基于遥感影像的城镇建筑物群分类  

Classification of urban buildings based on remote sensing images

作　　者：刘贾贾 刘志辉[1] 李凤[1] LIU Jiajia;LIU Zhihui;LI Feng(Earthquake Administration of Hebei Province,Shijiazhuang 050800,China)

机构地区：[1]河北省地震局,河北石家庄050800

出　　处：《自然灾害学报》2021年第6期61-66,共6页Journal of Natural Disasters

基　　金：河北省地震局地震科技星火计划项目重点项目(DZ20160621075);河北省地震灾害损失评估项目。

摘　　要：在地震灾害事件中,不同结构类型建筑的破坏程度不同,造成的人员和财产损失不同,掌握不同结构类型建筑物的数量有利于做好地震风险管理。本文以怀来县城为例,采用监督分类中决策树(CART)分类法、随机森林(Random Trees)分类法对建筑物群进行了分类,并与实际调查及目视解译结合的结果进行了对比分析,得出监督分类的两种分类方法均可用于城镇建筑物群的分类。In the event of an earthquake disaster,buildings of different structural types have different degrees of damage,resulting in different losses of personnel and property.The number of buildings of different structures is conducive to earthquake risk management.Taking Huailai County as an example,this paper uses the decision tree(CART)classification method and the random forest(Random Trees)classification method in the supervision classification to classify the building group,and combines the results of the actual investigation and visual interpretation.Comparative analysis shows that the two classification methods of supervised classification can be used for the classification of urban building groups.

关 键 词：遥感影像 城镇建筑物群 分类方法 精度评价 

分 类 号：P237[天文地球—摄影测量与遥感] P315.9[天文地球—测绘科学与技术]