In this paper, we propose an efficient terrain meshing method based on discrete point cloud, which is based on the existing meshing method of 3D reconstruction. Our meshing method uses the DTM (Delaunay Triangle Mesh) as a criterion to establish vertex constraints and evaluation functions. We start from a seed triangle and use the growing method to extend the outer-edge of each triangle until all the triangles have been finished. Every time, we select the point with maximum evaluation function value to generate a new triangle. In order to improve the efficiency, we design a retractable bounding box for spatial subdivision and optimize the search algorithm of edges and points. Besides, we identified the rendering surface while meshing. The experimental result shows that our method can effectively mesh the terrain discrete point cloud and keep the characteristics of the original 3D terrain well.