基于PCNN图像分割与边缘匹配的支票验证方法  

A Check Verification Method Based on PCNN Segmentation and Edge Matching

作　　者：毛华[1] 章毅[1] 

机构地区：[1]电子科技大学计算机科学与工程学院计算智能实验室,四川成都610054

出　　处：《计算机工程与科学》2008年第11期38-41,共4页Computer Engineering & Science

摘　　要：本文提出了一种新的支票验证方法。该方法包括从复杂背景中提取印文的算法和一种改进的方形印章的配准方法。在印文分割中,首先采用基于脉冲耦合神经网络(PCNN)同步特性的细化算法,定位出轮廓的外围轮廓;然后根据外围轮廓的颜色信息,采用全局阈值分割方法将印文提取出来。对于方形印章,采用边缘匹配方法对印文进行由粗到精的配准。实验结果表明,本文提出的支票验证算法能够从复杂盖印背景中分割印文,并且得到较好的配准效果。This paper proposes a new method for check verification, It includes extracting the seal imprint from the complex background, and a modified registration method for rectangular seals. In the segmentation step, the outer frame is firstly obtained by the thinning algorithm based on the synchronous characteristics of PCNN. Then the whole seal is extracted by the global threshold method using the color information of the outer frame. For rectangular seals, the raw-to-fine edge matching method is used to register the two seals. The experimental results show that the proposed check verification method can extract the seal imprint from the complex background, and the registration method is effective.

关 键 词：印章验证 脉冲耦合神经网络 图像分割 边缘匹配 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]