Abstract: After investigation existing approaches to fingerprint classification，this paper introduces an improved structural fingerprint classification method.Using the directional field information，Poincare index can detect the discontinuous point in the directional field——singularities.According to the number of singularities，their position and structural character of fingerprint’s ridge lines，the basic category of fingerprint is classified.It successfully provides the information of fingerprint category for the matching method，thus improves the matching speed and precision.Experiment results illustrate that the method can perform better.

Key words: fingerprint image, fingerprint classification, directional image, singularity

摘要： 在深入研究现有指纹分类算法的基础上，提出了一种基于结构的指纹分类算法。利用指纹图像的方向场信息，用Poincare索引值检测出方向场中不连续的点——奇异点，根据奇异点的数目和相互位置关系以及指纹脊线的结构特征确定指纹的基本类型。该算法为匹配算法提供指纹类型信息，可极大地提高算法的匹配速度和匹配精度，具有较好的使用意义。

关键词: 指纹图像, 指纹分类, 方向图, 奇异点