一种基于八邻域链码的掌纹图像定位方法(英文)

A Location Method Based on Eight Neighborhood Chain Code for Palmprint Image

准确地检测出手掌轮廓线的角点是对在线掌纹图像进行精确定位的关键.目前已有的角点检测算法不能自动地提取出定位有用的角点.通过分析手掌轮廓线的8-邻域链码在角点附近的变化规律,提出了一种快速而有效的掌纹图像的轮廓特征点检测方法.首先,采用LOG边缘算子检测出二值掌纹的图像的边缘并用轮廓跟踪方法获得掌轮廓线的8-邻域链码;其次,根据手掌轮廓的8-邻域链在特征点附近的变化规律精确地检测出手掌轮廓特征点;最后,根据检测出的特征点建立参考坐标系,对掌纹图像进行定位.实验结果表明该方法是快速而有效的.

The crucial issue of precise location for on-line palmprint image is to detect corner of hand-shape contour accurately. The existing corner detection algorithm can not extract corner useful for location automatically. Through analyzing the regularity of chain-code of hand-shape contour, a rapid and effective corner-detection method was proposed in this paper. Firstly, the method of LOG operator and eight neighborhood boundary-tracking algorithm is used of to extract the hand-shape contour and its chain-code. Then, according to the regularity of hand-shape contour chain-code around corner, the corner of hand-shape contour is detected accurately. Finally, establish a coordinate system according to the corner detected and locate the palmprint image. The experiment results demonstrate the effectiveness and rapidness of the proposed method.