基于车载LIDAR数据的建筑灾情应急测量研究

发布时间：2018-03-19 05:31

本文选题：车载LiDAR　切入点：建筑物灾情　出处：《河南理工大学》2014年硕士论文　论文类型：学位论文

【摘要】：重大自然灾害发生后快速获取建筑物灾情信息是灾害应急中最迫切的任务之一。灾害发生时伴随的恶劣气候因素严重制约了灾情信息的快速获取,因此灾害刚发生后的一段时间是建筑物灾损信息最匮乏的时期。车载LiDAR系统在获取建筑灾害信息方面具有明显优势,其能够在灾害发生后快速抵达灾害现场并迅速投入工作。本研究主要面向重大突发自然灾害的应急需求,以车载LiDAR系统获取的灾害环境的数据为信息源,重点研究了从原始点云数据中快速提取建筑物灾损信息的方法。论文选取若干建筑物和依照比例缩小后的建筑物模型作为研究对象,通过模拟试验、对比分析、数据处理等方法进行了车载LiDAR用于建筑灾情应急测量方法的探讨,研究取得了一定的进展。本文所做的工作主要如下:(1)基础地理数据辅助建筑物立面分割方法。首先将基础地理数据转换到车载LiDAR点云数据所在的坐标系下,实现坐标统一;然后从基础地理数据中提取有效的建筑物轮廓信息;最后通过设置合理的建筑物轮廓缓冲区阈值,对建筑立面点云进行分割。(2)利用投影点密度结合高程法进行精细滤波。根据点云数据投影的范围进行格网划分,设定格网单元内激光点的数量阈值以及格网单元内高程平均值阈值,将满足条件的格网进行保留,提取出建筑物点云数据。(3)进行建筑物灾害模拟试验。根据建筑物灾害后的表现形态,构建了不同灾损建筑物立体模型。制作了依比例缩小后精确的建筑物模型,模拟建筑物的各种灾损形态,采集了灾损建筑物点云数据。(4)采用随机抽样一致性算法(RANSAC)对点云数据进行立面分割,设计的判断准则主要根据点云平面方程和r半径密度;然后运用凸壳算法实现了建筑物立面特征线的提取。(5)根据提取出的建筑物立面灾损特征要素,总结出不同灾损形式下建筑物的形态特征,实现对灾损类型以及程度的判断,为重大自然灾害应急提供了依据。
[Abstract]:The rapid acquisition of building disaster information after major natural disasters is the most urgent Ren Wuzhi in disaster emergency. The severe weather factors associated with disasters seriously restrict the rapid acquisition of disaster information. Therefore, the period after the disaster occurred is the most scarce period of building damage information. The vehicle-borne LiDAR system has obvious advantages in obtaining building disaster information. It can reach the disaster site and put into work quickly after the disaster occurs. This research mainly faces the emergency demand of the major sudden natural disaster, and takes the disaster environment data obtained by the on-board LiDAR system as the information source. This paper focuses on the method of quickly extracting building damage information from the original point cloud data. In this paper, several buildings and the scaled down building models are selected as the research objects, and the simulation experiments are carried out, and the results are compared and analyzed. The data processing method is discussed in this paper, which is applied to the emergency measurement of building disaster with LiDAR. Some progress has been made. The main work of this paper is as follows: 1) the basic geographic data is used to assist the building elevation segmentation. Firstly, the basic geographic data is converted to the coordinate system in which the vehicle LiDAR point cloud data are located, and the coordinate unification is realized. Then the effective building contour information is extracted from the basic geographic data. Finally, by setting a reasonable threshold of the building contour buffer, Using projection point density combined with elevation method to carry out fine filtering. According to the range of point cloud data projection, the grid is divided. The threshold of the number of laser points and the threshold of the average height in the grid element are set to preserve the grid that satisfies the conditions. The building disaster simulation test was carried out by extracting the data of the point cloud of the building. According to the appearance of the building disaster, the three-dimensional models of different damage buildings were constructed, and the accurate building model was made after the scaling down. In order to simulate the damage patterns of buildings, the point cloud data of damaged buildings are collected. (4) A random sampling consistency algorithm (RANSACC) is used to segment the elevation of point cloud data. The criteria are mainly based on the point cloud plane equation and the radius density of r. Then the convex hull algorithm is used to realize the feature line extraction of the building facade. According to the extracted feature elements of the building facade, the shape features of the building under different disaster loss forms are summed up, and the types and the degree of the damage are judged. It provides a basis for emergency response to major natural disasters.
【学位授予单位】：河南理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TU198;TP391.41

【相似文献】