摘要
由一组二维轮廓线重建出物体的三维表面是医学数据可视化的一种主要绘制方式。当轮廓线比较复杂,例如当遇到非凸轮廓或相邻层轮廓线相差过大时,常用的三角化拼接方法就会失败。文章提出一种新的轮廓拼接方法能够处理任意形状的轮廓线。该方法的基本思想是对轮廓线进行凹凸性层次分析,然后将相邻轮廓线从外到内逐层拼接,从而构成一个三角化的物体表面。实验结果表明,该算法对于手动勾画和自动提取的轮廓线都可以给出较好的重建效果。
Three dimensional surface reconstruction from planar sectional contours is the main method in medical image visualization and many other applications.However,conventional triangulation algorithms suffer from limitations when planar contours are not convex,or when the contours in two successive slices are deformed very much.We propose a new triangulation approach aiming to handle arbitrary shaped contours in this paper.The method first calculates the convex hulls of every contour and its sub-contours,in order to decompose arbitrary shaped contours into levels of elementary convex or concave segments.It then triangulates these segments of two successive contours one level by one level with minimal-span strategy,and produces a triangulated mesh from the data points of the two contours.Finally we get a reasonable 3-D object surface from a series of planar sectional contours.The proposed algorithm is applied to object surface recovery from automatic extracted planar contours as well as manual drawn ones,and the experimental results demonstrate its practicality and effectiveness.
出处
《计算机工程与应用》
CSCD
北大核心
2006年第12期74-76,92,共4页
Computer Engineering and Applications
关键词
表面重建
轮廓重建
三角化
轮廓线
凸包
最短跨距准则
surface reconstruction,reconstruction from contours,triangulation,planar contour,convex hull,minimal -span strategy
