摘要
针对现有机载激光雷达点云滤波算法在林区适用性不强的问题,提出一种基于多分辨率层次插值的林区LiDAR滤波方法。该方法首先借助形态学迭代开运算和稳健z-score方法获取大量地面种子点;然后从低层到高层滤波过程中,通过薄板样条函数构造地面参考面,并借助自适应坡度阈值选择地面点;最后将分类出的地面点更新地面参考面,层层迭代直至滤波结果收敛。以ISPRS提供的6组山区基准数据为研究对象,将新方法滤波结果与近5年提出的10种滤波算法比较表明:新方法滤波结果精度最高,平均总误差和Kappa系数分别为1.89%和87.88%。在实例分析中,以6个不同林区点云数据为研究对象,将新方法与形态滤波算法(MF)和渐近不规则三角网加密滤波算法(PTD)比较表明:新方法平均总误差为6.82%,而MF和PTD平均总误差分别为9.21%和8.49%;且前者获取的DEM精度优于后两种方法。
To improve the filtering accuracy of the existing algorithms on dense forest areas,in this paper,a multi-resolution hierarchical interpolation-based filtering method for LiDAR data in forest areas was proposed.Firstly,a large number of ground seed points were obtained by performing mathematical morphological operations and robust z-score method.Then,in the hierarchical filtering,the ground reference surface was constructed in each layer through the thin plate spline,and the ground points were selected through the adaptive slope residual threshold.Finally,the classified ground points were used to update the ground reference surface and the multi-resolution hierarchical interpolation filter iterated through layers until the filter results converged.The new method was employed to filter the benchmark rural samples provided by ISPRS and its results were compared with 10 filtering algorithms proposed in the last five years.Results show that the new method has the highest accuracy and the average total error and Kappa coefficient were 1.89%and 87.88%,respectively.Moreover,the new method was used to process the point cloud data of six different forest areas,and the filtering results were compared with those of MF and PTD.Results show that the average total error of the new method was 6.82%while the average total errors of MF and PTD were 9.21%and 8.49%respectively.In addition,the accuracy of DEM obtained by the new method is significantly better than the DEMs obtained by the other two methods.
作者
陈传法
王梦樱
杨帅
王珍
CHEN Chuanfa;WANG Mengying;YANG Shuai;WANG Zhen(College of Geodesy and Geomatics,Shandong University of Science and Technology,Qingdao,Shandong 266590,China;BGI Engineering Consultants LTD,Beijing 10038,China)
出处
《山东科技大学学报(自然科学版)》
CAS
北大核心
2021年第2期12-20,共9页
Journal of Shandong University of Science and Technology(Natural Science)
基金
国家自然科学基金项目(41371367)
山东省自然科学基金项目(ZR2020YQ26,ZR2019MD007)
山东省高等学校青创科技支持计划项目(2019KJH007)。