基于RBF神经网络的电容层析成像图像重建算法

Image Reconstruction Algorithm Based on RBF Neural Networks for Electrical Capacitance Tomography

针对目前电容层析成像系统图像重建分辨率不高,精确度低的问题,提出了一种新的采用RBF神经网络对电容层析成像系统进行图像重建的方法.该神经网络采用改进的自适应遗传算法,优化选取隐层神经元的中心和宽度,用Tikhnov正则化方法训练网络权值.12电极的电容层析成像系统的仿真实验结果表明,该方法能明显改善成像质量,成像精确度较好,证明了该方法的有效性.

In view of the low precision of the reconstruction image of Electrical Capacitance Tomography （ECT） at present. A new method of Image reconstruction algorithm based on RBF neural networks for Electrical Capacitance Tomography is proposed. Adaptive genetic algorithm is used to optimize the centers and widths of the hidden units of RBF networks and the Tikhonov regularization method is used to train the weights of RBF networks. The simulation results for 12 -electrode electrical capacitance tomography system illustrate that the method can improve the reconstructtion image quality obviously and testify the effectiveness of the proposed method.