指纹图像方向提取方法的研究

指纹自动识别是图像处理技术、模式识别技术与计算机数据库技术的综合应用。指纹图像的方向信息对指纹的预处理、增强、分类及匹配有着重要的作用。因此，在指纹自动识别系统中具有重要的研究价值。文中在金字塔表示法的基础上提出了一种新的指纹图像方向提取方法。该方法通过改进局部方向的估计方法、确定传递权值等方式，提高指纹图像提取方向信息的准确性。实验表明，这种方法对指纹图像噪声有很好的健壮性，且在质量较差的指纹上提取方向信息的准确性方面相对于常用的局部梯度算法来说有明显提高。

Research on Extraction Method of Fingerprint Orientation Map

Automatic fingerprint identification system has integrated image processing, pattern recognition and database technique. The fingerprint orientation map information is very useful for fingerprint preprocessing, enhancement, classification and matching, so it is very important in automatic fingerprint identification system. This paper presents a fingerprint image local orientation estimation method, which is based on image pyramid. The proposed method can improve the accuracy of extracting directional information from fingerprint by elevating the estimation of local orientation and the value of propagation weight. The experimental results show that proposed technique is more robust against image noise and has better effect than most local gradient-based algorithms when elevating the accuracy of extracting directional information from poor fingerprint.