摘要
本文采用面向对象方法对高分辨率卫星影像道路信息提取。首先加入建筑物矢量数据对影像分割提取出建筑物,然后采用多尺度进行分割,对分割后的对象进行最近邻采样,得到总体分类图。最后根据道路特点构建道路知识库对道路信息优化。试验表明,面向对象的道路信息提取克服了"椒盐现象",取得了较好的提取效果。
This paper proposed an objected-oriented method to extract road from high-resolution satellite image. Firstly, class of building was got using chessboard segmentation by setting vector image of buildings as a thematic layer. And then it received a general classification by nearest-neighbored sampling to image objects. Those objects were got through multi-scale segmentation. Lastly, road was optimized using road-knowledge base. The experiment result showed that it could obtain better effect than traditional method.
出处
《测绘科学》
CSCD
北大核心
2011年第5期98-99,共2页
Science of Surveying and Mapping
基金
云南省自然科学基金资助(2007D042M)
关键词
道路信息
卫星影像
面向对象
信息提取
road information
satellite imagery
object-oriented
information extraction