摘要
无人机载LiDAR是一种新型遥感技术,可快速获取高密度、高精度的地物点云数据,随着获取精度越高,点云数据量越大。为减少参与点云滤波处理的数据量,提高滤波速度,本文结合点云回波次数、高度和点云回波强度三种属性信息,采用箱形图检验方法获取非地面点云,并使用反复建立三角网方法分别对处理前后的点云数据进行对比分析。结果表明:针对截取的试验区点云数据,可去除多次回波对应的噪声点云363个,得到第1次和第2次回波分别占点云总数的98.49%和1.49%;高度和强度离群的点云共占点云总数的1.46%。通过对点云进行滤波前期的处理,可降低滤波时间32.2 s,且保留地面点云2551个。本文研究利用箱形图检测点云属性获取离群值,可提高点云处理速度和降低地面点云漏分的现象,为后期直接利用统计方法对点云属性信息分析和实现快速处理海量点云提供参考。
Unmanned aerial vehicle(UAV)light detection and ranging(LiDAR)technology is a new means of remote sensing technology that can quickly acquire terrain point clouds with the characteristics of high-density and high-precision.But as accuracy of the data is higher,the amount of data is larger.In order to reduce the amount of data which involved in point cloud filtering and improve the filtering speed,this paper aims to extract a large number of non-ground point clouds.We introduced three kinds of attribute information to deal with point clouds data before filtering.There are echo times of point clouds,elevation values and echo intensity of point clouds,which can advantageously be exploited to enhance the precise of filtering approaches.Combining the three attribute information of point clouds,we used box-plot to identify outliers and then obtained non-ground point clouds.Firstly,the attribute information of original point clouds was extracted.Secondly,we used the box-plot to analyze outlier of point clouds corresponding to three attribute information.Finally,we used the repeated establishment of the triangulation network to filter dataset which combined point clouds separating the height and intensity outliers with corresponding to the second echo times,and then compared filtering results corresponding to before and after using box-plot method processing.The results show that the first and second echo times point clouds number respectively account for 98.49%and 1.49%of original point clouds total number after removing 363 point clouds of multiple echo times point clouds.Point clouds of high and intensity outliers account for 1.46%of original point clouds total number.Compared with the result of filtering original point clouds directly,it shows that the method which used in this paper before filtering reduces filtering time by 32.2s and increases the number of ground point clouds by 2551.That means box-plot method which can realize point clouds filtering pre-processing can improve the efficiency and speed of filtering
作者
李沛婷
赵庆展
田文忠
马永建
陈洪
LI Peiting;ZHAO Qingzhan;TIAN Wenzhong;MA Yongjian;CHEN Hong(College of Information Science and Technology/Division of National Remote Sensing Center,Xinjiang Production and Construction Corps/Geospatial Information Engineering Research Center,Xinjiang Production and Construction Corps,Shihezi,Xinjiang 832003,China;College of Mechanical and Electrical Engineering,Shihezi University,Shihezi,Xinjiang 832003,China)
出处
《石河子大学学报（自然科学版）》
CAS
北大核心
2019年第6期786-792,共7页
Journal of Shihezi University(Natural Science)
基金
中央引导地方科技发展专项(201610011)
兵团空间信息工程技术研究中心创建(2016BA001)
