基于地面LiDAR点云的北川县城震害建筑物提取研究

北川县城在汶川地震后变成一片废墟,本文以北川县城扫描获取的地面激光雷达数据为实验数据,在介绍地面LiDAR点云基本特征、面向对象方法思路的基础上,将面向对象方法引入到地面LiDAR点云建筑物提取中,从而实现了震害建筑物的自动提取。通过对比分析插值生成的特征的不同,构建适合于LiDAR点云震害建筑物提取的特征规则集,基于面向对象方法对北川地震遗址地面LiDAR数据进行震害建筑物识别,对结果进行分析,震害建筑物提取总体精度达92.3%,Kappa系数为0.873,提取精度满足地震遥感应急评估的需求,为震害评估增添新的可用遥感数据源。

Research on Building Extraction of Earthquake Damage of Ground LiDAR Point Cloud Based on Object-oriented

The seismic disaster information extraction method based on object-oriented method has been widely used in high-resolution optical image, while the application in the LiDAR point cloud especially the ground LiDAR point cloud data is less. Based on the introduction of the basic features of the point cloud and the method of object oriented. The laser point cloud data are decomposed and interpolated according to different features, and generated with Sdifferent band layers. Through different features analysis, Seismic damage buildings are identified based on object oriented method using ground LiDAR data of Beichuan earthquake site. By building the set of features which is suitable for the building of LiDAR point cloud,The paper extracts damage distribution. Through the analysis of the results, The overall accuracy of building earthquake damage building is 92.3%, the Kappa coefficient is 0.873,which meet the needs of earthquake remote sensing emergency assessment.