指纹的匹配方法研究

【作者】 杨敏；

【导师】 郭成安；

【作者基本信息】 大连理工大学 ， 信号与信息处理， 2004， 硕士

【摘要】 指纹识别技术作为生物识别技术的一种，以其特有的唯一性和不变性，成为当前最可靠的个人身份识别技术。指纹匹配作为指纹识别的核心技术之一，是自动指纹识别系统设计中的热点和难点。 本文在前人工作的基础上，完成了如下工作： (1)在指纹匹配的前期处理部分，选择并实现了一套快速、高效的指纹图像预处理、特征提取和后处理算法，为后续的指纹匹配处理奠定了基础。 (2)指纹匹配是本文的重点，在该部分对原有算法进行改进，提出了一种基于指纹全局星型结构特征的分步匹配算法。针对局部特征向量构造简单、易受局部形变干扰的问题，在第一步匹配中采用全局向量的同时，在此向量中增加了邻域特征点与中心特征点所在纹线方向夹角等反映纹线走向等参量，提高了匹配的准确性；针对固定限界盒门限造成误匹配率较高的问题，在第二步匹配中运用自适应门限，进一步提高了匹配的正确率。 (3)在指纹检索部分，提出了一种新的基于星型结构的指纹检索方法，一方面可以有效地进行指纹检索，排除不合格的指纹，使匹配速度得到显著提高；另一方面可以实现多维结构检索，满足不同系统的要求。 本文在实际微机指纹识别系统上实现了上述全套算法。实验结果表明，该套算法快速、有效，结果令人满意。更多还原

【Abstract】 Fingerprint identification as a biometric is becoming one of the most reliable personal authentication technologies due to the uniqueness and immutability of fingerprints. As one of the key techniques, fingerprint matching is still the focus in the design of automatic fingerprint identification systems.The main works given in the paper are as follows:(1) Select and implement a set of fingerprint pre-processing, feature extraction and post-processing algorithms, which is proved fast and effective, and is the basis of the fingerprint matching.(2) Propose a two-step global star-structure-based matching algorithm for fingerprint identification based on existing methods. In the first step of the algorithm, the global star structure feature vector is constructed. The ridge direction and the structural relationship parameters are added into the feature vector. And adaptive thresholds are used in the second step, which improves the robustness of the matching and makes the result more accurate.(3) Propose a novel fingerprint indexing algorithm based on star-structure. This algorithm can accelerate the process of matching greatly by rejecting the mismatched fingerprints efficiently. It can also realize multidimensional indexing and meet the requirements of different systems.All the proposed algorithms have been implemented on a fingerprint image processing system. The validity of the algorithms is confirmed by the experiment results given in the paper.更多还原