【作者】 王安东；

【导师】 高隽；

【作者基本信息】 合肥工业大学 ， 信号与信息处理， 2003， 硕士

【摘要】 随着计算机与信息技术的发展，生物特征识别技术的研究受到了广泛的关注。其中，指纹识别是生物特征识别中的一项重要内容，一直以来是国内外的研究热点。 传统的自动指纹比对系统是基于细节匹配的技术，需要进行复杂的图像预处理，算法复杂度高，耗时较长，不易硬件集成，因而有一定的实用局限性。协同识别技术是一门新兴学科，具有快速而高鲁棒性的识别能力。本文尝试利用指纹图的全局特征—指纹方向场信息，结合协同识别技术，进行指纹协同识别方法的研究。 论文的主要工作如下： ◆ 阐述了生物特征识别的发展历史、应用背景，并着重介绍指纹识别技术的发展、应用及研究现状。在收集整理指纹识别技术的文献资料基础上，重点研究了传统指纹识别方法中的关键步骤——图像预处理，验证了传统算法的性能，并对其优缺点进行了总结； ◆ 综述了协同模式识别基本理论。结合数字图像处理技术和协同识别技术，以指纹图像方向场作为指纹的特征信息，对“验证”和“识别”这两种指纹识别模式分别进行研究，并以VC++6.0为平台，编程实现了算法。实验中，采集真人的指纹图像作为样本，从算法的实用性和实时性两方面进行了探讨； ◆ 描述了算法在CCS集成开发环境中的软件仿真和在DSP—TMS320C54x系列器件上模拟运行的过程。实验结果显示算法运行时间短，复杂度小，硬件要求低，满足实用化的要求。更多还原

【Abstract】 With the development of electronic computer and information technology, the Identification technology using biometric have drawn more attention. Fingerprint-based identification technology as a very important method used biometric features, is always the hotspot.Classic automatic systems for fingerprint comparison are based on minutiae matching. They almost involve time-consuming fingerprint image’s pretreatment, which is very complicated algorithm implementations and need advanced hardware, so it is difficult to popularize. Synergetics identification method, as a rising subject, has rapid and robust capability on image identification. In this paper, fingerprint orientation field, which is a globe feature of fingerprint, has been adopted to identify fingerprint combining Synergetics identification method.The main work in this paper as follows: The evolutionary development and application background of Identification technology using biometric feature have been described, especially of Fingerprint-based identification. The information about fingerprint-based identification has been collected. Research of fingerprint image’s pretreatment that is key part of classic fingerprint-based identification has been done, and the performance of some algorithm has been compared; The approaches of pattern recognition using Synergetics have been summarized. Making use of image processing and synergetic identifying technique, I have worked on two different fingerprint-based identification modes, namely verification" and "identification". The approach of two modes has been proposed and implemented with C++ program language in this paper. For emulating the actual condition, the samples of different individual’s fingerprintshave been taken to make experiment with the approach in practicability and in good time subsequently. In addition, the algorithm was simulated on Code Composer Studio IDE, which provides a development platform that enables Tl customers to evaluate and develop applications for the DSP. The result of the simulation showed that the algorithms meet with the requirement of TMS320C54x, and it was possible to integrated into DSP practicably for popularizing.更多还原