无证矿山的遥感监测

发布时间：2018-06-10 21:07

本文选题：遥感 + 面向对象　；参考：《中国地质大学（北京）》2013年硕士论文

【摘要】：矿产资源是人类宝贵的不可再生资源，所以矿产资源的可持续发展是我国的既定方针。故提取违法开采图斑并取缔违法矿山势在必行。利用遥感影像并采用一套自动化的目标图斑提取手段对无证矿山进行提取具有很大的意义。鉴于目前对于无证矿山的提取大部分采用的仍是纯目视人工提取的手段，所以本文为寻求无证矿山图斑的自动化提取而开展了研究。论文中以九江北部的一小块地区作为实验区，采用SPOT-5影像，首先对该影像进行必要的影像预处理。其次，选取各地物的特征建立条件函数和隶属度函数，对处理后影像进行面向对象的分类，得到各类地物的图斑。再利用图像差值的方法对两期影像进行变化检测，检测出地类的变化图斑。之后，将上述两层图斑和矿权数据进行叠加，通过GIS平台的分析得到疑似无证矿山的图斑。最后对这些图斑用目视判读的方式来筛选并以野外加以验证。在面向对象的分类技术中，首先对实验区采用了面积比均值法，得到比较合适的分割尺度，确定合适的分类层次。然后对各类地物进行采样，选取它们的光谱上和纹理上的特征属性以确定它们的特征条件和隶属度函数，进而分类得到各类地物的图斑，最后对分类结果做了一个简单的精度评价。在变化检测技术中，，经分析后采用图像差值的方法，对各地类在两期影像上的各波段灰度进行计算，得到合适的差值波段，进行差值运算生成二值图，对二值图的散点进行简单分类，得到变化图斑。从本文的研究结果来看，这种面向对象分类结合变化检测的方式对于目标图斑的提取具有一定的可行性。并且对于自动化提取目标图斑的研究产生了深远的意义。
[Abstract]:Mineral resources are valuable non-renewable resources, so the sustainable development of mineral resources is the established policy of our country. Therefore, it is imperative to extract illegal mining map spots and ban illegal mines. It is of great significance to use remote sensing images and a set of automatic target spot extraction methods to extract unlicensed mines. In view of the fact that most of the current extraction methods for unlicensed mines are still manual and purely visual, In this paper, a small area in the north of Jiujiang is used as the experimental area, SPOT-5 image is adopted, and the necessary image preprocessing is carried out. Secondly, the conditional function and the membership function are established by selecting the characteristics of the objects in different places, and the object oriented classification of the processed images is carried out, and the image spots of all kinds of ground objects are obtained. Then the image difference was used to detect the changes of the two periods of images, and the change patterns of the ground types were detected. After that, the above two layers map spot and mining right data are superposed, and the suspected unlicensed mine map spot is obtained by the analysis of GIS platform. Finally, these patterns are screened by visual interpretation and verified in the field. In the object-oriented classification technology, the area ratio mean method is first used in the experimental area, and the appropriate segmentation scale is obtained and the appropriate classification level is determined. Then we sample all kinds of ground objects, select their spectral and texture attributes to determine their characteristic conditions and membership functions, and then classify them to obtain the image spots of all kinds of ground objects. Finally, a simple accuracy evaluation of the classification results is made. In the change detection technology, the gray level of each band of each region is calculated by using the method of image difference after analysis, and the appropriate difference band is obtained, and the binary map is generated by difference operation. From the research results of this paper, it is shown that this method of object oriented classification combined with change detection is feasible for the extraction of target spot. And it has a profound significance for automatic extraction of target spot.
【学位授予单位】：中国地质大学（北京）
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P237

【参考文献】