Research On Information Automatically Extract Method Of Surface Coal Mining In Xilinhot

In Inner Mongolia, mineral resources are rich. There are a lot of open-pit mining coals, which covers are not only large, but also the destructions of land resources and ecological resources are serious. Therefore, this thesis based on "Inner Mongolia mining exploration remote sensing survey and monitoring" project, select Shengli opencast coal mine in Xilinhot for the study area, develop the research on the information automatically extract method of mining area.With the spectrum、texture features and spatial characteristics of remote sensing data, test the object-oriented classifications of study area’s high-resolution images for performance. According to the test result, a set of classification combination with weighted voting method based on hierarchy technology is built. Then it is applied to the information extract of large-scale opencast coal mine in Xilinhot, to verify the effectiveness of this technology. The paper made the following findings:(1)Analysis the large information content of high-resolution remote sensing data deeply, then put the spectral information、texture information of experimental area’s worldview-2 data and spatial information of DEM data together. That provides data security for future study.(2)Classify the composite data by six kinds of object-oriented supervised classification algorithm, and test the resulting classifiers’ performance. After that, select out high-performance classifiers as the base classifiers, laying the foundation for hierarchical information extraction.(3)According to the base classifiers’ microscopic precision analysis results, make sure every class’ s optimal classifier. Using hierarchical means achieve the information of superior classes, that ensure the accuracy of the superior classes. At the same time, it avoids information interference between different classes. As to the weak classes have no optimal classifier, first measure the sub-classifiers diversity, then by the means of classification combination with weighted voting access to the extraction satisfactory.(4) Based on the above studies, use a large opencast coal mine(127.5km2) in Xilinhot to verify the effectiveness of this paper’s technology, the overall accuracy of information extraction reaches to 92.45 percent.