摘要
面向采样间隔未知的地面激光扫描点云数据,提出了一种使用随机邻域分析的采样间隔估算方法。在扫描数据中随机选取n个中心点并构建k邻域,计算每个中心点光束与相应邻域点光束的水平和竖直方向夹角;以区间宽度△统计水平和竖直夹角数目并构建直方图,取点数第2多的区间内的夹角均值作为当前△下的水平和竖直采样间隔;在此基础上,以多个△参数得到的采样间隔中位数作为最终估算结果。该文在不同扫描仪采集的数据上进行多组实验,并在MathWorks平台上共享实验代码。实验结果表明,所提方法的估算误差小于0.002°,对不同地物类型和参数设置具有很好的稳定性。
This paper focuses on the estimation of sampling interval from terrestrial laser scanning data and presents an estimation method based on the neighboring analysis of randomly selected points.We select n central points from the randomly scanning data and search for k nearest points of each central point.The horizontal and vertical(with Z axis)angles between the laser beam of each central point and corresponding neighboring points are calculated.Then histograms of horizontal and vertical angles are constructed respectively,with interval step△.The average angle of the interval with the second largest number of points is recognized as the corresponding horizontal or vertical sampling interval.Finally,the median value of the horizontal or vertical sampling values generated from a series of△values is used as the final estimation result.Tests are carried on the data obtained from different scanners and test code is shared on the MathWorks community.The test results show that the error of our method is smaller than 0.002°with good robustness with respect to different types of targets and parameter settings.
作者
陈茂霖
张昕怡
刘祥江
姬翠翠
赵立都
CHEN Maolin;ZHANG Xinyi;LIU Xiangjiang;JI Cuicui;ZHAO Lidu(School of Smart City,Chongqing Jiaotong University,Chongqing 400074,China)
出处
《应用科学学报》
CAS
CSCD
北大核心
2022年第6期984-995,共12页
Journal of Applied Sciences
基金
国家重点研发计划(No.2021YFB2600603)
国家自然科学基金(No.41801394)
重庆市教委科学技术研究项目(No.KJQN202000746)
重庆交通大学研究生科研创新项目(No.CYS22436)资助。
关键词
地面激光扫描
采样间隔估算
邻域分析
角分辨率
点云数据
terrestrial laser scanning
sampling interval estimation
neighboring analysis
angular resolution
point cloud
