基于面向对象层次分类法的输电线路走廊地表覆盖分类

Land-cover Classification of Electric Transmission Lines of Hierarchical Classification Based on Object-oriented

随着国产卫星遥感影像在电力部门的应用日益深入,利用高分辨率国产卫星遥感影像对输电线路走廊地区进行快速准确的地表覆盖分类成为现实。本文以湖北省输电线路走廊为研究对象,通过多种地物分类方法对比试验,综合利用高分辨率影像的光谱、纹理、形状等特征优势,以"高分一号"影像为例探索出面向对象的层次分类法,快速实现了地物的精确提取,可用于输电线路走廊地区山火评估及预防。研究结果表示,面向对象的层次分类法对于输电线路走廊地物快速分类是可行的并且能达到较高的精度。

With the wide application of domestic satellite remote sensing in the electricity sector, it has become a reality that high - resolution domestic satellite remote sensing images are used to classify the land cover in a fast and accurate way. In this work, the elec- tric transmission line in Hubei Province has been taken as the research target. The exploration of high - resolution No. 1 images gives an object - oriented method for hierarchical classification. A fast extraction of terrain features has been achieved through a comparative experiment of various terrain classification methods and a comprehensive utilization of characteristic advantages in high - resolution im- ages, such as spectra, texture and shape. As a result, the occurrence of mountain fires can be assessed and prevented. These results show that the object - oriented hierarchical classification method is feasible and accurate for the fast classification of terrain features a- long the transmission line corridors.