摘要： 提出了一种改进的奇异点的提取方法：先通过改进的Poincaré指数计算，挑选出符合条件的候选点，然后在此基础上，利用径向方向会聚性，前景、背景分割和概率分布来筛选候选奇异点。用该方法对不同质量的指纹图像进行实验，并与经典方法进行比较，结果表明该方法更加有效、可靠，也具有更好的鲁棒性。

关键词: 指纹图像识别, 奇异点, 方向图, Poincaré指数

Abstract: In this paper, a robust method for the detection of fingerprint sitgular points is presented. An improved Poincaréindex, the orientation variation measure and the use of gray value variance, and combination of them are proposed for fingerprint singular point detection and positioning. Experimental results on fingerprint images and comparison with traditional methods show less computational complexity and more robustness of the method are proposed.

中图分类号: 

• TP391.4