连续方向图法在残留指纹分割中的应用

The application of continuously distributed directional image in remaining fingerprint segmentation

针对残留指纹不易被分割的特点,提出一种连续方向图实现残留指纹分割的应用算法。该算法首先计算点方向图,然后利用直方图滤波技术对指纹点方向图进行滤波处理,得到连续分布方向图;其次对前景区域进一步分析,根据残留指纹和正常指纹方向一般不一致的特点,运用残留指纹与指纹区交界处方向信息变化较大的特性标记出残留指纹和指纹区边界点,然后对标记的一些虚假点进行后处理,最后利用这些边界点进行曲线拟合后去除残留指纹。实验结果表明这种连续分布方向图过渡平滑、自然,既具有很好的连续性、渐变性和抗噪性,又有较高的精度,可以有效地分割残留指纹。

For the problem that the remaining fingerprint cannot be segmented easily,the authors propose a segmentation algorithm for fingerprint images with remaining ridges based on continuously distributed directional image. The algorithm firstly calculates the point directional image,and uses the histogram filtering technology to filter directional image,and then the continuously distributed directional image can be obtained. Secondly,according to the direction inconsistency on the location where the remaining ridges border on the fingerprints in the foreground,we mark out the boundary between the remaining fingerprint and normal ridge based on the sharp directional information change. After dealing with false points in the post-processing,these true boundary points are used for curve fitting. Finally we remove the remaining fingerprint. Experiments have shown that such a continuously distributed directional image transits smoothly and naturally. It not only has good continuity,gradual change and noise immunity,but also has very high accuracy. And it can effectively segment the remaining fingerprint from the foreground.