机载LiDAR点云的滤波分类研究

发布时间：2018-02-04 19:08

本文关键词： 机载LiDAR 点云滤波分类　出处：《中国矿业大学（北京）》2013年博士论文　论文类型：学位论文

【摘要】：机载LiDAR系统是一项新兴的测绘技术。论文系统总结和分析了机载LiDAR系统的测距原理。基于LiDAR点云的高程直方图及点云间的相互关系滤除了点云噪声，提出了分类LiDAR地面点云的三种算法：改进参数的形态学法、二元二次趋势面结合形态学法、窗口迭代的克里金法，经过试验数据验证知，三种方法均取得了较好的分类结果。利用CUMT_PC法和CUMT_AT的方法分别分类出建筑物点云，并提取出了屋顶面域，试验结果表明两种方法的分类精度可以达到中等水平，，且屋顶面域的探测率和质量因子都高于92%，能达到自动识别的要求。根据点云的高程和反射强度信息分类出城市中的道路点云，提出了一种城市主干路网的提取算法，提取的城市道路网与参考数据对比后发现，所提取道路网的完整率和正确率较高。
[Abstract]:Airborne LiDAR system is a new technology of surveying and mapping. The principle of airborne LiDAR system ranging is summarized and analyzed in this paper. The elevation histogram based on LiDAR point cloud and the relationship between point clouds are filtered. Except point cloud noise. Three algorithms for classifying LiDAR ground point clouds are proposed: morphological method of improved parameters, binary quadratic trend surface combined with morphology method, window iterative Kriging method, and verified by experimental data. The CUMT_PC method and the CUMT_AT method are used to classify the building point cloud, and the roof area is extracted. The experimental results show that the classification accuracy of the two methods can reach medium level, and the detection rate and mass factor of the roof area are both higher than 92%. According to the elevation and reflection intensity information of the point cloud, the road point cloud in the city is classified, and an algorithm for extracting the urban trunk road network is proposed. Comparing the extracted urban road network with the reference data, it is found that the integrity and accuracy of the extracted road network are higher.
【学位授予单位】：中国矿业大学（北京）
【学位级别】：博士
【学位授予年份】：2013
【分类号】：P225.2

【引证文献】