一种基于神经网络判断的印鉴自动鉴别新方法  

A new automatic seal imprint verificationmethod based on neural networks

作　　者：常军民[1] 陈国定[1] 

机构地区：[1]浙江工业大学信息工程学院,浙江杭州310032

出　　处：《浙江工业大学学报》2005年第3期268-271,共4页Journal of Zhejiang University of Technology

摘　　要：印鉴鉴别是一个比较困难的课题,在文章中,分析了防伪印鉴的特点,从图像纹理和分析的角度出发,提出了一种基于印鉴边缘纹理的印鉴鉴别方法.主要包括印鉴图像的预处理,印鉴图像纹理特征的提取与选择及特征参数的分析比较.在此基础上,运用神经网络的判别方法对所提取的印鉴特征进行判别,以鉴别印鉴的真伪,该方法的算法简单,并经实验验证表明,该方法是从一种新的角度进行了印鉴鉴别,提出了鉴别印鉴的新途径.<Abstrcat> Seal imprint identification is a difficult topic. In this paper, features of seal imprint are analyzed, and a new scheme for automatic seal imprint identification using image texture and pattern recognition is presented. The major works includes the seal imprint preprocessing, computing, the texture features selection, and the features analysis and comparison. With this understanding, the features are judged and sorted by using neural networks. Main advantages of this method are that its algorithm is simple and this identification method is newly tried from a different angle. What is more, a new way of seal imprint identificaton is presented.

关 键 词：印鉴鉴别 纹理分析 灰度共生矩阵 神经网络 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]