基于同一性检测的指纹细节特征提取方法

EXTRACTING FINGERPRINT MINUTIAE BASED ON IDENTITY DETECTION

针对常规的指纹细节特征提取方法难以消除指纹图像预处理过程中产生的伪细节点问题,提出一种基于同一性检测的提取方法。首先,分别利用频域滤波器和小波域Gabor滤波器增强处理同一幅指纹图像;然后从得到的两幅图像中分别提取指纹各细节特征的位置和类型,取这两幅不同增强图像的特征点的交集以消除预处理环节产生的伪细节点;最后剔除指纹图像本身存在的伪特征。实验结果表明,该方法鲁棒性强,能显著提高所提取指纹细节特征的真实性和准确性。

As the false minutiae arising from fingerprint image pre-processing are difficult to be deleted by conventional fingerprint minutiae extraction methods, a new approach for minutiae extraction is put forward, which is on the basis of identity detection. Firstly, the proposed method uses frequency domain filtering and wavelet-based Gabor filtering to enhance .the same original fingerprint image respectively; Secondly, it extracts the minutiae of two different pre-processed fingerprint skeleton and detects the positions and types of these minutiae, then, it contrasts the characteristics of two sets of minutiae and takes their union to eliminate the false minutiae; finally it post-processes the union to remove the false minutiae from fingerprint image itself. Experimental results show that this method is of strong robustness and can improve the authenticity and accuracy of fingerprint minutiae extraction greatly.