Abstract: An effective new algorithm is proposed for improving the stability and reliability of Singular Points（SPs） location. It preprocesses the gray level fingerprint images, calculates the orientation field, and divides the orientation field into several homogeneous zones and selects candidate small areas. Candidate singular points are chosen out by Poincare Index Algorithm（PIA） and spurious singular points are further removed. The real singular points are precisely located. FVC2004 is used for algorithm testing and the results show that singular points location of the method is more precise, and the False Detecting Ratio（FDR）and the Missed Detecting Ratio（MDR） decrease 5.86% and 6.8% respectively, at the same time, the speed of the algorithm is 3.71~9.38 times faster than PIA, which basicallies satisfies the requirement of SPs accurate detection and location in fingerprint images.

Key words: fingerprint image, orientation field, singular points, homogeneous-zones-divide

摘要： 针对指纹图像奇异点快速精确定位的难题，提出一种简单实用算法。对指纹图像预处理，计算方向场并归域化，接着选出奇异点候选区，并以Poincare Index（PI）算法从中提取奇异点候选点集。对候选奇异点集去伪并精确定位。采用FVC2004指纹库进行实验验证，结果与PI算法对比，该算法鲁棒性更好，定位更精确，漏检率和误检率分别降低5.86%、6.8%，平均速度提高了3.71~9.38倍，基本满足高精度高速度的指纹奇异点定位要求。

关键词: 指纹图像, 方向场, 奇异点, 归域化