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.
出处
《计算机工程与应用》
CSCD
北大核心
2002年第20期218-220,共3页
Computer Engineering and Applications
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
Seal imprint verification,Image processing,Threshold determination,Noise removal
