基于有向点云数据的二次曲面拟合算法  

Algorithm of Quadric Surface Fitting from Oriented Point Cloud

作　　者：刘元朋[1] 赵辉[1] 陈良骥[1] 王金凤[1] 

机构地区：[1]郑州航空工业管理学院机电工程系,河南郑州450015

出　　处：《机床与液压》2008年第8期27-29,33,共4页Machine Tool & Hydraulics

基　　金：总装备部预先研究资助项目(41318.1.1.7);河南省科技攻关资助项目(0324250068)

摘　　要：针对有向点云数据的特点,提出了一种基于几何距离的二次曲面(如平面、球面、圆柱面)拟合算法。该算法利用非线性最小二乘法Levenberg-Marquardt进行求解,并基于二次曲面的高斯图特性给出了拟合算法的初始值估计方法。实验证明所提出的拟合算法能够有效地解决有向点云数据的二次曲面拟合问题。A new algorithm of quadric surface fitting was presented, The fitting models based on the geometric distance like planes, cylinders, spheres, cones to the oriented point cloud was addressed. Nonlinear least squares solution by Levenberg-Marquardt algorithm was given, novel techniques were presented to estimate initial parameters of the fitting models based on the speciality of the quadric gaussian image. Experimental results using real measured data demonstrate the effectiveness of algorithms.

关 键 词：点云 二次曲面拟合 高斯图 LEVENBERG-MARQUARDT算法 

分 类 号：TP311.11[自动化与计算机技术—计算机软件与理论]