提出了一种基于可见性约束的自动表面重构算法.此算法首先介绍了一种通过插入法实现的3维Delaunay三角网自动重构算法,然后利用给定的离散点,构建包围所有点的凸包.在完成初始的Delaunay三角剖分基础上,提出了利用特征点在影像中的可见性信息,对初始的3维模型进行修正,从而得到物体的实际表面模型.实验结果表明,该方法是有效的.

This paper proposed a visibility constrain based automatic surface model reconstruction algorithm. This algorithm firstly introduced an automatic 3-dimensional Delaunay triangulated network modeling construction algorithm which is realized by insertion method, then constructed convex hulls enveloping all the known irregular points. On the base of initial Delaunay triangulated network, we proposed a new method of using the visibility information of feature points in images to calibrate the initial 3-dimensional model, and thus getting the actual surface model of object. The experiment results showed its reliability.