【摘要】 地震破坏的建筑物在遥感影像上和空间上表现出的特征各异,致使遥感定量化估计其破坏程度较困难。本文介绍了基于LiDAR和多光谱影像相结合的多源遥感影像进行倒塌建筑物的面向对象识别的方法、分析处理步骤和特征参数选择,并以2010年1月12日海地地震后的太子港局部LiDAR数据和高分辨率卫星影像为例,提取了倒塌和未倒塌建筑物,经与高分影像目视解译结果比较,面向对象分类结果具有较高的分类精度。更多还原

【Abstract】 Due to the different characteristics of destroyed buildings in the remote sensing image,it is difficult to quantitatively estimate the damage degree of buildings.In this paper,a seismic damage detection approach,analysis processing step and characteristic parameter selection based on an object-based classification of Light Detection and Ranging(LiDAR)and multispectral remote sensing data was introduced.This approach was then applied to classify the data acquired in some area of Portau-Prince,the capital of Haiti after Haiti earthquake on January 12,2010.Overall classification accuracies and Kappa statistics for the collapsed buildings was 90%and 0.66,respectively,were achieved.Compared with visual interpretation results from images with high resolution,object-oriented classification results higher classification accuracy.更多还原