| Mineral resources, important parts of natural resources, are the basic guarantee for sustainable economic and social development of our country. The rectification of mineral resources development situation survey database(MRDSSD) and mineral reserves database(MRD) is a basic work carried out by the Ministry of Land and Resources for strengthening the management of mineral resources. But due to many factors, such as technology, humanity, history and so on, many mines in the MRDSSD have incomplete spatial reference information or even completely lost their location data. This leads directly to the great location deviation of mines after transited to the national coordinate system. Until now, many studies on Mine Localization have been done. Locating mines by using 3S technology and image matching technology can not only save money and manpower costs, but also greatly improve efficiency, and is of great practical significance for the mastering of the amount and location of the national mineral resources.In the existing Mine Localization methods, coordinate transformation method requires that there is enough spatial reference information for the mine to be located has; image matching method, which is based on feature points and locates mines by search corresponding areas in reference images, requires sufficient feature points and high similarity between DEM data. The existing Mine Localization technologies usually fail when mine has no spatial reference and its DEM has insufficient feature points, or there are vast differences between DEM data. So, this paper do research on Image Fuzzy Matching technology for the purpose of solving two difficult problems(the lack of feature points when DEM is matched without spatial reference and low similarity between DEM data) and providing precise DEM matching a fine initial condition. The main contents and results of this paper are as follows:(1) Data used in the rectification of MRDSSD and MRD was combed. Data of mines without complete spatial reference information makes up the database of DEM whose corresponding mines are to be located. The DEM database of typical data with accurate location was established to verify the algorithm proposed in this paper.(2) Sixty-nine DEM images of ASTER GDEM V2 covering Sichuan province were downloaded to establish the database of reference DEM.(3) After comparing various image matching methods, the image matching method based on rectangle features was selected to do the research of Mine Localization in this paper.(4) Firstly, five image similarity indicators were constructed based on rectangle features, and then training data set was established by collecting 6072 positive samples and 29400 negative samples. Secondly, this paper statistics every indicator's classification performance respectively by analyzing the training data set. Thirdly, the classifier using to discriminate whether matched or not was obtained after using Gentle Ada Boost algorithm training massive samples. Finally, a model of DEM image fuzzy matching was established.(5) By verifying of four typical data, the method used in this paper shows a capability in solving two difficult problems, the lack of feature points when DEM is matched without spatial reference and low similarity between DEM data. All the above may provide precise DEM matching a fine initial condition and new method for image fuzzy matching and mine localization. |
PDF Full Text Request