A Research On Mine Area Information Extraction Based On Remote Sensing Image Of Ikonos Using Object-oriented Classification

The application of high resolution remote sensing data for environmental monitoring mine is a development trend in recent years of monitoring work. Information extraction technology is the key in the application of remote sensing data. The traditional pixel classification method just considers spectral information, so the quantity of information extraction is small and the classification accuracy is low, which is difficult to meet the high resolution data information extraction. Therefore, some scholars introduce the "object oriented" extraction technology, which achieved very good results. The object-oriented information extraction technology research started late in our country, and concentrated more in the marine environment and monitoring area.In this article, it takes the IKONOS data of Baiyin coal mining area as the data source in2011, applying object-oriented classification technology for information extraction of land occupation and comparing with pixel information extraction methods. It makes the statistical analysis of land occupation of mine area at the end of the working and forms a dynamic analysis report at last. The main achievements are obtained as follows:1. Based on region growing algorithms of heterogeneity minimum, the image is segmented, achieving the image data of the multi-scale segmentation, the segmentation scales are10,30and50. It forms an interconnecting network layer between the layers and realizes information multi-scale, multi-level expression.2. Applying fuzzy classification techniques and contrasting to the traditional pixel-based classification method (maximum likelihood), data displays:the overall classification accuracy and the overall kappa coefficients of the object oriented were83%,0.8135, and the maximum likelihood is55%,0.4774. The results show that:the precision of object-oriented classification is higher and better than the pixel-based classification. The oriented-classification could meet the needs of mine remote sensing survey and monitoring.3. In2011, the extraction of all types of mining for map spot is812, and the area is972.93hectares. Contrasting to the2010visual interpretation map, the mining area cover decreases87.3hectares, while the site of solid waste increases88.01hectares. The analysis points out,2010Mine Remote Sensing Survey and the monitoring work have gained some effects, the phenomenon of illegal mining and occupation of cultivated land decreases. Meanwhile, the area of solid waste occupation is increasing, which should arouse the attention from government departments.