【作者】 王芃芃；

【导师】 隋立春；

【作者基本信息】 长安大学 ， 摄影测量与遥感， 2011， 硕士

【摘要】 机载LiDAR点云数据的分类滤波是其数据后处理的重要组成部分,也是后续数字产品生产的基础,因此滤波算法的实践和深入研究具有一定的价值和意义。本文深入分析了机载LiDAR技术的现状和滤波算法的国内外现状、机载LiDAR系统原理、点云数据特点和评判滤波后的误差指标等,以国际摄影测量与遥感协会提供的标准样本点云数据为基础进行滤波算法的研究与分析,并定性和定量分析了滤波算法的效果。本文的研究工作体现在如下：1、系统阐述了机载LiDAR技术的背景和应用意义、国内外LiDAR点云数据滤波算法的研究现状、机载LiDAR系统原理和组成,总结了LiDAR数据后处理常用软件,论述了机载LiDAR点云数据的特点和处理流程,为试验和后续数据处理提供了理论依据。2、针对机载LiDAR点云数据的复杂性和离散性等特点,本文研究了数学形态学原理应用于机载LiDAR点云数据的处理,通过算法测试了国际摄影测量与遥感协会提供的标准样本点云数据,研究了数学形态学滤波算法的流程。针对数学形态学中的窗口、格网间距和高差阈值的不同进行了大量的实验。3、在对机载LiDAR点云数据预处理的基础上,本文探索并实验了顾及因果关系的二维自回归模型算法应用于机载LiDAR点云数据的后处理。更多还原

【Abstract】 The filtering of airborne laser scanning points cloud is one of the critical data processing technologies, therefore, the deep research of filtering algorithms is practically valuable and useful. In the area of photogrammetry and remote sensing, it is full of frontier research task. This paper analysed and summarized airborne LiDAR status, systematic composition, research actualities of filtering, the acquisition theory and characteristics of point cloud, and the standards of judging filtering errors. This paper tested the algorithms of point cloud filtering and post-processing and made the precision analysis. The research work embodies as follows:1. On the basis of introducing the airborne LiDAR technique research background and significance, domestic and foreign LiDAR point cloud data filtering algorithms, the research status, the system theory and composition and their main parameters comparisons, and list the LiDAR data post-processing commonly used software, this paper discusses the airborne laser scanning point cloud data processing steps, which provides the theory basis for the experiments.2. Aiming at the characteristics of the complexity and dispersion of the point cloud data, this paper studied the morphological filtering used in processing of LiDAR point cloud, tested standard sample data given by ISPRS according to algorithm, did lots of experiments about window, grid spacing, elevation difference threshold as the three parameters to weed out the non-ground points, we can get different ground point cloud data on the same block of data by giving different window size, grid spacing and elevation difference threshold for filtering. Contrasts of the errors displayed the better filtering results.3. According to procedure, on the basis of pre-processing of airborne LiDAR point cloud data, this paper explored Causal Auto-Regressive process Model as one method of the Robust Estimation to process the point cloud data.更多还原