面向对象的高分辨率影像采煤塌陷地提取

Coal mine subsidence area extraction from high-resolution remote sensing imagery based on object-oriented classification method

煤炭开采引起了塌陷等一系列的地质环境问题,与常规监测方法相比,遥感技术可以实现大范围、高效率、周期性的动态监测。在遥感影像分类方法中,面向对象的遥感影像分类方法能更好地利用高分辨率遥感影像中丰富的纹理和几何结构信息。针对煤炭开采导致的地表塌陷地的特点,在归纳整理遥感影像中塌陷地判识准则的基础上,重点探讨了面向对象的遥感影像分类方法中塌陷地的自动提取规则。综合利用ERDAS IMAGINE9.2、ENVI4.7和ENVI4.4 ZOOM进行数据处理,以安徽省淮南矿务集团潘三矿区为实验区,用该方法利用SPOT5影像进行了塌陷地信息提取实验,结果证明,面向对象分类方法能有效地从高分辨率遥感影像中自动提取塌陷地相关信息。

Coal mine activities have caused a series of geological and environmental problems,such as ground surface subsidence,et al.Compared with conventional monitoring techniques as leveling and GPS,the remote sensing technique can dynamically monitor the mine subsidence with wider coverage and higher efficiency.The object-oriented classification method,which can make better use of the texture and geometry shape information of the remote sensing imagery,is more and more utilized in high resolution image classification.Based on the characteristics of mine subsidence area,how to recognize the mine subsidence area from remote sensing imagery is studied,and the automatic extraction rules for object-oriented classification method are proposed.The Pansan mine of Huainan Mining Group in Anhui province,China,is selected as an experimental area,the SPOT 5 images are used to extract mine subsidence area with the proposed method,all of the images are processed using ERDAS IMAGINE9.2,ENVI4.7 and ENVI4.4 ZOOM.2.322 6 km2 mine subsidence area are extracted,compared with field survey data,the result is very convincing.