摘要
针对现有的LiDAR点云数据抽稀算法存在无法有效保留地形特征点或地形分类精度不高等问题,文章提出一种顾及地形特征的点云数据抽稀算法:利用点云中的局部极值点与点云边界点作为种子点构建不规则三角网(TIN);利用一定原则逐渐选取非种子点中的地形特征点加密TIN;然后采用一种临近三角面的平面测试策略剔除三角网中可能存在的冗余点,得到最终结果。测试结果表明:该算法在保证地形精度的前提下,能够有效地减少冗余点数量;同时,为了提高算法的实用性,该文通过大量试验给出了算法中所需参数的最优配置。
Aiming at the problem that existed point cloud thinning methods cannot either efficiently keep the terrain feature points or realize high-precise terrain classification, the paper proposed a method considering terrain features： the boundary of the raw point cloud and the extreme points of each grids were used as seed points to build the TIN; and the TIN was interpolated by the terrain features points found out in non-seed points using a specific method step by step; then the planar redundant points was removed by using a planarity testing method of adjacent triangle plane to get the final result. Testing result showed that the method could efficiently reduce the number of redundant points with ensuring the accuracy of terrain classification. Moreover, the paper gave the optimal parameters of the algorithm through a lot of experiments.
出处
《测绘科学》
CSCD
北大核心
2016年第9期140-146,共7页
Science of Surveying and Mapping
基金
国家自然科学基金项目(41271374)
中国测绘科学研究基本科研业务专项(7771504)
测绘地理信息公益性行业科研专项(B1506)
