一种基于独立分类特征的指纹多级分类算法

A MULTI-LEVEL FINGERPRINT CLASSIFICATION ALGORITHM BASED ON INDEPENDENT CLASSIFICATION FEATURES

为了提高基于大容量指纹库的自动指纹识别系统的检索效率,提出一种基于独立分类特征的指纹多级分类算法。依据评测指标对输入指纹图像进行质量评估,若指纹质量不合格,则提醒用户重新输入;若指纹质量合格,则分别利用指纹图像的纹型类别、奇异点间脊线数、中心区域脊线平均频率3个相互独立的分类特征实现多级分类,从而逐级减小检索空间。实验结果表明,该分类算法检索效率高、鲁棒性强,为大容量指纹库提供了一种快速有效的索引机制,具有很强的实用性。

In order to improve retrieval efficiency of the automated fingerprint identification system for large-scale fingerprint database,a multi-level fingerprint classification algorithm based on independent classification features is proposed.Quality evaluation indexes could be applied to evaluate the quality of input fingerprints,remind the users to re-enter fingerprints if the quality is poor.For the good quality fingerprints,three independent features including grain type,number of ridges between singular points and average frequency of ridge line in central region are separately used to realise multi-level classification,which could decrease the retrieval space gradually.Experimental results show,the proposed classification algorithm has high retrieval efficiency and strong robustness,which provides a rapid and effective index mechanism for large-scale fingerprint database.It has good practicality.