Coal mining has brought a series of environmental problems. Local government departments have issued relevant governance policies, but the premise of scientific prevention and control is to correctly grasp the actual distribution of various ground objects in the mining area. Using classification methods to extract ground object information based on remote sensing images can effectively realize mining area monitoring and provide reference for land and space planning and environmental protection in the mining area. Therefore, it is very important to select the appropriate scale and method to identify the ground object information of remote sensing image. in this paper, Landsat 8 images of the Wucaiwan mining area and GF-2 images of the Tebian coal mine were taken as the research objects, and unsupervised classification, supervised classification and object-oriented classification were used to identify and monitor the mining area's surface. The results showed that: (1) the classification effect of the Mahalanobis distance method was the best in terms of comprehensive operation process and classification accuracy. This method had high classification accuracy for GF-2 and Landsat 8 images. When classifying GF-2 images, the kappa coefficient reached 0.90, and the overall classification accuracy was 94.27%. When classifying Landsat 8 images, the kappa coefficient reached 0.85, and the overall classification accuracy was 90.02%. (2) The factors causing the classification error were 'homospectral foreign bodies' and 'mixed pixels'. (3) When combined with the actual needs and image characteristics, the extensive use of medium and high-resolution remote sensing images to identify and monitor the surface elements of mining areas can greatly improve the work efficiency and minimize the image costs. (4) The construction layout of tailings pond in the Tebian coal mine was conducive