印章鉴定系统的图像处理研究  

Image Processing Technology in Seal Imprint Verification

作　　者：周琨[1] 张大力[1] 

机构地区：[1]清华大学自动化系,北京100084

出　　处：《计算机工程与应用》2002年第20期218-220,共3页Computer Engineering and Applications

摘　　要：在银行的印章鉴定系统中,正确识别票据印章的前提条件是得到高质量的预处理图像。预处理的主要过程是将印章图像三值化,去噪声并进行分割,然后将分割图像送入后续识别模块进行识别。由于票据模式的多样性,背景的复杂性和图像采集的任意性,使得图像的预处理面临较大的困难。文章主要研究了预处理过程中阈值的确定方法,以及去除印章区域复杂噪声的算法。实践证明,在去噪声算法中,通过多种滤波器的有效组合,能得到较高质量的处理图像,从而满足了识别算法的需要。In the seal imprint verification system of bank,obtaining high quality preprocessed image is a prerequisite to correctly verify the seal imprint of documents.The primary procedure of preprocessing is binarization, noise removal and image segmenation, and then segmented image is sent into the following recognition module to recognize.Because of the diversity of pattern,complexity of background and randomicity of sampling,it is difficult to acquire an ideal preprocessed result.In this paper,the threshold determination algorithm and complex noise removal method,a new way to solve this problem,are proposed.With the combination of multiple filters,noise can be suppressed to a very low level,thus satisfying the requirements of recognition algorithm.

关 键 词：印章鉴定系统 图像处理 阈值确定 噪声消除 模式识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]