Study On Mine Area Utilization Information Extraction Based On High-Resolution Remote Sensing Image Classification And Application

The traditional pixel-based remote sensing classification algorithm is mainly used to extract the low-resolution remote sensing satellite data, And with high-resolution commercial satellite, a growing number of high-resolution remote sensing images are used by the majority of users. High resolution image data volume, spectral information-rich, making extraction method based on low-resolution remote sensing satellites increasingly do not apply to high-resolution images, especially when the information extraction in the mine. Because the mine information is complex, when relying solely on spectral information extraction, inevitably produce " the same feature makes different spectral spectrum" and "the same spectrum of different objects" phenomenon, which led to the extraction of the results have been unsatisfactory. So the image information extraction method is applied to the object-oriented classification of remote sensing images, which greatly improved the image of high extraction accuracy, especially in terms of information extraction mine, which has unparalleled advantages.In this paper, we use high resolution image data experiment Beijing’s FangshanMentougou mine, And integrated use of object-oriented classification techniques to extract information of mine, then compared with the traditional pixel-based classification approaches and manual interpretation methods. first of all, we pretreated the data in the test area, on this basis, pixel based supervised classification and unsupervised classification are conducted, and manual interpretation are added;Then, we then focus on the experimental area of multi-scale segmentation, When in the multi-scale segmentation, we focus on explaining the comparative selection of parameter factor, Then select the study of texture, spectral characteristics,shape features and other characteristics as a parameter, establish affiliation classification layers, using specified parameters and fuzzy Mathematics extract mine information, Finally, by using the confusion matrix, assess the accuracy of classification results based on comparative Analysis of traditional methods and object-oriented classification method, and with the results of manual interpretation of assistance in explaining, Obtained that using object-oriented classification method works better than traditional classification based element method, especially in the mining information extraction, and have better results, could meet the production and application of remote sensing survey in the mine.