【作者】 王双；

【导师】 刘铁根；

【作者基本信息】 天津大学 ， 光学工程， 2007， 硕士

【摘要】 印鉴在中国、日本、韩国等东方国家有着广泛的应用。在我国,印鉴的作用尤其重要。但随着社会经济的发展,银行业务量日渐增大,传统的人工鉴别支票印鉴真伪的方法,已不适应银行业务的需要。而市场上现有的银行印鉴自动识别系统又存在判据单一、识别错误率高等缺陷。因此,对支票印鉴图像的提取和配准方法进行进一步的研究,具有较高的现实意义和理论意义。本文在搜集和分析国内外相关研究报告及学术论文的基础上,结合图像处理和模式识别的知识,专注于支票印鉴图像的提取和配准方法研究。论文中首先对支票印鉴图像的复杂特性进行了分析。为了消除盖印条件的影响以及支票本身背景暗花、手写或印刷字体等形成的噪声干扰,本文采用色彩分量变换法、灰度线性化、中值滤波、最大类间方差法、形态学闭运算及贴标签法等系列方法流程,对支票印鉴图像进行预处理,得到适宜识别的二值低噪声图像。在获得二值印鉴图像的基础上,本文提出了种子扩散孔洞填充算法和改进的多级Hough变换检测圆的方法,并运用其提取出预留印鉴的特征信息作为先验知识,将支票中的圆形印鉴和方形印鉴依次提取出来。这一印鉴图像的提取过程,改正了现有方法多是针对单个印鉴进行处理的缺点。针对提取出的圆形印鉴印文缺损,本文建立了三个基准图像,运用形态学的细化算法及膨胀运算对其进行处理,找到缺损位置并进行补偿。在分析了传统的印鉴配准方法的基础上,本文提出了一种先粗略配准再精确配准的双层印鉴图像配准算法,其旋转角度可精确到0.1o,从而实现了印鉴图像的精确配准。为了对配准效果有一个量化的评价,本文引入了判决中常用的内差和外差概念,通过对内差、外差百分比设定阈值的方法进行印鉴真伪判决。对大量支票印鉴样本经过上述预处理、提取及配准方法处理后,取得了比较满意的判决结果,可以证明该方法得到的配准图像效果良好。更多还原

【Abstract】 In the oriental countries, such as China, Japan and Korea, seals are extensively used for the purpose of entity identification. In China, seals play a more important role. With the development of social economy, the portfolio of the bank increases day by day. So the traditional method of manual check seal identification cannot readily meet the needs of banking transactions. While there is only a single criterion in the existing bank seal automatic recognition system of market with the deficiencies of low precision. Thus, carrying on further research of extraction and registration algorithm for check seal identification has higher practical meaning and theoretical meaning.On the foundations of the research reports and scientific papers of domestic and international, we focus on extraction and registration algorithm for Check Seal verification using theory of image processing and pattern recognition. In this paper, we analyze the complex characteristics of check seals firstly. To eliminate the difference of producing conditions and the disturbance caused by background and writing in check image, many methods are used in the pre-processing of check seal verification, such as color components transformation, linearity transform to gray image, medium value filter, Otus, close calculations and labeling algorithm of mathematical morphology. After the processes above, the good binary seal image can be obtained.We bring forward the algorithm of the Seed Spread Filling Holes and Multi-Hough Circle Detection. Using these algorithms, we calculate the characteristic of seal reserved and get the rotundity and squareness seal from check image. The seal extraction processing corrects the existing method’s shortcomings of treatment seal on individual. In order to repair the seal character defect, we establish three benchmarks image to find the defect’s location and repair it using thinning and expansion algorithm.On the basis of traditional registration algorithm, we bring forward a double-level registration method including rough and precise registration method. The rotate angle of precise registration method can be précised to 0.1o.We introduce the concepts of difference inside and difference outside and use the percent of difference inside and difference outside to judge whether the seal is real or fake. The result of a mass of check seals is satisfied. Therefore, the above methods and algorithmic have got quite good registration effect.更多还原