论文部分内容阅读
针对形态学运算在机载Li DAR滤波中最大窗口尺寸的选择问题,提出了一种基于形态学开重建的迭代多尺度点云滤波算法。通过循环迭代多尺度开重建运算,克服开重建对矮小地物的误判问题,自动获取开重建运算的最大窗口尺寸,解决了对城市区域滤波的窗口适宜尺寸选择问题,提高了地物与地面点分类的正确性。使用ISPRS提供的城区样本测试数据开展实验,结果表明:其Ⅰ类、Ⅱ类及总误差均值分别达到3.10%、6.05%和4.11%,在Ⅱ类误差不显著增加的情况下,Ⅰ类误差和总误差均值同比均为最小,整体分类与自动识别性能优于常规滤波算法。
Aiming at the selection of maximum window size for morphological computation in on-board Li DAR filtering, an iterative multiscale point cloud filtering algorithm based on morphological reconstruction is proposed. By iterative iterative iterative multi-scale reconstruction operation to overcome the problem of misjudgment of the small objects by reconstruction, the maximum window size of the reconstruction operation is automatically obtained, the suitable size selection of the window for the urban area filtering is solved, Point the correctness of the classification. The experimental data of urban samples provided by ISPRS were used to carry out experiments. The results showed that the average error of class I, class II and total error reached 3.10%, 6.05% and 4.11%, respectively. Under the condition of no significant increase of class II error, The mean total error is the smallest compared with the same period of last year, and the overall classification and automatic identification performance are better than the conventional filtering algorithms.