基于机载Lidar的建筑物震害空间特征提取研究

Airborne LiDAR will be an important data source after the earthquake catastrophe in the future because of its technological superiority. The successful applications of remote sensing (RS) technology in many home and abroad earthquake emergency works show that disasters can be quantitatively assessed using high-resolution airborne LiDAR data. Owing to the diversity of damage display, identification of damaged building is more complex and different than the regular building. The existing damage extraction methods mostly use the digital image processing methods which do not take full advantage of the high accuracy spatial information and not quantify the damage degree of buildings. Therefore, we will make use of the two main characteristics of buildings in the post-earthquake airborne LiDAR data , which are the similarity of building vertical section (side elevation) and the abrupt change of building height caused by earthquake damage, and analyze the three-dimensional seismic damage characteristics of different structure types of buildings in each damage degree. The spatial geometric description features in the building side elevation representing the distinctive difference of damage degree by the procession of post-earthquake real LiDAR data are determined. The construction methods of standard intact building which are best matched with the detected buildings and the similarity measure algorithm of building side elevation are researched. In addition, the applicability of the technology method and the stability of algorithm in ordinary building are detected and discussed in quantitative recognition of damage degrees in complex structure buildings. It is expected that the research results will promote the development of quantitative damage identification technology based LiDAR, which have important implications for post-earthquake building damage assessment.

机载激光雷达(LiDAR)的技术优势使其成为未来特大破坏性地震后的重要数据源，国内外多次地震应急遥感应用实践证明，高精度LiDAR能够实现灾害定量评估。损坏建筑物特征表现的多样性使其震害识别比规则建筑物提取复杂、难度大，现有震害提取方法以数字图像处理方法为主，未充分利用LiDAR的高精度空间信息，并且未对建筑物破坏程度进行量化。据此，本研究将使用实际地震后机载LiDAR数据，利用建筑物垂直剖面（侧立面）相似和震害导致房屋高度突变两个主要特性，分析不同类型不同破坏程度建筑物的三维震害特征，试验确定基于建筑物侧立面的震害可分性空间几何描述特征，开展与检测房屋最佳匹配的未破坏标准房屋构建方法和建筑物侧立面相似度量算法研究，在此基础上，探讨上述技术方法在复杂建筑物震害程度定量识别中的适用性。项目预期成果将促进基于LiDAR数据的建筑物震害定量识别技术的发展，提高地震灾害评估的准确性。

机载激光雷达(LiDAR)的技术优势使其成为未来特大破坏性地震后的重要数据源，国内外多次地震应急遥感应用实践证明，高精度LiDAR能够实现灾害定量评估。损坏建筑物表现的多样性使其震害识别比规则建筑物提取复杂、难度大，现有震害提取方法以数字图像处理方法为主，未充分利用LiDAR的高精度空间信息，并且未对建筑物破坏程度进行量化。本项目收集整理了2010年海地7.0级地震后获取的机载LiDAR数据资料，在震害LiDAR点云滤波处理基础上，研究了融合遥感影像的建筑物点云提取方法，通过选取不同破坏类型和屋顶形态建筑物典型破坏样本，分析了不同类型不同破坏程度建筑物的遥感影像和三维点云分布特征。利用建筑物垂直剖面（侧立面）相似和震害导致房屋高度异常两个主要特性，研究了建筑物点云坐标自动归一化方法，建筑物侧立面点云连接线Douglas-Peuker简化算法，研究了考虑房屋整体和局部破坏变化的侧立面相似度度量方法，建立了以建筑物侧立面点云简化后图形拐点数量和紧凑度作为评价各个侧立剖面形状相似度的匹配模型、以质心距离作为位置相似度匹配模型、以面积、周长作为大小相似度匹配模型，从整体上给出建筑物损坏程度和局部结构破坏情况量化分析。此外，建立了基于局部表面拟合的法向量、天顶方向夹角及其二次震害表征参数，发展了邻近点坡度、体积、高度均值偏离度、回波次数比等震害特征及其二次特征参数，通过典型样本的震害敏感性试验研究，确立了有效的建筑物震害检测空间特征。研究成果将促进基于LiDAR数据的建筑物震害定量识别技术的发展，提高地震灾害评估的准确性。