【作者】 石旦旦；

【导师】 王万良； 徐新黎；

【作者基本信息】 浙江工业大学 ， 控制理论与控制工程， 2010， 硕士

【摘要】 随着互联网技术及信息科技的飞速发展,传统的基于信物、口令及密码的身份安全系统已经不能满足人们的需求。生物识别技术具有唯一性且终身不变等特点,满足作为身份验证的要求。指纹识别作为最古老也是相对较成熟的生物识别技术,已经被广泛应用于金融、司法等安全系统。然而,当前指纹识别技术还不能满足日益增强的性能要求,各方面技术还有待改进。本文对指纹识别技术中的几个关键技术进行研究,提出了可行的解决方法,具体归纳如下:1.对指纹图像预处理工作进行研究,着重介绍了方向场提取技术及根据方向场进行指纹增强的技术。并且根据方向场信息进行指纹的中心点提取,与以往的中心点提取方法相比,算法简单,速度快。2.针对传统的基于脊线细化的检测方法的耗时长、伪节点多等缺点,提出一种基于差分码的细节特征点提取方法。首先跟踪脊线边缘得到差分码,再利用差分码对不同的边界特征具有独特的差分码这一特点检测转弯点,这些点就是指纹细节点位置。通过与三种不同的方法进行实验对比,结果表明,本文方法有更好的提取准确性和时间性能。3.传统的采用点模式匹配方法,需要寻找参考细节点,并且还要通过旋转平移匹配细节点对,存在时间耗费大,参数计算繁琐等缺点。针对这一问题,本文以中心点为基准构造螺旋向量,并将方向场和细节点数同时纳入匹配。匹配时只需比对两个螺旋向量即可,实验结果验证了该匹配算法的有效性。最后,对论文的研究工作进行了总结和展望。更多还原

【Abstract】 As the rapid development of internet technology and the information technology, the traditional security systems such as keepsake, user name and password can be no longer meet people’s needs. So the bio-identification technology was born, which is unique and remains a lifetime. These features of biological recognition technology make it an excellent key in identity authentication system. As the oldest and the relatively more mature biometric technology, fingerprint recognition technology has been widely used in the financial, judicial and other security systems. However, the current fingerprint identification system performance can not meet the increasing requirements, many aspects of technology could be improved.In this paper, several key technologies of fingerprint recognition are studied and several viable solutions are put forward. Specifically summarized as follows:1. In this paper, Fingerprint image pre-processing work is on the study, highlighting the orientation field and in accordance with the orientation field extraction technology to enhance fingerprint technology. Using the information of orientation field to detect core. Compare with the traditional way, this method is simpler and faster.2. The traditional way based on ridge thinning has the defect of time-consuming and many spurious minutiae. In view of this limitation, a new algorithm for extracting minutiae based on differential code is proposed. At the beginning, the differential code is got by tracing ridge boundary; subsequently, the turning point is detected through the differential code with the special characteristic for different features, where the minutiae existing. Three different methods are chosen to compare with the proposed approach. The results show that it can detect minutiae with less computational cost and lower false detection.3. The way we generally using to match fingerprint is point mode, which needs to find a reference point and match the details by translating and rotating points. The calculation is complex and it costs a lot of time. To address this issue, this article constructs a spiral vector and makes core as the reference point. We use orientation field and minutiae number both as the matching materials. We should only compare two spiral vectors to get the matching result. Analysis of the experimental data shows the validity of the algorithm.Finally, a summary is made and some future works are presented.更多还原