中文摘要：

为提高基于DEM高程数据的三维大地形绘制效率，提出基于K-means聚类和动态LOD的三维地形建模方法。对高程数据集使用K-means聚类分析方法，发现地形数据的自相似性，建立数据采样点与地貌特征分类的自动关联；依据采样点的地貌属性值完成粗化处理，构建不规则三角网模型；采用局部优化算法进行有限域内的三角形分裂以实现三角网的动态更新，结合克里金插值法实现插入点的精确计算，形成LOD层次细节多分辨率模型。实验结果表明，该方法对不同类型地形都获得了较高的数据简化率，提高了三维模型的绘制效率，具有较好的地形特征保持性。

英文摘要：

To improve the efficiency of rendering for the three-dimensional massive terrain based on DEM data,a method based on K-means clustering analysis and dynamic LOD was presented to construct the terrain model.The elevation data sets were firstly trained using K-means clustering analysis,and the self-similarity among the terrain data was found,then the associated relationship was established automatically between sampling points and the classification of geomorphology.Secondly,the terrain data were roughened based on the values of the additional property of the geomorphic features in the sampling points,and the terrain was built by multi-resolution triangulated irregular network and the level of detail model.Finally,the details of the local region were reconstructed according to the thinning evaluation function of the nodes,completing the dynamic updating for the triangulation.For the new insertion point,the Kriging interpolation method was used to compute its value.The experimental results show that the method gains good performance in reducing different types of terrain,improves the efficiency of rendering three-dimensional model and retains more landscape features.