一种基于点模式的指纹分阶段匹配算法的研究

Study of Fingerprint Grading Matching Algorithm Based on Point Pattern

介绍了一个指纹中心点定位(Core)以及用中心点为中心构造特征向量进行初匹配,并且以此作为最佳匹配参考点来进行二次匹配的算法。本算法的特点:1.介绍指纹中心点的准确定位。2.以中心点作为最佳匹配参考点将匹配分为两步进行:初匹配利用了细节点间的结构关系,克服了图像的平移和旋转的影响;二次匹配引用了界限盒思想,增强指纹匹配算法对形变的适应能力。本算法把点模式的优点和基于结构的特征点之间的相对距离不变性、所跨越纹线数目的不变性、特征点类型的不变性很好结合起来。实验结果显示本算法具有较强的适应性和较高的拒识率。

Introduce a core location and as a core to construct the feature vector first matching, and as a best matching point to make second matching. This algorithm＇s trait is： 1. Correctly location core;2. Using a core as a best matching point which divide grading matching into two steps： first matching make use of minutia structure relation which overcome influence on revolving and motion; second matching excerpt bounding box idea, which enhance fingerprint matching algorithm to adapt transfiguration, This algorithm combines merit of point pattern with structure merit（ feature point relative distance fixedness, spanning ridge amount fixedness, feature point type fixedness）, The experiment result shows that algorithm adapts well to FAR.