Research On Mining Development Area Information Extraction Based On Object-oriented Classification

With the application of remote sensing technology in mining monitoring more and more widely,the demand for remote sensing classification software is also getting higher and higher,but the use of existing classification software in mining area extraction still need to rely on a wealth of mine interpretation experience and professional geological knowledge.In this paper high-resolution image is taken as the data source,and the information of mining area is extracted automatically by object-oriented remote sensing classification technology.This can save time and improve work efficiency.Firstly,the paper studies on the principle and implementation of the image segmentation algorithm to improve the precision by using marker watershed algorithm;Secondly,the classification algorithm is adopted to realize the automatic extraction of mine features based on high-resolution image,that is,stope,mine construction,transit site,solid waste.The main contents and conclusions are as follows:(1)The image is preprocessed by the anisotropic diffusion algorithm and uses the evaluation index to select the best filter parameters.The good edge preserving of the algorithm is helpful to extract the shape features and reduce the over-segmentation phenomenon to a certain extent.(2)The watershed algorithm is a spatial domain segmentation method,which is widely used in the segmentation of remote sensing image,but its most obvious drawback is that the over-segmentation phenomenon is serious.In this paper,we use the marker watershed algorithm to extract the mine development area as the foreground feature.Finally the experimental results show that the watershed algorithm has better segmentation performance.(3)Analysis and combination of mine features.This paper involves the spectral,shape and texture characteristics of the mine development area.On the one hand,it selects the reasonable combination of characteristics,on the other hand,according to the vector basic data the mine interpretation sign is created,and then it will extract the characteristics of mine interpretation sign.(4)The fuzzy C-means clustering algorithm is used to classify the objects.On the one hand,the algorithm has good convergence and high efficiency,and it is suitable for multi-dimensional feature classification.On the other hand,there are some kinds of interactive connectivity and fuzziness in the different mining classes of the actual form,just right the algorithm is developed in the fuzzy set theory,which is suitable for solving this kind of uncertainty problem,so we choose the fuzzy C-means clustering algorithm.The results show that the total accuracy of the classification can reach 89.75% and the kappa coefficient is 0.8462.The large mine area that is extracted from the high resolution image of Hengfeng County,Shangrao City,Jiangxi Province can achieve the accuracy of 74.54%,kappa coefficient of 0.6356.