Dynamic Monitoring Of Rare Earth Mining Based On Object-oriented Classification Method

Ion-adsorbed rare earth ore is not only a very important strategic resource in the global,but also a valuable mineral resource to our country.In recent years,China’s rare earth reserves continue to decrease,How to reflect the development activities of rare earth mine in real time accurately has become the prerequisite for the sustainable development of China’s ion adsorption rare earth resources.Remote sensing technology as a rapid and dynamic to provide multi-temporal,large-scale real-time information monitoring means to make up for the lack of traditional regulatory means,Especially with rich spatial details of the high-resolution remote sensing satellite data,but also for the rare earth mine monitoring provides a great convenience.However,the use of traditional visual interpretation and pixel classification based on the high-resolution image data for rare earth mine information extraction,or cost manpower and time,or classification accuracy is a bit worse,they can not achieve the desired results.This article is based on object-oriented classification method,using of ENVI Feature Extraction module and five high-resolution remote sensing data from 2005 to2015,extracted the rare earth mining area in Xunwu of Jiangxi Province and carried out the dynamic monitoring of rare earth mining based on the extraction results,The main results are as follows.(1)This article first completed the multi-source multi-temporal remote sensing data preprocessing work and then studied the best image segmentation scale of object-oriented classification.Combined with the characteristics of rare earth mining area,through the comparative analysis,the optimal segmentation parameters and the combined parameters of each remote sensing image are determined,and the multi-scale segmentation of the image is realized,which provides the data base for object-oriented classification.(2)Combined with the actual situation,this article select the terrain,spectrum and geometric characteristics to establish a set of rules that conform the characteristics of the rare earth mining area to achieve object-oriented classification.Among them,the terrain features include elevation,slope,surface cutting depth and terrain relief,spectral characteristics include NDVI index and spectral reflectance,geometric features include area,rect_fit and elongation.Finally,combined with the rule set,the membership function method is used to realize object-oriented classification.(3)This article used the traditional supervised classification maximum likelihood classification to the rare earth mining area of the study area,compared the results of object-oriented classification and traditional supervised classification and evaluate accuracy,The result shows that the object-oriented classification is more accurate and the effect is better.(4)Using the first classification after the comparison method to carry out dynamic monitoring of rare earth mining,The result shows that in the period of2005-2015,the mining area of Xunwu rare earth shows a tendency to first decrease after increase and then tend to stabilize.Finally,this article puts forward some reasonable suggestions for the environmental protection and management of the area to provides data support and technical support for the mining management department.The results show that the use of object-oriented classification technology,combined with high-resolution image extraction of rare earth mining area and dynamic monitoring is completely feasible,and the results are ideal.