提出了一种新的基于方向场离散傅里叶变换 (Discrete Fourier transform, DFT) 和角点检测的指纹奇异 点定位算法。对方向场做 DFT 变换到频域处理能更好地压制噪声。通过角点提取、基于 目标标注的聚类方法来定位指纹图像的奇异点。从方向场的极/零点模型出发,导出了若干新颖的性质,说明了奇异点和 角点之间的关系。实验结果表明,奇异点定位精度有所提高,处理时间更短。

A new method for locating fingerprint singular points based on discrete Fourier transform (DFT) of orientation field and corner detector is proposed. Orientation field is processed in frequency domain via DFT so that noise is better suppressed. Fingerprint singular points are located by corner detection and clustering based on object labeling. Novel properties, which reveals the relationship between singular points and corners, are derived based on pole/zero model of orientation field. Experimental results show improved accuracy of singular points location and a reduced processing time.