混合整体最小二乘曲面拟合在点云滤波中的应用  

Mixed LS-TLS surface fitting in point cloud filtering

作　　者：吉文来 芮明 李明峰 JI Wenlai;RUI Ming;LI Mingfeng(School of Geomatics Science and Technology,Nanjing Tech University,Nanjing 211816,China;Changzhou Institute of Surveying and Mapping,Changzhou 213002,China)

机构地区：[1]南京工业大学测绘科学与技术学院,江苏南京211816 [2]常州市测绘院,江苏常州213002

出　　处：《测绘通报》2020年第8期50-54,91,共6页Bulletin of Surveying and Mapping

基　　金：江苏省测绘地理信息科研项目(JSCHKY201809)。

摘　　要：针对传统点云滤波过程中拟合曲面不连续,参数求解不合理及高差阈值不明确的问题,本文提出一种基于混合整体最小二乘的曲面拟合约束法。利用中心网格加权坡度值构建了多尺度格网,探讨了依据网格邻域选取地面种子点的标准,分析了混合整体最小二乘曲面方程求解和根据滤波结果逐级调整高差阈值的方法和步骤。试验结果表明,该方法提高了复杂区域的点云滤波精度,为构建数字高程模型提供了精确数据。Aiming at the problems of the discontinuity of the fitted surface,the unreasonable parameter solution and the unclear threshold of the height difference in the process of point cloud filtering,a method for constraint surface fitting based on mixed LS-TLS(least squares-total least squares)is proposed.The multi-scale grid is constructed based on the weighted slope value of the center grid,the criteria for selecting the ground seed points based on the grid neighborhood is discussed,and the methods and steps for solving the mixed LS-TLS surface equation and adjusting the height difference threshold according to the filtering result are analyzed.Experiments show that the precision of point cloud filtering in complex regions is improved and benefit for the generation of digital elevation model.

关 键 词：点云数据 多尺度格网 混合整体最小二乘 曲面拟合 滤波 

分 类 号：P23[天文地球—摄影测量与遥感]