摘要： 提出一种基于指纹统计特性的识别算法。该算法根据奇异点的位置和方向，提取指纹图像的感兴趣区域(ROI)，并使用二维主成分分析(2DPCA)的方法进行统计特征的提取和识别。在FVC2002指纹数据库上进行实验，结果表明：相对于PCA，该方法的计算速度更快。相对于传统的基于特征点的方法，该方法的实现更为简便。

关键词: 指纹识别, 二维主成分分析, 奇异点

Abstract: This paper proposes a fingerprint recognition algorithm based on statistics. Region of interest of a fingerprint image is extracted according to the position and direction of the singular points. Then the two-dimensional principal component analysis approach is applied to the training images represented by ROIs to get the statistical feature space. Fingerprint recognition is performed in the feature space using Euclidean distance classifier. Experimental results obtained on the fingerprint database of FVC2002 prove that the method has the advantages of fast computation speed compared with PCA, and simple implementation compared with traditional minutiae-based approach.

Key words: fingerprint recognition, two-dimensional principal component analysis, singular points

中图分类号: 

• TP18