基于区域Voronoi图的复杂形体的三维重建算法

Reconstruct Complex 3D Object Algorithm Based on Region Voronoi Diagram

针对目前由二维轮廓线重建三维实体的方法要求轮廓线间距小、形状相似,对轮廓分叉及孔洞的解决有诸多限制且计算量大费时等问题,提出了Delaunay三角剖分与体素重建的思想相融合的方法。首先将断层图像轮廓近似表示成平面多边形,并利用Voronoi图进行划分,然后对其进行三角剖分。最后通过最短路径把这些三角形连接到相邻断层轮廓线的顶点上,扩展成四面体,从而直接完成三维重建。试验结果表明,该方法能较好地解决轮廓的分叉、孔洞等问题,且减少了数据量。

In view of the various problems in the reconstruction of 3D solid from c2D contours, a method of 3D reconstruction by combining the Delaunay triangulation and volume reconstruction is presented. First, the cross-section contours are approximated with planar straight-line graph, and parted by using the Voronoi diagram, then generated by a conforming constrained Delaunay triangulation. At last, the triangle and vertex of slice contours are connected by a nearest neighbor to form tetrahedron and achieve reconstruction. It is proved that the branching and correspondence problems can be better solved by this method, and the data are also reduced.