欢迎访问《图学学报》 分享到:

图学学报

• 几何设计与计算 • 上一篇    下一篇

基于点云的球面三维逆向建模

  

  • 出版日期:2013-06-29 发布日期:2015-06-11

3D Converse Modeling for Sphere Point Cloud

  • Online:2013-06-29 Published:2015-06-11

摘要: 球面拟合是三维逆向建模中面临的亟待解决的复杂难题之一。它广泛应用
于零件检测、建筑物结构恢复建模、医学血管和细胞模拟领域。论文改进了对三维球面点云
进行直接拟合的方法,得到球面的相关几何参量,并在速度和精确度上都获得了提高。论文
对拟合相关参数和噪声影响进行分析,并否定了对选定的数据集分组整合的算法。LM
(Levenberg-Marquardt)算法是最广泛应用的最小二乘法二次曲面拟合的方法之一,本算法在
与LM 算法的对比中凸显了算法在时间和某些情况下精度的优越性,为进一步研究三维图形
拟合重建恢复等工作打下了基础。

关键词: 拟合, 球面, 三维逆向建模

Abstract: Sphere fitting is one of the complex and waiting prompt solution problems in the
field of 3D Converse Modeling, which can be used for the parts detection, the recovery of the
architecture, medicine and so on. The paper offers a method to directly fit sphere, called DF
algorithm, and acquire the geometric parameters with improvement at speed and accuracy. The
paper also pays attention to data analysis for the cause of the results and the comparison with the
LM (Levenberg-Marquardt) algorithm, which is the one of the most widely used nonlinear least
square methods, laying foundations for the rebuilding the parts and the gives of the further work.

Key words: direct fitting, sphere, 3D converse modeling