季婧婧+娄震
摘 要: 由于彩色印章的存在,传统的印章滤除算法常常会弱化票据中印章的遮挡字符,降低票据识别正确率,这很难满足实际应用的需要。针对这种情况,提出一种基于彩色图像二次分割的算法,该算法能够实现印章的有效滤除。算法首先对票据背景和印章部分进行滤波。然后,基于印章颜色分析进行图像的第二次分割,判断并提取出印章遮挡的字符。根据票据背景平均灰度值自适应的增强遮挡字符,并滤除干扰印章部分,实现遮挡字符细节得以保留,从而提高了票据识别的正确率。实验结果表明,相比于现有的方法,该文采用的方法将识别准确率提高了约10%,实验证明了该方法的有效性,并具有较好的应用前景和商业价值。
关键词: 银行票据; 多色印章; 颜色判定; 遮挡字符
中图分类号: TN919?34; TP391.41 文献标识码: A 文章编号: 1004?373X(2014)22?0005?05
Filtering of color seal on bank notes based on re?segmentation
JI Jing?jing, LOU Zhen
(Department of Computer Science, Nanjing University of Science & Technology, Nanjing 210000, China)
Abstract: As a result of the existence of color seal, traditional seal filtering algorithms usually weak the characters covered in the Bank Notes, and reduce the recognition accuracy of the notes. Previous methods are hard to meet practical needs. According to this, an algorithm based on color image second?segmentation algorithm is presented. The algorithm can realize efficient filtering of the color seal. It separates the background of notes and color seal first, and then conducts image second?segmentation based on the color analysis of the seal to judge the result and extract the covered characters. According to the mean gray value of bank note background, the covered character is enhanced adaptively and the area disturbed by the seal is filtered, so the recognition rate of the bank notes is improved. The experimental results show that the recognition accuracy is increased by about 10% by the method in comparison with the existing methods.
Keywords: bank note; color seal; color judgement; covered character
0 引 言
银行票据处理是金融商业中非常重要的一项工作,其过程需要充分体现票据处理的效率性、严密性以及安全性。长期以来,金融部门的票据处理过程效率低下,直接影响了国内银行的整体竞争力。票据的自动分割和票据信息提取能够极大地提高批量票据处理的速度及工作效率,减少人为干预。因此,寻求一种有效而实用的票据处理方法变得越来越迫切。银行票据通常都是通过印章来确认票据的有效性,而印章的遮挡是导致票据信息提取质量严重下降的主要原因之一。因此,对票据印章滤除的研究具有很大的实际意义。……