三维碎片拼合的轮廓提取和特征点检测

摘要

针对三维空间碎片的拼接,提出一种轮廓提取和特征点检测的方法.首先,采用一种简单、有效的方法来构建三维碎片数据的拓扑关系,并在此基础上提取碎片的轮廓曲线.对存在缺陷、拓扑关系难以建立的三维测量数据,文中给出了相应的修改、调整方法.同时,提出了一种对轮廓进行重采样的方法来计算空间轮廓上每一点的曲率和挠率.将曲率和挠率相结合得出轮廓曲线上各点的全曲率,然后由全曲率值来检测碎片轮廓的特征点.轮廓曲线的描述就由其上每一点的曲率和挠率以及特征点来表征.实验表明了该算法的有效性和稳定性.%An approach to contour extraction and feature point detection in the 3-D fragment reassembly is pro-posed. A simple and effective technique is used for building the intrinsic topology of the fragment data suitable forcontour extraction. For the scanned data in which the topology is difficult to be achieved, the corresponding solu-tions are given to manage this problem. A robust approach is used for the curvature and torsion calculation of thediscrete contour in a 3-D space. Finally, a method is developed for detecting feature points of the fragment con-tour based on total curvature. Therefore, the contour description combines the simple global information with lo-cal feature points. Experiments with real contour curves extracted from 3-D fragments demonstrate that the pro-posed method is robust and efficient.