基于稳健加权总体最小二乘的点云数据平面拟合  

PLANE FITTING OF POINT CLOUDS BASED ON ROBUST WEIGHTED TOTAL LEAST SQUARES

作　　者：欧江霞 李明峰[1] 王永明[2] 徐燕[1] 

机构地区：[1]南京工业大学地球空间信息研究中心,南京210009 [2]武汉大学测绘学院,武汉430079

出　　处：《大地测量与地球动力学》2014年第3期160-163,共4页Journal of Geodesy and Geodynamics

基　　金：国家自然科学基金项目(41274009);江苏省研究生科研创新计划项目(CXLX13_422);南京市科技计划项目(201101069);江苏省测绘科研项目(JSCHKY201108)

摘　　要：针对经验权值偏差及扫描数据异常点对传统加权整体最小二乘点云数据平面拟合精度的影响,提出既可自适应修正观测向量权值及系数矩阵权值,又能剔除异常数据的稳健加权整体最小二乘算法。将新算法应用于拟合不同扫描距离获取的点云数据,算例表明,与常规方法相比,该算法的单位权中误差较小,平面拟合精度较优。A method of conventional weighted total least squares was proposed based on the robust weighted total least squares for eliminating the influences of point clouds anomalous points in laser scanning data on fitting precision. The method can be used to correct adaptively the weight of observation vector and and coefficient matrix, and remove anomalous data. The result of fitting analysis with the point clouds from different distances shows that the fitting accuracy with the method is better than the conventional methods.

关 键 词：点云数据 平面拟合 最小二乘 稳健估计 权值 

分 类 号：P207[天文地球—测绘科学与技术]