Kohonen网络在电容CT图像重建中的应用

Kohonen网络在电容CT图像重建中的应用 APPLICATION OF KOHONEN NETWORK IN ECT IMAGE RECONSTRUCTION 潘敏 Pan Ming 1 first-author 卢振陶 Lu Zhentao 1 苏祥芳 Su Xiangfang 1 王延平 Wang Yanping 1 武汉大学电子信息学学院 武汉大学电子信息学学院 College of Electronic Information,Wuhan University College of Electronic Information,Wuhan University 提出一种采用Ｋｏｈｏｎｅｎ网络算法进行１２电极电容ＣＴ系统的图像重建的方法．该网络由输入和输出两层神经元组成，输入为１２电极电容ＣＴ系统测量所得的６６个电容值，输出对应管道介质的空间分布模式．该网络采用竞争算法无监督学习，实验结果证明，该网络有较强的抗噪声能力，可重建出较精确的介质分布图像 A method about using self organizing featrue maps neural network to rebuild images of 12 electrode ECT system is described.The network is make up of two layers:input layer and outpur layer.The input is 66 capacitance data measured from 12 electrode ECT system,and output corresponds to the distributing modes of media in the pipeline.The network adopts competition algorithm to study without teaching.The test results show that the network has good noise immunity and can rebuild precise images of media distribution. ECT Kohonen网络 自组织特征映射 竞争算法 ECT Kohonen network self organizing feature maps competition algorythm TP18 国家自然科学基金(61672102) 1999-01-01 2021-04-01 1