容的误差平方和、重建图像的局部非均匀性函数和总变差,并将这4种准则函数引入Hopfield网络的能量函数中,由此推导出Hopfield网络的动态方程,在此基础上得到ECT图像重建迭代算法,最后通过仿真实验对所提方法进行了验证。仿真实验结果表明利用此方法获得的重建图像误差和相关系数比LBP算法和Landweber迭代算法得到的相应指标要好。由此可见,该方法是一种有效的、精确度较高的ECT图像重建方法。

Image Reconstruction by using Multi-criteria of Hopfield Network for ECT

In view of low image reconstruction quality of electrical capacitance tomography ( ECT ) , an image reconstruction algorithm of ECT based on multi-criteria of Hopfield network is put forward.Firstly, the basic principle of ECT image reconstruction and Hopfield network was analyzed.Secondly, according to the characteristics of the ECT image reconstruction, four criterion functions were determined,including image entropy, the error sum of squares between measuring capacitances and estimating capacitances, lo-cal heterogeneity function of reconstruction image, and total variation of reconstruction image.These four criteria functions were introduced to the energy function of Hopfield network,and thus the dynamic equa-tions of Hopfield networks were deduced.On this basis, the iterative algorithm for ECT image reconstruc-tion was gained.Finally, the proposed method was verified through simulation experiment.Simulation re-sults show that the error of image reconstruction and correlation coefficient are better than corresponding indicators obtained by LBP and Landweber iterative algorithm.This ECT image reconstruction method is proven to be highly effective and accurate.It also provides an effective method and means for ECT image reconstruction algorithm.