面向基元的高空间分辨率矿区遥感影像土地利用分类  

Element-Oriented Land-Use Classification of Mining Area by High Spatial Resolution Remote Sensing Image

作　　者：陈启浩[1,2] 刘志敏[1] 刘修国[1,2] 罗红霞[1] 高伟[1,2] 王红平[1,2] 

机构地区：[1]中国地质大学信息工程学院,湖北武汉430074 [2]地理信息系统软件及其应用教育部工程研究中心,湖北武汉430074

出　　处：《地球科学（中国地质大学学报）》2010年第3期453-458,共6页Earth Science-Journal of China University of Geosciences

基　　金：教育部“新世纪优秀人才支持计划”项目(No.NCET-07-0772);国家重点“863”项目(No.2007AA120503)

摘　　要：为了合理开发矿产资源和有效监测矿区生态环境,采用面向基元的分类方法,对广西横县某矿区的高分辨率航空遥感影像进行了土地利用分类.通过优化分形网络演化多尺度分割方法,高效提取了矿区两个尺度上的影像基元层;基于基元信息,详细分析了各地表地物光谱特征、空间特征以及类相关特征,建立了研究区土地利用的分类知识库;采用决策支持的模糊逻辑推理法进行分类,使分类的精度从53%提高到了90%.表明面向基元的方法能较好地利于高空间分辨率矿区影像的各种特征进行高精度的土地利用分类.For the purpose of rational exploitation of mineral resources and effective monitor of ecological environment in mining areas,we did experiments about land-use classification of the high spatial resolution airborne image from a mining area in Heng County of Guangxi province using element-oriented method.By optimizing the evolution of multi-scale fractal network segment process,two levels of image elements were extracted efficiently.Based on the multi-scale image elements,the land-use classification knowledge base of the study area was established through analyzing spectral,spatial and class-relation features in this area.The classification precision improved from 53% to 90% by decision supporting fuzzy logic reasoning of the knowledge base.The experiments show that the element-oriented method can obtain high precision land-use classification for taking full advantage of various features of the mining area from the high spatial resolution image.

关 键 词：高空间分辨率 面向基元 矿区土地分类 分形网络演化 知识库 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]