Classification of the mining land cover based on density segmentation

The Datun mining area is the test area in this paper,and the purpose is to obtain the mining land cover classification map in 2003.The data source used in this paper is the Landsat enhanced thematic mapper plus(ETM)remote sensing data.By obtaining the normalized difference vegetation index(NDVI)of the study area,the land cover classification information is extracted using the density segmentation method.Because the results cannot distinguish the construction land and wetland,this paper obtains the humidity information of multi-spectral data through the tasseled cap transformation.From the density segmentation image of humidity information,the construction and wetland types can be clearly distinguished.Finally,combining the two classification maps,the visual interpretation using Landsat ETM fusion image with 15-m resolution helps to get the final classification results.After classification accuracy assessment,the overall accuracy calculated from the classification confusion matrix is 86%.This result can be applied in actual project.