最优尺度下的面向对象建筑物提取  

Object-oriented Building Extraction Based on Optimal Scale

作　　者：李慧 李志娟 张东华 LI Hui;LI Zhijuan;ZHANG Donghua(Administration of Surveying,Mapping and Geoinformation of Inner Mongolia,Hohhot 010010,China;Department of Land Science and Technology,China University of Geosciences(Beijing),Beijing100083,China;College of Water Conservancy and Civil Engineering,Inner Mongolia Agricultural University,Hohhot 010010,China)

机构地区：[1]内蒙古自治区测绘地理信息中心,呼和浩特010010 [2]中国地质大学(北京)土地科学技术学院,北京100083 [3]内蒙古农业大学水利与土木建筑工程学院,呼和浩特010010

出　　处：《遥感信息》2022年第3期72-76,共5页Remote Sensing Information

摘　　要：针对基于面向对象方法提取建筑物准确度与精度不高的问题,提出了一种面积方差最优尺度分割法。该方法利用SPOT 5卫星遥感影像,根据建筑物特征对影像进行多尺度分割,以所有类别分割对象与实际目标匹配程度作为衡量的依据获得最优尺度。在不同尺度等级的对象层次上,根据对象的多特征组合,完成对建筑物分层分类提取。实验结果表明,该方法大大减少了分类的不确定性,获取了较高的精度,为建筑物准确提取提供了思路。Aiming at the problem of low accuracy of building extraction based on object-oriented method,an area variance optimal scale segmentation method is proposed.Using SPOT 5 satellite remote sensing image,this method segments the image on multiple scales according to the building characteristics,takes the matching degree between all categories of segmented objects and the actual target as the measurement basis,and obtains the optimal scale by constructing this index.As a basis for measurement,the optimal scale is obtained by constructing this indicator.At the object level of different scale levels,the hierarchical classification and extraction of buildings is completed according to the combination of multi-special features of the object.The experimental results show that the proposed method can greatly reduce the uncertainty of classification,and obtain higher accuracy,which provides an idea for accurate building extraction.

关 键 词：面向对象 影像分割 最优尺度 建筑物提取 分类 

分 类 号：P237[天文地球—摄影测量与遥感]