一个新的基于细节特征的指纹匹配方法

自动指纹识别系统(automaticfingerprintidentificationsystems,AFIS)的精度和效率主要依赖于指纹的匹配算法.指纹匹配涉及的两个关键问题是指纹的对齐和匹配方式.根据同一个指纹的不同采样,其脊线形状保持高度的相似性的特点,利用两条脊线对应点的距离构造了一个判据,用来评价两条脊线形状的相似性,以实现指纹的最优对齐;针对传统指纹匹配算法中伪细节点的混入和真实细节点的遗漏影响指纹匹配精度的问题,提出了一种基于编辑距离原理的指纹细节特征匹配方法,对指纹库Fingdb和FingerDUT进行了测试,等错误率分别为0.62%和2.75%,证明该方法具有较高的可靠性和有效性.

A new fingerprint matching method based on minutiae of fingerprints

The accuracy and efficiency of automatic fingerprint identification systems (AFIS) mainly depend on the fingerprint matching processing. The alignment of fingerprints and the matching method are the two key problems to matching working performance. Because the shapes of ridges from the same fingerprint keep similarity in different acquisition images, a novel criterion for evaluating the similarity of two ridge shapes is proposed, which is constructed based on the distances of corresponding point pairs on ridges. An optimal alignment of two fingerprints can be implemented by aligning the most similar ridges on two fingerprints. In the conventional fingerprint matching approach, the detailed position change because of ridge distorting and the failure matching because of the extra pseudo minutiae interfusing and/or the real minutiae missing, can affect the precision of matching. So a new method based on editing distance for calculating the similarity score between two fingerprints is presented. Experiments have been done on two fingerprint databases, Fingdb and FingDUT, and the equal error rates (EER) are 0.62% and 2.75% respectively. The results show that the new approach is robust and effective.