为了弥补Maio等人提出的直接灰度提取细节算法(DGMD)在指纹脊线曲率变化较大时不能正确的提取细节点的缺点,本论文提出了一种基于Hessian矩阵的指纹细节点(端点和分叉点)提取算法。该算法结合Hessian矩阵的特征向量和指纹纹线本身的几何方向的特点,提取跟踪曲线的细节点。本算法在FVC2000(国际指纹竞赛数据库)上作了测试,实验结果表明,指纹脊线曲率变化较大时,算法能正确跟踪指纹脊线方向突变的中心线,从而使细节点提取的总的准确率可达95.5%,具有一定的实用价值。与DGMD算法和单独的Hessian矩阵法算法相比,本方法有很好的鲁棒性和较高的准确性。

An algorithm of fingerprint minutiae extraction using Hessian matrix was proposed in order to overcome the shortcoming of Direct Gray-level Minutiae Detection (DGMD) proposed by Maio,etc.,which had the failure of minutiae extraction as a result of abrupt changes of the fingerprint ridge curvature. Combining the characteristics of Hessian matrix eigenvectors and the geometry direction of fingerprint,the proposed algorithm tracked the ridges of fingerprint and extracted minutiae accurately. The experimental results based on FVC2000 show that the proposed algorithm can track the fingerprint ridge whose curvature changes abruptly. So the algorithm has achieved the total correct minutiae extraction rate of 95.5%. Moreover,the algorithm has better robustness and accuracy than DGMD algorithm and Hessian matrix algorithm.