矩阵码的自组织特征映射识别研究  

Matrix Code Recognition with Self-Organization Feature Mapping Network

作　　者：胡友兰[1] 胡俊翘[1] 黄树槐[1] 王运赣[1] 

机构地区：[1]华中理工大学机械科学与工程学院

出　　处：《华中理工大学学报》1996年第3期17-19,共3页Journal of Huazhong University of Science and Technology

摘　　要：基于Ｋｏｈｏｎｅｎ自组织特征映射网的自组织性、自适应性及对输入信号的高容错能力，运用无导师动态竞争学习方法对较模糊的矩阵码进行识别，取得了较好的效果．对网络的训练采用了带噪声的矩阵码数据，大大提高了系统的辨识能力．试验表明，该方法为矩阵码的识别提供了一条有效的途径．atrix code recognition is the key technique of the applied matrix code system. By using the self-organization, self adaptation and high fault-tolerant capability to input signals of the Kohonen self-organization feature mapping network and with the unsupervised dynamic competitive learning method, good results of the recognition of the obscure matrix code have been obtained.As the network is trained with matrix code data containing noise,the identifying ability of the system is greatly improved. Experimental results show that the method provides an effective approach to recognizing the matrix code.

关 键 词：图像识别 矩阵码 神经网络 自组织映射 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]