基于四层BP神经网络的催化剂活性图像识别  

Image of Catalyst Activity Recognition Based on Four Layers BP Neural Network

作　　者：李含光[1] 赵晓瑜[1] 高宝建[1] 吴昊[1] 

机构地区：[1]南京信息工程大学计算机与软件学院,南京210044

出　　处：《武汉理工大学学报》2010年第16期68-71,共4页Journal of Wuhan University of Technology

摘　　要：在工业生产中,催化剂具有十分重要的作用,长期以来对其活性很难做准确的判别。针对这一难题,提出了利用图像处理和BP神经网络相结合的方法来判别催化剂活性,同时在神经网络的隐含层选取双隐层,构成了4层神经网络。详细描述了4层BP神经网络的算法、系统框架和主要技术。The catalyst plays an important role in industrial production,but it is very difficult to judge its activity for a long time.To solve this problem, the method of combining image processing and BP neural network is put forward to judge the catalyst activity,and choose two hidden layers to constitute a fourlayer neural network.In this paper,the algorithm of four layers BP neural network,system framework and main technique are described in detail.

关 键 词：4层BP神经网络 图像识别 纹理特征 共生矩阵 催化剂活性 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]