摘要
在起伏不定的陡坡地表,地面点云与植被点云相互关联,难以将二者准确分离。鉴于此,针对地形复杂、地势起伏大的陡峭山坡,该文提出了一种基于多回波及Fisher判别的滤波算法。首先结合格网划分思想划分点云网格,在每个网格内,根据点云的回波次数和回波强度进行初步分类,划分出待定样本;其次,利用Fisher判别分析法将点云投影到一维空间,通过判定临界值将植被点云与地面点云分离,实现陡坡点云的滤波;最后,从定性与定量两方面将该算法与数学形态学滤波算法、迭代三角网滤波算法和布料模拟滤波算法进行比较,结果表明:在陡峭山坡情况下,该文提出的算法总误差最小,整体滤波效果更好,验证了该方法在陡坡点云滤波中的适应性与有效性。
Multiple echo information of laser pulses can reveal the type of the tested objects,which provides a new idea for point cloud filtering.In the surface of undulating steep slope,the ground point cloud is associated with the vegetation point cloud.It is difficult to separate them accurately.Thus,for the conditions of steep slopes with complex terrain and surface,a filtering algorithm based on multiple echo information and Fisher discrimination was proposed to solve the problem.Firstly,the point clouds were divided into grids based on grid partition theory,and the grid spacing should be apposite.Then,in order to divide out the samples to be determined,the cloud points in each grid were classified initially by echo times and echo intensity of the clouds.In addition,in order to separate the ground point cloud from the vegetation point cloud and achieve the filtering of steep point cloud,the point clouds were projected into one-dimensional space by Fisher discrimination analysis.Finally,the algorithm was compared with the mathematical morphology filtering algorithm,the iterative triangulation filter algorithm and the cloth simulation algorithm from both quantitative and qualitative aspects.The results show that in the case of steep slopes,the total error of the proposed algorithm is minimal and has a better overall filtering effect.The adaptability and effectiveness of the method in cloud filtering of steep slope are verified.
作者
孙渡
李永强
吴珍珍
刘会云
刘洋洋
黄腾达
SUN Du;LI Yongqiang;WU Zhenzhen;LIU Huiyun;LIU Yangyang;HUANG Tengda(School of Surveying and Land Information Engineering,Henan Polytechnic University,Jiaozuo 454000,China)
出处
《地理与地理信息科学》
CSCD
北大核心
2018年第2期34-39,共6页
Geography and Geo-Information Science
基金
河南省基础与前沿技术研究项目(162300410184)
测绘地理信息公益性行业科研专项经费项目(201412020)
关键词
三维激光扫描
多回波信息
FISHER判别
点云滤波
3D laser scanning
multiple echo information
Fisher discrimination
point cloud filtering
