A contour-parallel offset (CPO) tool-path linking algorithm is derived without toolretractions and with the largest practicability. The concept of "tool-path loop tree" (TPL-tree) providing the information on th...A contour-parallel offset (CPO) tool-path linking algorithm is derived without toolretractions and with the largest practicability. The concept of "tool-path loop tree" (TPL-tree) providing the information on the parent/child relationships among the tool-path loops (TPLs) is presented. The direction, tool-path loop, leaf/branch, layer number, and the corresponding points of the TPL-tree are introduced. By defining TPL as a vector, and by traveling throughout the tree, a CPO tool-path without tool-retractions can be derived.展开更多
针对现有无人机测绘点云偏移校正方法多依赖曲率特征筛选配准点、在复杂地形条件下特征辨识度低从而导致校正精度不足的问题,提出一种基于改进快速探索随机树(Rapidly-exploring Random Tree,RRT)算法的点云偏移校正方法。首先,利用改进...针对现有无人机测绘点云偏移校正方法多依赖曲率特征筛选配准点、在复杂地形条件下特征辨识度低从而导致校正精度不足的问题,提出一种基于改进快速探索随机树(Rapidly-exploring Random Tree,RRT)算法的点云偏移校正方法。首先,利用改进RRT算法在待校正点云中自适应筛选子区域并完成匹配,获取可靠对应点对;其次,采用最小二乘法与奇异值分解(Singular Value Decomposition,SVD)相结合的方法求解偏移变换矩阵;最后,以该变换矩阵为核心,对边缘点进行二次优化,实现校正点云的整体整合与优化。实验结果表明,在城区建筑密集区、山地林地等七类典型场景中,该方法的平均绝对误差(Mean Absolute Error,MAE)均达到最优水平;在人工注入0.1~0.6 m梯度平移偏移条件下,均方根误差(Root Mean Square Error,RMSE)仅由0.32 cm增至0.71 cm,在不同场景及偏移量级条件下均表现出良好的高精度校正性能。展开更多
摘要A contour-parallel offset (CPO) tool-path linking algorithm is derived without toolretractions and with the largest practicability. The concept of "tool-path loop tree" (TPL-tree) providing the information on the parent/child relationships among the tool-path loops (TPLs) is presented. The direction, tool-path loop, leaf/branch, layer number, and the corresponding points of the TPL-tree are introduced. By defining TPL as a vector, and by traveling throughout the tree, a CPO tool-path without tool-retractions can be derived.
摘要针对现有无人机测绘点云偏移校正方法多依赖曲率特征筛选配准点、在复杂地形条件下特征辨识度低从而导致校正精度不足的问题,提出一种基于改进快速探索随机树(Rapidly-exploring Random Tree,RRT)算法的点云偏移校正方法。首先,利用改进RRT算法在待校正点云中自适应筛选子区域并完成匹配,获取可靠对应点对;其次,采用最小二乘法与奇异值分解(Singular Value Decomposition,SVD)相结合的方法求解偏移变换矩阵;最后,以该变换矩阵为核心,对边缘点进行二次优化,实现校正点云的整体整合与优化。实验结果表明,在城区建筑密集区、山地林地等七类典型场景中,该方法的平均绝对误差(Mean Absolute Error,MAE)均达到最优水平;在人工注入0.1~0.6 m梯度平移偏移条件下,均方根误差(Root Mean Square Error,RMSE)仅由0.32 cm增至0.71 cm,在不同场景及偏移量级条件下均表现出良好的高精度校正性能。