震后机载LiDAR点云建筑物震害提取因子研究

发布时间：2018-03-13 09:16

本文选题：机载激光雷达　切入点：遥感　出处：《中国地震局地震预测研究所》2016年硕士论文　论文类型：学位论文

【摘要】：地震灾害作为危害人类最严重的自然灾害之一,震后灾情信息的及时获取对应急救援、评估具有非常大的现实意义,遥感技术的快速、大范围、全天时等优势在多次抗震救灾工作中得到了充分应用并发挥着越来越重要的作用。地震造成的建筑物倒塌是造成人员伤亡和经济损失的重要原因之一。机载激光雷达(Light Detection and Ranging,Li DAR)受天气影响较小,具有穿透性、作业速度快、测量精度高、数据产品丰富等的优势。机载Li DAR系统获取的高精度三维点云数据为建筑物震害研究提供了重要的数据源。机载Li DAR点云数据建筑物震害信息提取根据使用的数据源的不同,分为基于震后单时相Li DAR点云、多时相Li DAR点云、Li DAR点云与其他数据结合的方法。本文首次提出震害提取因子概念,根据建筑物震害点云特征差异建立了机载Li DAR建筑物震害三维空间特征的二次特征参数并对及进行了震害定量提取研究,利用海地地震实例数据试验确定有效震害提取因子。主要内容包括有:1.针对地震后机载Li DAR数据特点,初步建立了震后机载Li DAR点云数据预处理方案,在去噪和滤波基础上,利用点云的平面密度参数在尽可能保留地震破坏地物信息的基础上去除地面点,获取建筑物点云。2.在对未倒塌与倒塌建筑物点云特征进行差异分析的基础上,通过计算未倒塌与倒塌建筑物点云试验样本的强度、高度、法向量、体积、坡度等值,统计分析未倒塌与倒塌建筑物这几个数值的变化情况,建立了强度标准差值的变化(SSDI)、高度标准差值的变化(SSDH)、法向量与天顶方向夹角标准差值的变化(SSDA)、体积标准差值的变化(SSDV)、坡度标准差值的变化(SSDS)因子,对各个因子进行定量分析,确定了SSDI、SSDH、SSDA、SSDV与SSDS因子用于震害检测的因子阈值。3.建立的震害提取SSDI、SSDH、SSDA、SSDV与SSDS因子根据前述章节中经试验选择合适因子利用人工确定因子阈值方式和支持向量机方式对试验区建筑物震害点云进行提取试验。试验结果证明本文提出的SSDH、SSDA,SSDV与SSDS因子可以作为检测震害点云的有效因子进行震害提取,不同因子的组合试验效果优于单因子识别效果,利用支持向量机方法可实现建筑物震害的快速检测,建筑物震害检测总体精度达到90%。
[Abstract]:Earthquake disaster is one of the most serious natural disasters that endanger mankind. The timely acquisition of disaster information after earthquake is of great practical significance to emergency rescue and evaluation, and the rapid and wide range of remote sensing technology. The all-day advantage has been fully applied in many earthquake relief work and is playing an increasingly important role. The collapse of buildings caused by earthquake is one of the important causes of casualties and economic losses. Airborne laser is one of the important causes of casualties and economic losses. The radar light Detection and angling Li DAR is less affected by the weather. It has the advantages of high penetration, high working speed and high measuring precision. The high accuracy 3D point cloud data obtained by airborne Li DAR system provide an important data source for building damage research. Based on the method of combining single phase Li DAR point cloud and multitemporal Li DAR point cloud with other data, the concept of seismic disaster extraction factor is proposed for the first time in this paper. Based on the difference of damage point cloud characteristics of buildings, the secondary characteristic parameters of 3D spatial characteristics of airborne Li DAR buildings are established, and the quantitative extraction of earthquake damage is carried out. The effective seismic disaster extraction factor is determined by using Haiti earthquake case data test. The main contents include: 1. According to the characteristics of airborne Li DAR data after the earthquake, the preprocessing scheme of airborne Li DAR point cloud data after earthquake is preliminarily established, which is based on de-noising and filtering. The plane density parameters of point clouds are used to remove the ground points on the basis of preserving the information of earthquake damage objects as much as possible, and to obtain the point clouds of buildings. 2. On the basis of the difference analysis of the characteristics of point clouds of uncollapsed and collapsed buildings, By calculating the strength, height, normal vector, volume, slope equivalent of the point cloud test samples of the uncollapsed and collapsed buildings, the variation of the values of the uncollapsed and collapsed buildings is analyzed statistically. In this paper, the variation of strength standard difference and height standard difference are established. The variation of normal vector and zenith angle standard difference is discussed. The change of volume standard difference and slope standard difference are analyzed quantitatively. The threshold value of SSDI SSDV and SSDS factors for seismic damage detection was determined. 3. SSDI SSDH SSDH SSDV and SSDS factors were selected according to the experiments in the preceding chapters to determine the factor threshold and support vector machine (SVM). The test results show that the SSDH SSDV and SSDS factors proposed in this paper can be used as effective factors to detect seismic damage point clouds. The effect of combination test of different factors is better than that of single factor recognition. By using support vector machine method, the rapid detection of earthquake damage of buildings can be realized, and the overall accuracy of seismic damage detection of buildings can reach 90%.
【学位授予单位】：中国地震局地震预测研究所
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：P315.9

【相似文献】