机构地区: 五邑大学计算机学院
出 处: 《数字技术与应用》 2012年第11期121-122,124,共3页
摘 要: 文中提出了一种票据印鉴预处理、特征提取的方法。该方法包括从包含文本、数字的票据中对印鉴图像的二值化、边缘检测、形态学处理、特征提取等;利用每种印鉴拥有固定的面积大小,避免了复杂背景下的噪声干扰和人为因素的干扰,实验数据表明在误识率和快速性之间具有良好的平衡性和算法的鲁棒性,有利于票据数字化存储。 In this paper a method of preprocess and feature extraction of seal image on receipt is proposed. The method includes binarization,edge detection,morphology processing and extraction seal image from recerpt image with text and figures. The size and width are constant on the same kind of the seal image. Based on this noise interference and human _caused disturtance could be avoided in complex background on receipt image. At last the experimental data show that the extracting seal system in this paper has better robustness and balance between error rate and rapidity.The method can help to digitalize and store the receipt.
领 域: [自动化与计算机技术] [自动化与计算机技术]