A Point Cloud Thinning Method Of Airborne LiDAR Based On Multivariate Terrain Features

The airborne LiDAR(Light Detection And Ranging)can quickly acquire high-precision,high-density terrain information,which is an efficient terrain detection technology.The point cloud data collected by airborne LiDAR is often massive.However,many application areas do not need such high-density point cloud data.The large amount of data increases the difficulty of data processing at a later stage.In order to improve the efficiency of data processing operation under the condition of meeting the accuracy,and reduce the difficulty of later data processing,it is very necessary to carry out the point cloud thinning of airborne LiDAR.The current point cloud thinning algorithm is mostly based on single terrain feature.However,a single terrain feature parameter cannot realize the comprehensive expression of complex terrain.Therefore,this paper proposes an airborne LiDAR point cloud thinning algorithm based on multivariate terrain features.After filtering,a multivariate terrain feature complexity model was constructed using principal component analysis and as an indicator of point cloud selection,achieving high-precision thinning of point clouds.The main work in this paper focuses on the following points:1.An improved cloth simulation filtering method was researched with setting the rigidness parameters.Based on the cloth simulation filtering algorithm,an improvement was made on the rigidness parameters of the simulated cloth.That is,on the basis of partitioning and the type of terrain,the matching IMRI(Improved Rigidness)value was selected so that the simulated cloth and the terrain feature are more consistent,and the accuracy of the point cloud filtering is improved.2.An airborne LiDAR multivariate terrain feature extraction method was achieved.Through the quadric surface fitting of the local terrain,the geometric rules of the terrain data and the fitting surface were established;using the LM(Levenberg-Marquardt)algorithm to iterate the parameters,and the optimal terrain fitting parameters are obtained.Based on the fitted model,six types of terrain feature information(local point density,elevation standard deviation,slope,Gaussian curvature,mean curvature,and roughness)were extracted rapidly3.A point cloud thinning algorithm based on multivariate terrain features was proposed.Four types of terrain feature parameters including elevation standard deviation ?H,slope(p,Gaussian curvature Cg and roughness K,were selected as variable factors.Based on principal component analysis and multi-factor theory,a multivariate terrain features complexity model T(?H,?,CG,Kr)was constructed.Taking the obtained T value as point cloud dilution criterion,and a point cloud thinning criterion was designed,eventually the high-precision thinning of point cloud was achieved.Comparing the results of this algorithm with the slope-based point cloud thinning and TerraScan software's thinning results.The experimental results show that the proposed algorithm has higher precision and robustness.