基于RF的面向对象方法的建筑物提取研究

建筑物作为城市地理空间的主要组成部分,建筑物信息的提取研究对城市规划、智慧城市等方面具有重要意义。基于高空间分辨率遥感影像的建筑物信息提取,分类方法的选择是影响分类精度的关键因素。文章采用目前分类精度较高,且相对成熟的算法之一——随机森林(Random Forest,RF)结合面向对象的影像分析方法实现德国恩茨河畔法伊欣根某部分城区的建筑物信息提取。结果表明:基于RF的面向对象方法的建筑物提取效果较好,总体精度和Kappa系数分别达到0.96,0.84。

Buildings are the main components of the urban geographic space,and the research on the extraction of buildings information is of great significance to urban planning,intelligent city and so on.In the extraction of building information based on high spatial resolution remote sensing images,the selection of classification methods is the key factor affecting the classification accuracy.In this paper,Random Forest(RF),one of the relatively mature algorithms with high classification accuracy,combined with object-oriented image analysis method,is used to extract building information from a part of Vaihingen an der Enz,Baden-Wurttemberg,Germany.The results show that the object-oriented method based on RF has a good effect on building extraction,and the overall accuracy and Kappa coefficient are 0.96 and 0.84 respectively.