[Objective] Through the analysis and research for optimal point cloud density mico DEM, this paper aimed to select the best point cloud density from it to achieve efficient, rapid access to micro-surface topography observations purpose. Upon which the computational cost was expected to be reduced under the precondition of observation accuracy.[Methods] Previous research of point cloud data proved that the optimal level of resolution about the data was 4 mm. The raw data were compressed by 7 ratings:100%, 75%, 50%, 25%, 10%, 5% and 1% to generate the corresponding DEM. Three most common DEM accuracy indicators:average error, root mean square error, standard deviation were used to evaluate the accuracy of the DEM above.[Results] (1) When the point cloud compression level was set above 15%, the average error unchanged and maintained at a low value with the decrease of point cloud density, when it was compressed <15%, the average error increased rapidly with the decrease of point cloud density.(2) When the compression of the point cloud was >10%, standard deviation was in a flat status with the decrease of the point cloud density. When it was <10%, the standard deviation of point cloud density decreases rapidly with the increasing of point cloud density.(3) When point cloud was compressed over 20%, with the decrease of the point cloud density, root mean square error was in the flat segmentation; when it was <20%, the root mean square error rapidly increased with the decrease of point cloud density.[Conclusion] The comparison of different compression ratings verified that the optimal compression level to generate micro-topography DEM should be 20%.