【作者】 袁枫；

【导师】 高井祥； 张继贤；

【作者基本信息】 中国矿业大学 ， 地图制图学与地理信息工程， 2010， 博士

【摘要】 地球空间信息技术是当今世界各国研究的热点之一,信息的获取、处理和应用是其研究的三大主题。上世纪80年代末,机载激光雷达技术(Light Detection and Ranging,LIDAR)在三维地球空间信息的实时获取方面取得了重大突破,为获取高时空分辨率地球空间信息提供了一种全新的技术手段。作为一种新型的主动式直接对地观测技术,机载LIDAR正逐步得到广泛的应用。机载LIDAR系统获取的数据是一系列空间分布不规则的离散的三维点云,如何处理大量的点云数据,从中提取有用的地形和地物信息等,并研究机载LIDAR数据以及与其它数据源的融合在地形测绘、土地利用、城市规划及建设等领域的应用,是当前急需解决的问题。基于此,本论文开展了机载LIDAR数据处理与土地利用分类的研究。本文系统地分析了机载激光雷达技术的原理和特点,对机载激光雷达数据处理以及土地利用分类的关键技术和方法进行了深入的研究和探讨。论文主要研究内容和成果包括:1、系统总结了在机载LIDAR数据条带平差、机载LIDAR数据滤波以及土地利用分类方面的国内外的研究现状。2、介绍了机载激光雷达系统的系统组成和工作原理,并着重介绍了ALS50-II系统,系统总结了机载LIDAR的数据处理的流程和点云数据的特点,详细分析了机载LIDAR数据中的误差。3、在LIDAR数据条带平差的误差理论研究的基础上,针对经过检校后的LIDAR数据中存在的GPS定位误差和INS测姿误差,从LIDAR的严格传感器模型出发,提出一种无需原始观测值的条带平差的数学模型;针对LIDAR数据中连接点难以选取的问题,研究了基于最小二乘3D表面匹配原理的连接点选取方法。试验结果表明该模型能够提高机载LIDAR数据的精度,有效地消除相邻条带数据间存在的方向和位置偏移。4、探讨了机载LIDAR数据滤波的原理及难点,在本文算法假设的基础上,设计并实现了一种机载LIDAR数据的自适应滤波算法。该方法首先将原始LIDAR点云数据内插成规则格网数据,接着基于规则格网数据进行光滑分割,并建立区域邻接矩阵和高程指向矩阵,然后根据分割段之间的几何拓扑关系对分割段进行分类,最后根据分类得到的地面点内插得到地形表面,再从原始点云数据中精确提取出地面点,实现自适应滤波。试验结果表明,该方法整体上优于已有的典型滤波算法,能够有效地进行滤波,得到的DEM保留了地形特征细节,效果较好。5、研究在融合高分辨率的机载LIDAR数据和RCD105获取的彩色航空影像的基础上,进行土地利用分类的方法。首先根据试验区的实际情况,选取一定数量的土地利用类型模板样本,对其进行统计分析,提取出各种典型地物的波谱特征以及空间分布特征;设计和发展了一组特征提取的空间算子,提取土地利用类型的波谱特征、空间分布特征、形状、尺寸等;根据框架理论和试验区内土地利用类型建立了土地利用类型的框架系统,实现了土地利用分类的原型试验系统,并利用试验区的数据进行土地利用分类整个流程的试验,输出试验区的土地利用类型分布图。更多还原

【Abstract】 Geo-spatial information technology is the focus research in the geo-science world presently. Information capturing, processing as well as application are the three main items. At the end of 1980s, Airborne LIDAR technology got break through in real time acquiring of 3D earth spatial information, providing a new technique means for acquiring earth spatial information with high time and spatial resolution. As a new, active and direct earth observation technique, airborne LIDAR is applied widely gradually.The data set obtained by airborne LIDAR system is 3D discrete sub-randomly spatial distributed point cloud. At present, how to process LIDAR point cloud data to extract topographic information and different object information from point cloud, and how to apply airborne LIDAR data and other data source to topographic surveying and mapping, land use, and city construction and planning are key problems of LIDAR research. The application research of LIDAR data in many fields such as topographic mapping, urban construction and forestry programming and so on is the active studying topic too. Therefore, this dissertation develops the research on DEM extraction and classification of land use of airborne LIDAR data.This dissertation comprehensively analyzes airborne LIDAR technology and characteristics. Then, the paper carries on through researches in critical technology and methods of data processing and classification of land use. The main studies and contributions are described as follows:1. The paper systematically summarized the research status strip adjustment of LIDAR data, filtering method of LIDAR data and classification of land use home and abroad.2. This article introduces the composition and working theory of airborne LIDAR system and the composition of ALS50-II system emphatically. Then the process flow of airborne LIDAR data and characteristics of points cloud are discussed and errors in airborne LIDAR data are analyzed.3. Based on errors theory of strip adjustment of airborne LIDAR data, according to errors in LIDAR data after calibration, it puts forward a mathematical model of strip adjustment. This model comes from rigorous sensor model of LIDAR and doesn’t need any raw measurements. Considering difficulties of tie points selection from LIDAR data, it takes research on method of tie points selection based on least squares 3D surface matching theory.4. Theory and difficulties of filtering of airborne LIDAR data are discussed. Based on the assumption of algorithm of this paper, it designs and realizes a self-adaptive filtering method of airborne LIDAR data. This method first interpolates irregular LIDAR point cloud data into grid, and then takes smooth segmentation on grid data and builds region adjacency and height pointing matrixes. After that, it classifies segments according to geometry and topology relation between segments. At last, terrain surface is interpolated according to terrain points classified, and then terrain points are extracted from raw point cloud precisely.5. Based on fusion of high resolution airborne LIDAR data and color airborne image acquired by RCD105 camera, it does research on classification of land use. First according to the actual situation of test area, it selects a num of patterns of different land use types, and carries on statistical analyzes on them, extracting spectrum features and space distribution features. Then it designs a group of operators to extract spectrum features, features of space distribution, shape, size of land use types. According to frame theory and land use types of test area, frame system of land use types is built. A prototype system of classification of land use is realized, and land use types are recognized. At last, experiment using data of test area is carried out to test the whole flow of classification of land use are distribution map of land use types are derived.更多还原