森林复杂场景的布料模拟点云滤波改进方法  

Improved Method of Cloth Simulation Point Cloud Filtering for Forest Complex Scene

作　　者：王海军 肖海 罗争光 刘心成 WANG Haijun;XIAO Hai;LUO Zhengguang;LIU Xincheng(The Second Surveying and Mapping Institute of Hunan Province,Changsha 410100,China)

机构地区：[1]湖南省第二测绘院,湖南长沙410100

出　　处：《地理空间信息》2024年第12期38-41,共4页Geospatial Information

基　　金：湖南省自然资源领域推进碳达峰碳中和重大技术研究资助项目(〔2022〕5号)。

摘　　要：针对机载LiDAR地面点滤波算法在复杂场景中,全局统一的滤波参数设置难以满足不同地形特征变化的地面点分类精度的问题,提出一种基于最低点中位数曲线判断地形变化并结合布料模拟的地面点滤波算法。首先构建点云最低点中位数曲线,通过该曲线的坡度判断地形变化状态,依据不同地形设置局部布料模拟滤波参数,实现地面点精准分类,然后将通过插值生成的数字表面模型和数字高程模型相减获得归一化数字表面模型。实验结果表明:本研究提出的改进方法获取的地面点分类结果和归一化数字表面模型的精度都明显高于坡度滤波、二次曲面滤波和全局布料模拟滤波方法。Aiming at the problem that the global unified filtering parameter setting of airborne LiDAR ground point filtering algorithm is difficult to meet the classification accuracy of ground points with different terrain features in complex scenes,we proposed a ground point filtering algorithm based on the lowest point median curve to judge the terrain changes and combined with cloth simulation.Firstly,we constructed the median curve of lowest point of point cloud,and judged the terrain change state by the slope of curve.Then,according to different terrain,we set the local cloth simulation filtering parameters to realize the accurate classification of ground points.Finally,we obtained the normalized digital surface model by subtracting the digital surface model and digital elevation model generated by interpolation.The experimental results show that the accuracy of ground point classification results and normalized digital surface model obtained by the improved method are significantly higher than that of slope filtering,quadric surface filtering and global cloth simulation filtering.

关 键 词：机载LIDAR 地面点滤波 归一化数字表面模型 布料模拟 

分 类 号：P231[天文地球—摄影测量与遥感]