面向对象的高空间分辨率遥感影像信息提取——汶川地震城市震害房屋案例研究  

Information extraction of object-oriented high resolution remote sensing image: A Case Study on urban damaged buildings in Wenchuan earthquake

作　　者：曾涛[1] 杨武年[1] 黎小东[1] 刘汉湖[1] 彭立[1] 简季[1] 

机构地区：[1]成都理工大学地质灾害防治与地质环境保护国家重点实验室/遥感与GIS研究所,四川成都610059

出　　处：《自然灾害学报》2010年第5期81-87,共7页Journal of Natural Disasters

基　　金：地质灾害防治与地质环境保护国家重点实验室"科技减灾;重建家园"开放研究基金项目(DZKJ_0806);四川省教育厅重点基础研究基金项目(07ZA007)

摘　　要：随着遥感技术的快速发展,尤其是高分辨率民用遥感卫星的成功发射和应用,遥感技术在建筑物识别、震灾调查和快速预估评估方面发挥出越来越重要的作用。然而,传统面向像元的分类方法在对高分辨率遥感影像进行分类时,存在着不能充分利用影像信息、分类精度降低、速度慢等局限性,根据高分辨率遥感影像的特点,提出了面向对象的高空间分辨率遥感影像建筑物震害信息提取方法。该方法首先通过影像分割将影像划分为互不相交的影像对象,然后根据这些影像对象的影像特征如光谱、纹理、形状和上下文等信息进行分类,提取出破坏建筑物和未破坏建筑物,试验表明,面向对象的分类方法应用于建筑物震害信息提取较传统分类方法有更高的精度,具有很大的应用潜力。With the rapid development of remote sensing technology,especially more and more launches and applications of high resolution satellite,the technology plays an important role in identification of buildings,investigation and pre-evaluation of earthuake disaster.However,the traditional pixel-oriented approaches show a lot of limitations on classification of high resolution satellite images,and they can not take full advantages of image information.According to the characteristics of high-resolution remote sensing images,this paper presents a new object-oriented scheme of damaged building information extraction from high-resolution remote sensing image.The scheme has two steps.The first step is segmenting the whole imagery into image objects which do not intersect mutually.The second step is extracting damaged buildings and undamaged ones with the features used to classify,like spectrum,texture,shape and context.The experimental results indicate that the new object-oriented information extraction technique has high accuracy compared to the traditional classification methods and has a great application potential.

关 键 词：建筑物震害 高分辨率影像 面向对象 影像分割 信息提取 

分 类 号：P315.9[天文地球—地震学]