利用高分辨率DSM和DOM自动提取城市建筑物区域

Automatic Extraction of Urban Building Area Using High-resolution DSM and DOM

建筑物区域的精确提取是数字地形图自动化绘制的基础。本文通过融合无人机航摄处理得到的高分辨率DSM(digital surface model)与DOM(digital orthophoto map)数据,提出一种基于阈值分割、SVM(support vector machines)分类和形态学处理的建筑物区域提取算法,并选取ISPRS提供的Potsdam数据集进行试验,通过定量化精度分析验证了该算法的可行性及可靠性。

The accurate extraction of building area provides the base for automatic drawing of digital topographic map.This paperproposes an algorithm of building area extraction,which is based on threshold segmentation,SVM classification and morphology processing by combining high-resolution DSM and DOM data obtained by UAV aerial photography.The Potsdam data set provided by ISPRS is selected for experiment,and the feasibility and reliability of the proposed algorithm are verified by quantitative accuracy analysis.