针对传统地表移动监测方法周期较长、工作量大的问题，通过无人机LiDAR和点云滤波获取地面点云，并构建沉陷DEM，实现地表沉陷监测的方法具有快速、高效的优势；由于现有点云滤波和插值算法构建的沉陷DEM模型仍会包含噪声，限制了该技术在矿区的普及,因此，进一步研究了沉陷DEM噪声的去除方法，对比分析了多重滤波与经典滤波方法。实验分析结果表明：在几种去噪方法中，中值滤波组合维纳滤波的去噪效果最好，保留了下沉盆地的细节特征，能够满足矿区地表形变监测的基本要求。

Aiming at the problems of long cycle time and large workload of traditional surface movement monitoring methods, the method of acquiring ground point clouds and constructing subsidence DEM through UAV LiDAR and point cloud filtering enables surface subsidence monitoring fast and efficient. Because of the subsidence DEM models constructed by existing point cloud filtering and interpolation algorithms still cover noise, which limits the popularity of this technology in mining areas, therefore, it is significant to further study the removal method of subsidence DEM noise, compare and analyze the multiple filtering and classical filtering techniques. Experimental analysis results show that the median filter combined with Wiener filter has the best denoising effect among several denoising methods, which can keep the details of the subsidence basin and meet the basic requirements of surface deformation monitoring in mining areas.