提出一种两阶段的奇异点检测方法。将指纹图像分块，求出各块的方向构成块方向图，并在块方向图的基础上利用邻域方向的分

布分析结合改进的Poincare Index方法来确定奇异点所在的候选区域，对候选区域中的像素再通过计算局部方向变化率来确定奇异点的精

确位置。将此方法用于对FVC2004 DB1_A指纹数据库的图像，实验结果表明这种方法对指纹图像中的噪声有很好的鲁棒性，并且计算简

单快速，易于实现。

-A two-phase singularity detection. Fingerprint image block, calculate the direction of each block constitute a block pattern, and pattern in the block based on the use of the distribution of the direction of the neighborhood together with the improved Poincare Index method to determine the singular point where the candidate regions of the candidate region of pixels re-calculated to determine the rate of change of local direction of the precise location of singular points. This method is used to FVC2004 DB1_A fingerprint database of images, experiments show that this method of noise on the fingerprint image has good robustness and computing simple and fast, easy to implement.