印鉴自动识别算法研究(英文)  

A New Approach to Automatic Seal Imprint Identification

作　　者：郭田德[1] 许传祥[2] 高自友[3] 

机构地区：[1]中国科学院研究生院华罗庚应用数学与信息科学研究中心,北京100039 [2]北京大学数学学院,北京100871 [3]北方交通大学交通运输学院,北京100044

出　　处：《自动化学报》2004年第1期8-19,共12页Acta Automatica Sinica

基　　金：Hi-Tech Research and Development Program of China (863-2002AA103061) ; President Foundation of Graduate School of Chinese Academy of Sciences (Yzjj200105)

摘　　要：印鉴在许多东方国家被广泛使用了多年.印鉴自动识别是模式识别中的一项非常困难的课题.文中对计算机印鉴自动识别的各个阶段进行了仔细研究,并给出一系列相关算法.为了把“伪造印鉴多出的笔划部分”与“由于印泥渗出而多出的部分”区别开来以及将“伪造印鉴少出的笔划部分”与“由于印泥少而盖印不清晰的部分”区别开来,文中提出了全差图、内差图、多差图、少差图等一系列差图的概念,并对这些差图进行了处理.在此基础上,完成了一套完整的计算机印鉴自动识别方法.实验结果表明,该方法可能在实际应用中是可行的.Seal imprints have been extensively used in many oriental countries. The automatic i-dentification of seal imprints is a very difficult problem of pattern recognition. Every phase of the automatic identification of seal imprints is studied carefully and a corresponding series of algorithms is presented . In order to make a distinction between surplus parts of strokes in forgery seal imprint and surplus parts of seal paste in genuine seal imprint, and make a distinction between missing parts of strokes in forgery seal imprint and missing parts of seal paste in genuine seal imprint, several difference images and image processing methods are proposed, i. e.> full difference image, inner difference image, surplus difference image, missing difference image, etc. On such a basis, an automatic seal imprint identification scheme is presented. Experimental results confirm that the proposed scheme are feasible for practical applications.

关 键 词：印鉴识别 模式识别 特征提取 差图 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]