针对山区点云的渐进加密三角网滤波改进算法  

Improved Filtering Algorithm of Progressive TIN Densification for Point Cloud in Mountain Areas

作　　者：王欢 张翰超 张艳 朱宏 WANG Huan

机构地区：[1]自然资源部第三航测遥感院,成都610100 [2]中国测绘科学研究院,北京100036

出　　处：《地理空间信息》2020年第12期27-30,I0005,共5页Geospatial Information

基　　金：自然资源部第三航测遥感院实验室开放基金资助项目(SYS2017KFJJ06)。

摘　　要：利用机载LiDAR点云滤波快速提取高精度数字地面模型已在测绘领域得到广泛应用。针对渐进加密三角网滤波算法在处理地形起伏较大地区出现山脊被削平现象,提出一种基于规则约束的山脊处种子点提取方法。通过优化种子点,使得地面点加密在更加完整的初始地表上进行,滤波结果呈现山脊被削平现象则大大改善,提升了渐进加密三角网滤波算法对山区地形的处理能力。实验结果表明,通过基于规则约束的山脊处种子点提取方法提取到的种子点正确率达93%,利用滤波结果生成的数字高程模型(DEM)与参考DEM高程差值的RMSE由3.72 m提升到0.88 m,证明了通过优化山脊处种子点可切实提高PTD算法对山区地形的滤波精度。Airborne LiDAR point cloud is widely used in the field of surveying and mapping.To solve the problem of the ridge was flattened in the region with large topographic undulations,we proposed a method of extracting seed points at ridge based on rule constraint.By optimizing the seed points,the ground points were encrypted on a more complete initial surface,and the filtering result show that the flattening of the ridge was greatly improved,which could greatly improve the processing ability of the progressive TIN densification filtering algorithm for mountain terrain.The experimental result shows that the accuracy of seed points obtained by the seed point extraction method based on rule constrain at the ridge reaches 93%,and the RMSE of elevation difference between the generated DEM and the reference DEM from 3.72 m to 0.88 m,which can prove that optimizing seed points at the ridge can improve the filtering accuracy of PTD algorithm.

关 键 词：LIDAR 渐进加密三角网 山区 种子点 

分 类 号：P237[天文地球—摄影测量与遥感]