基于度序列的身份证号码识别研究

发布时间：2018-01-08 15:20

本文关键词：基于度序列的身份证号码识别研究　出处：《北京交通大学》2017年硕士论文　论文类型：学位论文

【摘要】：身份证号码识别技术蕴藏着巨大的经济价值,它在政府办事部门、酒店入住登记等方面具有非常强的应用背景。虽然每个人的身份证号码由18个数字组成,但是其本质上最多出现10个不同的数字。因为待识别图像清晰度与拍摄环境、设备等因素相关,所以身份证号码识别系统应具有鲁棒性。我们希望识别率达到100%,同时对于使用身份证号码识别系统的用户来说,识别系统也应具有界面友好、操作简单等优点。本文首先介绍了身份证号码识别技术的背景和研究意义;其次讨论了身份证号码识别系统的预处理,包括图像二值化、去噪、定位、分割等。对如何选择合适的二值化阈值以及如何对图像去噪进行了论述;并在数学形态学的基础上,讨论了图像细化、分割算法。最后阐述了图像识别的两种方法:结构方法和统计方法。根据身份证号码的数字特征,在统计方法的基础上,本文提出了一种新的改进方法:基于度序列的身份证号码识别。第一,根据度序列的概念,单个数字可以看成是由简单图构成的;第二,设置一个元素均为1的3×3模板,使模板中心点与待识别图像像素为1的点重合,用给定模板与该像素点的8-邻域所有点进行逻辑“与”运算,然后对模板的结果进行计算,即可得到该点的度数;第三,利用Microsoft Visual Studio 2010版本中MFC平台开发功能和C/C++高级语言程序进行界面设计、算法实现,最终在电脑屏幕上显示身份证号码识别结果。经实验结果表明,该方法的身份证号码识别率在88%以上,具有一定的应用性。新的改进方法的优点是把原始图像数据通过计算机软件产生一组不变量——度序列。由于它不需要存储整个身份证图像,因此在很大程度上降低了计算机的存储空间,同时也保护了个人的隐私。
[Abstract]:ID number recognition technology has great economic value, it has a strong application background in government departments, hotel check-in registration and so on, although each person's ID number is composed of 18 digits. But in essence there are up to 10 different numbers, because the clarity of the image to be identified is related to the shooting environment, equipment and other factors. Therefore, the identification system should be robust. We want to achieve a recognition rate of 100, and for users using the identification system, the identification system should also have a friendly interface. Firstly, this paper introduces the background and significance of ID number recognition technology. Secondly, the preprocessing of ID number recognition system is discussed, including image binarization, denoising, location, segmentation and so on. On the basis of mathematical morphology, the algorithm of image thinning and segmentation is discussed. At last, two methods of image recognition: structure method and statistical method, according to the digital feature of ID number, are discussed. On the basis of statistical method, this paper proposes a new improved method: identification of ID number based on degree sequence. Firstly, according to the concept of degree sequence, a single number can be regarded as a simple graph. Secondly, a 3 脳 3 template with an element of 1 is set so that the center point of the template coincides with the point of pixel 1 of the image to be identified, and the logical "and" operation is performed with the given template and all the points in the 8-neighborhood of the pixel point. Then the results of the template are calculated and the degree of the point can be obtained. Thirdly, using the MFC platform development function of Microsoft Visual Studio 2010 version and C / C high-level language program to carry on the interface design, the algorithm is realized. Finally, the identification results are displayed on the computer screen. The experimental results show that the identification rate of the method is more than 88%. The advantage of the new improved method is that the original image data is generated by computer software to produce a set of invariant-degree sequences because it does not need to store the entire ID image. As a result, the storage space of the computer is greatly reduced, and the privacy of the individual is also protected.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】