基于最小二乘的点云叶面拟合算法研究  

Leaf Surface Fitting of Point Cloud Based on Least Square Method

作　　者：刘俊焱 云挺[1] 周宇[1] 薛联凤[1] 李正军[1] 

机构地区：[1]南京林业大学信息与科学技术学院,江苏南京210037

出　　处：《西北林学院学报》2014年第5期70-77,155,共9页Journal of Northwest Forestry University

基　　金：国家自然科学基金(31300472);江苏省自然科学基金(BK2012418)

摘　　要：利用地面激光扫描仪(TLS)获取户外树木的大量点云数据,从中截取树叶点云数据并以此来进行曲面拟合,构建树叶真实三维模型。主要针对空间散乱点云数据的曲面拟合方法进行研究。在最小二乘法、正交最小二乘法、移动最小二乘法等3种二维曲面拟合方法基础上,针对空间散乱点云数据,提出新的曲面拟合方法。通过比较这3种方法对点云数据曲面拟合后效果,得出结论:针对树叶散乱点云数据,移动最小二乘法能够有效的拟合出树叶的曲面。Point cloud data of outdoor trees were obtained by terrestrial laser scanner（TLS）,by which the data of leaf part were intercepted for blade curve surface fitting and real 3Dmodel construction.The curve surface fitting method on irregular space scattered point cloud data were studied.Improvements based on least square method,orthogonal least square method and moving least squares were made,and new method on irregular spaces scattered point cloud data were put forward.Experimental results showed that the moving least squares method could fit the blade＇s curved surface effectively.

关 键 词：激光扫描仪 点云数据 曲面拟合 最小二乘 正交最小二乘 移动最小二乘 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]