【摘要】 建筑物倒塌是地震灾害中最为严重的一种破坏形式,直接造成震区生命财产的损失.震区建筑物的破坏监测对灾情速报、灾害等级划分、灾害评估、应急抢险、次生灾害监测和灾后重建规划等具有重要的指导意义.传统的面向像元的分类方法不能够很好地提取出建筑物信息,而面向对象的分类能够融合除像元光谱信息以外的形状、纹理和位置关系等多种信息,分类更加客观准确.利用玉树Ms7.1地震前后航空航天多源遥感数据,采用面向对象的分类方法,提取建筑物的覆盖面积,然后统计出建筑物的倒塌率为58.6%,与影像目视解译倒塌率60.4%的结果吻合较好.更多还原

【Abstract】 The building collapse is the most severe earthquake damage;Earthquake directly caused the loss of life and property,and it is important to quick report of the disaster monitoring,classification,assessment,emergency rescue,secondary disaster monitoring and reconstruction planning etc..Conventional pixel-oriented classification method is not able to extract the building information exactly,but object-oriented classification can integrate the shape,texture and position information besides spectral information,which is more objective and accurate.In this paper,Yushu Ms7.1 non-homologous aerospace remote sensing data before and after the earthquake is used to extract the building coverage by using object-oriented classification method.The building collapsed rate was expected 58.6% and close to 60.4% obtained from visual interpretation.更多还原