【作者】 成春桃；

【导师】 刘正光；

【作者基本信息】 天津大学 ， 模式识别与智能系统， 2009， 硕士

【摘要】 近年来,随着全球经济和信息技术的不断发展,安全问题日益突出,越来越多的领域需要可靠的身份识别。生物识别技术为此提供了一种解决方案,而自动指纹识别技术凭借其唯一、稳定、安全等特点成为其中的最佳选择之一。针对识别速度慢、识别率低等问题,本文对指纹识别算法进行了重点研究并做了相关改进,完成了一套切实可行的算法,同时,将其应用于基于EFI的计算机开机身份验证。算法方面的改进具体体现在以下几方面:(1)为了避免质量过差的指纹图像的输入,本文引进指纹质量评估体系,这样有效提高了系统的识别效能。(2)奇异点在指纹匹配中有重要作用,本文引入了奇异点的检测,设计了基于奇异点的粗匹配,提高了系统的识别率和平均识别速度。(3)在预处理时,对基于Gabor滤波器的增强算法加以改进,在保证增强效果的前提下减少了该步骤的时间消耗;细化方面将快速细化算法和改进的OPTA细化算法有机结合,采用新模板,使得指纹细化更彻底;去除指纹伪特征时进行自动阈值的选取,提高了算法的通用性。(4)在匹配过程中,使用粗匹配、方向匹配和细节匹配相结合的多级匹配算法,对不同的指纹采用不同的匹配方法;在确定两幅指纹图像是否匹配时采用复合判定法。实验结果表明,该算法在识别率和识别速度方面都取得了相对令人满意的效果,但该算法对指纹位置有一定要求。本文的项目背景是设计基于EFI的指纹识别应用系统。本文对EFI驱动模型进行了重点研究,实现了采集指纹和指纹识别的驱动程序开发,成功地应用了该项新技术,具有当今基于EFI层面产品开发的前沿技术水平。更多还原

【Abstract】 Recently, with the development of continuous development of the globaleconomy and information technology, security precautions have also becomeincreasingly prominent in recent years. Operations of more and more fields depend onidentification technology. Biometric technology provides significant solutions, andautomated fingerprint identification is one of the best options because of itsuniqueness, permanence and safety.In order to solve problems of slow speed and low rate of recognition, this paperfocus on the research on fingerprint recognition algorithms and proposes correlativeimprovements, finally finishes a whole set of practical algorithm. At the same time,this algorithm is applied to the computer boot authentication based on EFI.Improvements of the algorithm consist of the following main contents:(1) In order to avoid inputting those images with too low quality, this paperintroduces fingerprint quality evaluation system, as a result the performance of thesystem is improved.(2) Singular point is very important for fingerprint matching. This paperintroduces singular point detection and designs rough matching based on singularpoint, so as to improve rate and average speed of recognition.(3) In preprocessing part, fingerprint enhancement algorithm based on Gaborfilter is improved to reduce its consuming time under the premiss of ensuringenhancement effect. In thinning aspect, this paper makes the result closer to thefingerprint grain line center by integrating the rapid thinning algorithm and theimproved OPTA thinning algorithm. Automatic threshold is used while eliminatingfalse features for fingerprint, so as to make the algorithm more universal.(4) In the process of matching, this paper uses the multi-hierarchy matchingmatching of the integration of the rough matching,the direction matching and thedetail matching; in order to determine whether two fingerprint images are matching,the multiple determination method is used. The experimental results show that thealgorithm acquires a relatively satisfactory result in aspects of recognition rate andrecognition speed.The fingerprint recognition system based on EFI is designed and applied in this paper. This paper focus on studying EFI driver model, implements drivers offingerprint capture and fingerprint identification, and applies the new technology onthis project successfully. The project presents the high level in the area of productsbased on EFI.更多还原