【作者】 于博；

【导师】 段玉波；

【作者基本信息】 东北石油大学 ， 模式识别与智能系统， 2011， 硕士

【摘要】 随着网络时代的到来,传统的身份识别方法比如交易密码,银行账号等已经逐渐不能满足现今人们对安全的需要。生物特征识别是利用人体自身成长过程中自带的生理特征来进行身份识别,这种方式无需记忆方便易行,且具有长期稳定型,与传统身份识别相比具有无可比拟的优势,是未来身份识别发展的主流方向。指纹识别在人类发展的历史长河中,有着悠久的历史和重要的作用,1960年至今,人们已经为指纹数据建立了电子档案。人工查对指纹的工作单调、枯燥,往往要花很长的时间,并且容易出现错误和判断。因此,人们建立了辅助人工鉴别的计算机指纹分析系统。时至今日,指纹识别已经完全依赖于计算机技术,优秀的算法成为提高鉴别精度和速度的关键。本文设计了基于TMS320VC5501 DSP的嵌入式指纹识别系统,分析了指纹的特征,建立了指纹的数学模型。对指纹图像进行预处理,通过对采集到的指纹图像进行分析,使用模式算法进行指纹分割、均值、取其对象等方法获取需要的真实的指纹信息。采用空间低通滤波法,对指纹图像进行平滑处理。独创性的使用蛇模型方法来实现对指纹图像的特征提取。使用模板匹配的方式对已提取的特征点进行有效的匹配。本文通过CCS3.0软件,成功的把指纹预处理算法以及特征值提取算法移植到TI公司的TMS320VC5501型高速DSP处理器中,实现了指纹识别系统的硬件匹配并达到了预期的效果。更多还原

【Abstract】 With the advent of the Internet age, the traditional identification methods like trading passwords, bank account number has been gradually unable to meet the needs of modern people to safety. Biometrics is the use of the body’s own process of growth that comes with physical characteristics for identification, you don’t have to remember anything in this way and it’s easily to work. What’s more, it has a long-term stable, compared with the traditional identification with the unparalleled advantage, is the future Identifying the direction of the mainstream of development.The development of fingerprint identification in human history, has a long history and important role, since 1960, it has been established for the fingerprint data in electronic files. Artificial fingerprint check work monotonous, boring, like always take a long time, and error-prone judgments. Therefore, people build a secondary analysis of artificial fingerprint identification computer system. Today, fingerprint identification has been completely inseparable from the computer, the best method to improve identification accuracy and speed as the key.This article is designed based on TMS320VC5501 DSP embedded fingerprint identification system that analyzes the characteristics of the fingerprint to establish the fingerprint of the mathematical model. The fingerprint image preprocessing, collected through the analysis of fingerprint images, fingerprint segmentation algorithm using the mode, mean, whichever is the object and other methods of getting to the real fingerprint. Low-pass filter using space on the fingerprint image smoothing. Original use of full snake model approach to achieve the fingerprint image feature extraction. Template matching method using the extracted features have been effectively match point. By CCS3.0 software, the success of the preprocessing algorithm and the fingerprint feature extraction and matching algorithms to TI’s TMS320VC5501 DSP processor speed to achieve the hardware matches the fingerprint identification system and achieve the desired good results.更多还原