基于FCM和BP神经网络的棉麻纤维识别方法研究  

Study on method of cotton and bast fiber identification based on FCM and BP neural network

作　　者：殷士勇[1,2] 王文中[3] 

机构地区：[1]华东理工大学计算机科学与工程系,上海200237 [2]盐城工业职业技术学院机电工程系,盐城224005 [3]盐城工业职业技术学院财务处,盐城224005

出　　处：《黑龙江大学自然科学学报》2013年第3期405-409,共5页Journal of Natural Science of Heilongjiang University

基　　金：江苏省科技厅资助项目(BN2011056)

摘　　要：提出一种基于模糊c均值(FCM)和BP神经网络的棉麻纤维识别方法。首先,根据纤维横向和纵向截面形态的不同,提取6个特征参数,然后运用模糊c均值算法将样本聚类成3类,再将聚类后的数据作为BP神经网络的输入进行训练和预测,最后进行仿真实验。结果表明,将两种算法结合起来用于纤维的识别具有明显优势,是值得推广的纤维识别方法。A method for cotton and bast fiber identification based on fuzzy c-means （FCM） and BP neural network is proposed. Firstly, according to the different forms of fiber transverse and longitudinal, six characteristic parameters have been extracted, then using the fuzzy c-mean algorithm, the samples has been clustered into 3 categories, and then the data clustered has been trained and forecasted as the input of BP neural network, at last the simulation experiment has been done. The experimental results show that the two combined algorithms has obvious advantages to fiber recognition, and the fiber identification method is worthy of popularization.

关 键 词：FCM 模糊聚类 BP神经网络 棉麻纤维 纤维识别 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] TS101.1[自动化与计算机技术—计算机科学与技术]