一种改进的自适应三角剖分算法  

Improvement to adaptive triangulation algorithm

作　　者：蒋恒恒[1] 李奇敏[1] 汤宝平[1] 

机构地区：[1]重庆大学机械传动国家重点实验室,重庆400030

出　　处：《计算机应用研究》2011年第10期3950-3953,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(50905190)

摘　　要：现有的大多数散乱点云三角剖分算法存在细节特征表现不足和适应性不强的问题,为此改进了一种自适应的三角网格剖分算法。此方法将Shepard曲面插值与多尺度分析方法相结合;引入改进的八叉树搜索思想,加细搜索进而估算出点云中每个测量点的曲率;生成带自适应分辨率的分层空间栅格,最终实现自适应的三角网格重构。实验结果表明,经改进的算法,形成的三角网格质量较高,能够较好地再现原三维物体的细节特征,且效率较高,适用广泛。This paper improved the adaptive triangulation algorithm for three dimensional unorganized point clouds,since the most existing algorithms were not very adaptable,and they were difficult to express the detail characters of the real surface well.In the proposed method,it combined 4D Shepard surface with multi-resolution analysis,and implemented the modified octree algorithm,which the curvature of each point in the point cloud was estimated.Then constructed a hierarchical grid with adaptive resolution for generating a triangular mesh from point clouds.Experimental results show that the improved algorithm is greatly advanced and generally applicable,and forms high quality triangle grid surface and reproduces initial 3D object＇s detail characters,which is suited to popularize in CAGD and surface modeling.

关 键 词：点云 三角剖分 Shepard插值 八叉树搜索 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]