Study On Mine Area Information Extraction Based On Object-oriented High-resolution Remote Sensing Image Classification And Its Application

High spatial resolution remote sensing data is an important direction of the development of remote sensing technology. And high-resolution remote sensing research and applications is also increasing. Because of its large volume of data, rich information, complex spectral, it has become a difficult deal. Traditional pixel-based classification of remote sensing technology for high-resolution remote sensing image processing result is not satisfactory. Thus, some scholars use object-oriented information extraction technology to classify the remote sensing image, greatly increased the accuracy of high-resolution remote sensing image classification. Object-oriented classification of remote sensing image is divided into a series of image objects, using objects'spectrum, the shape and texture characteristics, adopting fuzzy classification to achieve classification information extraction. This method is particularly suitable for high-resolution remote sensing data for information extraction.In this paper, taken SPOT5 image of a surface mine in Benxi as an study area, and used of object-oriented classification to study the mine survey information extraction, and with conventional methods were compared. First of all, according to the feature type of the study area, with the method of Area Ratio Mean analysis the optimal segmentation scale, select three scale levels to form multi-scale segmentation level of network. Then, light of the specific characteristics of surface features, select the object's spectral characteristics, texture, shape features and class-related features and so on, establish classification rule base, adopt the membership functions method and the standard nearest neighbor method to classify image to extract the study area object classes. Finally, the accuracy of the classification results were evaluated, statistic each polygon area and coordinates of the classification maps and compare to that of the field verification of the visual interpretation, further analysis of polygon classification accuracy for areas and coordinates. In addition, using the traditional classification methods (maximum likelihood) to deal the study area of remote sensing image, and compare the accuracy to object-oriented classification of the results. Concluded that the object-oriented classification than the pixel-based classification more precise, more suitable for high-resolution remote sensing classification. The location attributes and shape characteristics of the extracted object classes have a higher consistency to the true surface features. The information of Object-oriented classification could meet the plan production and application of mine remote sensing survey.Established the object-oriented classification rules and technical processes of the information of the open-pit mine, which makes up for the blank of the open-pit mine in the application of remote sensing survey in China. It laid a good foundation for further study and application.