基于QuickBird数据的输电线路径优选中地表信息提取

Decision tree classification method with QuickBird image applied in electric transmission path design

针对输电线路径优选的目标,利用QuickBird数据,基于地物类型光谱特征分析,结合遥感影像的纹理特征,采用决策树分类算法,提取影响输电线路径选择的主要地物要素.研究影响输电线选线的相关要素(如居民区、道路、水体等)及其背景地物要素(如耕地、空地等)的光谱特征和纹理特征,确立以4个波段亮度值、归一化植被指数(NDVI)和纹理对比度参数作为特征变量,建立了基于光谱和纹理组合的决策树分类模型,有效地实现居民地、道路和水体信息的提取,并将自适应滤波方法用于分类后处理,优化了分类结果.总体精度由82.09%提高到92.83%,Kappa系数由0.760 8提高到0.904 1.该精度能够满足输电线路径初选优化的要求,为提取影响输电线路径初选地物要素提供了高效快速的技术方法和基础地理数据.

With the advancement of remote sensing technology,more and higher spatial resolution data such as QuickBird have been applied in engineering projects.The distribution of residential areas,roads and water bodies should be taken into account when electric transmission paths are designed.Recently,the knowledge based on the interpretation of these images has become an effective and efficient approach to realize the automatic interpretation which can integrate spectral and other associated information such as texture.According to the objects of the electric transmission path design,the decision tree classification method is adopted to extract residential areas,roads and water bodies from QuickBird data.First,relevant variables including the grey values of four bands,NDVI（normalized difference vegetation index） and contrastively textural information are selected and extracted.Then based on the analysis of spectral and textural characteristics,the decision tree model is built.The results of classifications are optimized by using an adaptive filter to reduce speckle while preserving the edges in residential areas and roads.The overall accuracy is improved from 82.09% to 92.83% and the Kappa coefficient increases from 0.760 8 to 0.904 1.The results are satisfied with the electric transmission path design.