Object-oriented Monitoring Of Land Damage And Recovery Of Ionized Rare Earth Mines

The application of rare earth elements in modern industrial production is very extensive and it is an extremely important strategic resource.Ion-absorbing type rare earths ore is the main type of rare earth in south China,As rare earth mining areas are mostly located in mountainous and hilly regions,traditional supervision methods are not time-efficient,and remote sensing technology can effectively make up for its shortcomings.In the past,visual interpretation and pixel-based classification were usually used in image interpretation.The former has high accuracy but low efficiency,the latter improves the timeliness,but its accuracy of information extraction is low.With the appearance of high spatial resolution remote sensing data,its rich spatial texture information effectively improves the information extraction accuracy.This paper selects the rare earth mining area in Ningdu county of Jiangxi province as the research area.Using 3 TM images in 2001,2005 and 2009,one GF-1 image in 2014,two GF-2 images in 2015 and 2017 as data sources.Using the Feature Extraction module of Envi,the Feature Extraction module was used to extract the ground Feature information of the mining area by object-oriented classification method.The range of the rare earth mining area was dynamically monitored and the land recovery situation was obtained.The main results are as follows.(1)In the process of data preprocessing,the fusion method of high resolution data is determined through experiments.The object-oriented classification image segmentation scale is studied on this basis.According to the characteristics of mining area,the optimal segmentation parameters and merging parameters of images from different data sources are determined by multigradient tests.Object segmentation is completed to provide the basis for object-oriented classification.(2)Selecting suitable spectral feature,texture feature and spatial feature based on actual situation of rare earth mining area.The spectral characteristics include spectral reflectance and NDVI index.Texture features include information entropy;Spatial features include area length,rectangularity and extensibility.Combined the features into a rule set,then completed object-oriented classification by K Nearest Neighbor method.(3)Using the maximum likelihood method,one of the traditional supervised classification,to get the area of rare earth mining.By comparing the result of object-oriented classification and supervised classification,the former has higher classification accuracy and better extraction.(4)According to the result of dynamic monitoring of rare earth mining area in study area,the rare earth mining area is increasing during 2001 to 2009.Based on the data from 2014 to 2017,it can be found that since the mine integration was carried out in the research area,the mining area tends to be stable.Finally,this article puts forward some reasonable suggestions for land restoration.The experimental study in this paper shows that the method of object oriented classification is effective in high resolution image extraction and dynamic monitoring.