结合高分辨率遥感影像与DLG数据的面向对象道路面提取研究

Road Polygon Extraction Based on Object-oriented Method in High Spatial Resolution Remote Sense Image and DLG Data

从高分辨率遥感影像上提取道路一直是遥感影像处理的研究热点之一,但受制于有限的光谱信息和形态特征,难以进行快速、高效地自动或半自动提取。本文以高分辨率遥感影像WorldView-2作为数据源,充分利用DLG(数字线划图)的道路空间位置和属性信息,采用面向对象分类技术,探究了提取道路面的应用方法。对各种方法进行了比较分析,讨论了其优缺点,实验证明利用DLG数据作为约束条件能高效地提取出道路面。本研究提出的道路面提取方法能直接应用于地理国情监测的地表覆盖等研究。

Extracting road information based on high spatial resolution remote sense image is one of the research hotspots in remote sensing image processing.It is difficult to obtain road information quickly and efficiently because of the limitation of the spectral information and morphological characteristics.Firstly,based on object-oriented method,this paper explored the technique of extracting road polygons in high spatial resolution remote sense image WorldView-2 taking advantage of DLG data. Secondly,it performed comparative analysis for the four methods of extracting road polygons.Experimental results show that it is efficient to extract road polygons when using DLG data as spatial constraints.Those methods of extracting road polygons can be used in land cover monitoring of national geographic conditions monitoring.