Abstract: The direct outputs of UAV oblique photography usually include 3D mesh, TDOM, DSM and so on. However, DEM that can be used for design and planning can't be directly generated without manual edits. The dense matching point clouds produced by the oblique images processing procedure are not fully used. The dense matching point clouds, similar to the structure of LiDAR point clouds, may be freely to set the point density. Regardless of the data amount, the density of dense matching point cloud can be multiple time of that of LiDAR point cloud. Besides, the dense matching point clouds have texture information without separate color assignment, which has a certain auxiliary effect on artificial visual editing after automatic classification of ground points. This paper compares and analyzes the dense matching point cloud and LiDAR point cloud, separately based on the basis of the oblique photographed data and LiDAR data in the same survey area, and verifies the feasibility of dense matching point cloud for ground point filtering and DEM production in the housing area and sparse forest area.

Key words: LiDAR point cloud, dense matching point cloud, ground cloud point filter, DEM