摘要： 为提高大规模网络化指纹识别系统的检索速度和准确率，提出一种基于多级索引的指纹分类算法，将质量特征分为质量高和质量差2类，对于质量高的指纹利用指纹图像的3个特征——指纹类别、宏观曲率和平均周期建立三级索引，以逐级缩小检索空间。实验结果表明，该算法准确率高、检索速度快，具有良好的实时性。

关键词: 指纹检索, 质量特征, 宏观曲率, 平均周期

Abstract: In order to improve the index speed and the accuracy of the network-based fingerprint recognition system for the large population, a fingerprint classification algorithm based on multi-level index is proposed. Quality feature is used to classify fingerprints into two types: low quality and high quality. For the low quality fingerprints, three features, including ridge shape, macroscopic curve and average period are used to establish three-step index, which decreases the index space gradually. Experimental results show this algorithm has more accuracy, and improves the index speed effectively. It has better real-time performance.

Key words: fingerprint index, quality feature, macroscopic curvature, average period

中图分类号: 

• TP391