基于纹理的面向对象分类的稀土矿开采地信息提取

Extraction of Rare Earth Mining Areas Using Objects-oriented Classification Approach Based on Texture Characteristics

原地浸矿法和池浸法/原地堆浸法(非原地浸矿法)是江西省稀土矿开采常用的方式。以江西省定南县为研究区,根据稀土矿开采方式和特点,采用面向对象分类的方法,结合纹理信息、面积大小及上下文关系等特征,对研究区2010年的ALOS影像进行稀土矿开采地的信息提取,有效区分了原地浸矿法和非原地浸矿法开采区,且两者的总体精度高达85%。并采用回溯法提取了2001年的稀土矿开采地的信息。最后根据稀土矿区分布图,对稀土矿开采现状以及2001—2010年的十年开采变化情况进行了分析。该研究可为该区域及相关矿产资源开采区的环境遥感监测提供数据支持和技术借鉴。

In-situ leaching method and non in-situ leaching method including pond leaching method and in-situ heap leaching method are common exploitation ways of rare earth ore in Jiangxi Province. By choosing one county in Jiangxi Province as a study region, an objects-oriented classification approach is applied, combining with texture information, object size and context features, to extract rare earth mining areas information from 2010＇ s ALOS fu- sion image of the study. The result shows that this method can distinguish between the in-situ leaching mining areas and non in-situ leaching mining areas effectively according to the difference of rare earth ore exploitation methods. The effectiveness of the information extraction approach was verified using confusion matrix and field survey. As a result, both verification methods showed the overall classification accuracies are as high as 85%. Furthermore, rare earth mining areas information from 2001 ＇ s SPOT fusion image was extracted by backtracking method, based on above information extraction results of the year 2010. Finally, according to the rare earth mining areas distribution maps, the current exploration status of rare earth mine and exploitation changes within ten years period, ranging from 2001 to 2010, were analyzed. This study can provide date support and technical reference for mineral resources environment remote sensing monitoring in the study area and other related areas with similar cases.