Abstract: Proposed a method to detect the core point and its orientation from fingerprint image.Through using hierarchical analysis of the orientation coherence，picked out the points of low orientation coherence，then utilized poincare index to distinguish the style of the low orientation coherence points.After the style have been distinguished，developed a new algorithm to confirm the orientation of the core point，through using the orientation information of a cirque，the centre of a cirque is the core point.Experimented in FVC2002 fingerprint database，result show that this method detect core point correctly and quickly，orientation of core point is precision and fastness，have the ability to defeat the noise，it can badly conform the request of fingerprint alignment.

Key words: fingerprint recognition, orientation filed, orientation coherence, core point

摘要： 提出了一种指纹中心点检测以及确定其方向的算法。该算法通过分层分析指纹方向场的一致性，检测出方向场中方向一致性较低的点，然后利用poincare索引辨别检测出方向一致性低的点的类型。类型确定后，提出了一种使用以指纹中心点为圆心的环形域中的方向信息，确定指纹中心点方向的新算法。在FVC2002指纹图像库中进行实验，结果表明，该算法检测指纹中心点准确、快速，所确定的中心点方向精确、稳定，对噪声具有较高的鲁棒性，能够很好地满足指纹对齐的要求。

关键词: 指纹识别, 方向场, 方向一致性, 中心点