摘要
在分析现有滤波算法的基础上,结合数据驱动和模型驱动式算子各自的优点,提出基于点云空洞修复和TPS变形模型的数学形态学机载LIDAR点云滤波,该方法首先提取和修复由水域造成的大面积点云空洞,采用多尺度形态学开算子作用于修复的数据,得到近似裸露地表面;然后利用2D空间的TPS变形模型,以近似地表面为基础,插值原始点云,根据插值与原始点云高程的差值大小去识别地面点和非地面点。通过定量分析,验证该方法不仅有较高的滤波精度,而且也能较好的保留裸露地表的细节特征,同时该方法有助于辅助人工处理,提高数据处理的质量。
Based on the basic analysis of existing filtering algorithm,combiningrespective advantages of datadriven and model-driven,a filtering method is proposed based on mathematical morphological of the repairing point cloud hole and TPS model.This method first extracts and repaires the point cloud,using multi-scale morphological opening operator acting on the restoration of data to form an approximate bare ground surface.Then by the TPS deformation model of 2D space based on approximate surface,interpolation of the original point cloud,the difference between the original point cloud elevation interpolation,ground point and non-ground points are identified.In this paper,through quantitative analysis shows that the method not only has high filtering precision but also better retains the minutiae of bare surface.Meanwhile,the method can also help auxiliary artificial processing and improve the quality of data processing.
出处
《测绘工程》
CSCD
2016年第7期28-32,共5页
Engineering of Surveying and Mapping