模糊几何和纹理特征在指纹细节点后处理中的应用

Fingerprint Minutiae Post-processing Using Fuzzy Geometry and Texture Feature

指纹识别一般基于指纹细节点匹配 ,当指纹图像质量较差时 ,细节点的可靠提取十分困难 ,通常会产生大量的虚假细节点 为提高细节点的精度 ,给出一种在原始灰度指纹图像上进行细节点后处理验证的方法 在每个自动提取出的细节点上取其在原始灰度指纹图像上的局部邻域 ,分析邻域图像的模糊几何特征和纹理特征 ,然后用MLP神经网络对提取出的局部邻域特征进行分类 ,实现细节点类型验证 实验结果表明 :文中方法能有效地去除大量的虚假细节点 。

Fingerprint recognition is usually based on minutiae matching. Reliable minutiae extraction is very difficult for poor fingerprint images and many false minutiae will be extracted. To improve the accuracy of minutiae, a new minutiae post-processing method based on the analysis of initial grayscale fingerprint image is presented. The fuzzy geometry features and texture features were extracted from each minutia's local neighborhoods. The verification of minutia's type was realized by classifying these features with a MLP neural network. Experimental results show that the proposed method can filter out most false minutiae and is more accurate than other methods.