基于方向图的指纹纹线检出算法研究

指纹纹线检出大多是先按指纹方向图增强指纹图象，然后进行图象分割。但因指纹脊线内和谷线内的图象增强方向可能不一致，使得图象分割的效果不好。作者提出一种直接按指纹方向图进行指纹纹线检出的算法，它先采用拓宽的Prewitt算子计算指纹图象的梯度矢量图，然后用局部最大梯度法计算指纹方向图，最后用阶跃函数模拟从原始指纹图象和指纹方向图中进行指纹纹线检出。

A Study of Orientation Based on Fingerprint Detection

One of the key techniques of automatic fingerprint identification is fingerprint detection. Most of the fingerprint detection is first to enhance a fingerprint image referring to the fingerprint orientation, then to segment the fingerprint image. A problem of the methods is that the directions of image enhancement within a ridge and within a valley do not conform. For this reason, the authors propose a method of fingerprint detection referring to fingerprint orientation. In this method, first, a widened Prewitt operator is applied to calculate the gradient vectors of a fingerprint image. Then, a local maximal gradient vector approach is applied to calculate the fingerprint orientations. Finally, a step function fitting method is used to detect the fingerprints referring to the fingerprint orientations.