基于震后高分辨率卫星遥感影像的建筑物瓦砾快速提取方法

A Rapid Method for Extracting Building Rubble Based on Post-earthquake High-resolution Satellite Imagery

利用震后高分辨率卫星遥感影像提取建筑物损毁空间分布和破坏程度信息,对于地震灾情评估具有重要作用.本文以2010年海地地震巨灾为例,选用震后高分辨率卫星遥感影像Geoeye-1为数据源,在分析建筑物瓦砾可分离性的基础之上,利用监督分类方法提取损毁建筑物的瓦砾.结果表明,在震后高分辨率卫星遥感影像中,瓦砾是建筑物损毁的明显震害标志;瓦砾的生产者精度为87.23%,大于总体分类精度63.14%;瓦砾的Kappa系数为0.62,高于总体Kappa系数0.54.研究表明,基于瓦砾纹理特征的遥感信息提取方法能够从震后复杂的城市地物类型中识别出大部分瓦砾,该方法得到的结果可以应用于地震灾情应急评估,辅助应急救援等.

Using post-earthquake high-resolution satellite imagery to detect the spatial distribution and extent of building damage can play an important role in the assessment of earthquake disaster loss. The paper takes Haiti Earthquake catastrophe in 2010 as an example,and uses Geoeye-1 very high resolution satellite imagery to extract the rubble of damaged building based on supervised classification method,after analyzing the detachability of building rubble. The results show the rubble is a clear sign of building damage in post-earthquake high-resolution satellite imagery. In the precision evaluation,the producer accuracy and Kappa coefficient of rubble are 87. 23% and 0. 62 respectively,higher than the total classification accuracy（ 63. 14%）and the whole Kappa coefficient（ 0. 54）. Most of rubble could be extracted from post-earthquake complex urban building type based upon rubble texture. The method used in the paper could be applied in earthquake disaster emergency assessment,and could assist in emergency rescue as well.