一种基于遗传算法的指纹比对算法  

Fingerprint matching based on the genetic algorithm

作　　者：洪波[1] 荣钢[1] 黄韬[1] 

机构地区：[1]清华大学自动化系

出　　处：《清华大学学报（自然科学版）》2001年第3期93-96,共4页Journal of Tsinghua University(Science and Technology)

摘　　要：指纹比对在身份验证与身份识别中具有非常重要的应用。该文提出了一种基于遗传算法的指纹比对算法。该算法针对指纹比对的特点 ,设计了相关的目标函数与遗传算子 ,搜索了两幅指纹间可能存在的平移、旋转等几何参数变化关系 ,并在此基础上确定了二者之间的特征点匹配关系 ,从而可判断两幅指纹匹配的程度。实验结果表明 ,该方法可以在变换与局部变形存在的情况下 ,正确判定两幅指纹之间对应的特征点。与直接利用遗传算法搜索特征点对应关系的方法相比 ,新搜索方法收敛所需的迭代次数仅为遗传算法的1/ 5 。Fingerprint plays an important role in identity verification. A matching method based on genetic algorithm is proposed. Taking consideration of the characteristics of fingerprint matching, specific objective function and genetic operators are introduced to search the possible geometric relation between two fingerprints. Point pair relations are then determined according to these parameters. Thus similarity between the two fingerprints can be decided. Experiment results show that this method can find the correct feature point pairs under transformation and local distortion. Compared with the genetic algorithm based method that search point pair relation directly, this parameter based method converges at a speed 5 times faster.

关 键 词：指纹比对算法 遗传算法 特征点匹配 适度函数 身份识别 指纹匹配 匹配度 适度函数 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]