| Airborne laser scanning(ALS)adopts top-down spatial sampling method,which can quickly obtain the tree height and canopy information of the forest,and the laser point cloud has limited description of the internal trunk and branch information of the forest stand due to the influence of the laser intensity and the density of the forest stand.Terrestrial laser scanning(TLS)is able to obtain interior point cloud information of forest stands at close range and high density through bottom-up scanning,making it difficult to obtain complete canopy information.The fusion of ALS and TLS point clouds can complement each other’s observation perspectives,which can realize the accurate extraction of parameters theoretically such as the number of forest trees,tree height and diameter at breast height(DBH).The study was conducted with three fir plots and one natural broadleaf forest plot in Central South University of Forestry and Technology Lutou Experimental Forestry Site.Using ALS and TLS to collect point cloud data,the registration of ALS and TLS point clouds was performed.The parameters of DBH and tree height were extracted on the basis of individual tree segmentation.The accuracy of parameter extraction was evaluated by combining the actual measurement data of the plots.The main outcomes and conclusions are as follows:(1)A terrestrial-airborne LiDAR point cloud fusion method based on the shape feature points of canopy gap was proposed.The original point cloud was preprocessed;next,the canopy point cloud was separated using height threshold and a canopy point cloud density model was generated to extract the canopy gap vector;then the weighted effective area algorithm was used to extract the canopy gap shape feature points;finally,the registration of the point cloud was achieved based on the Coherent Point Drift algorithm and the Iterative Closest Point algorithm.The experimental results showed that the proposed method can achieve the automatic fusion of terrestrial-airborne LiDAR point clouds.(2)The point cloud registration method of canopy gap shape feature points was superior to the registration method of canopy feature points.Compared with the registration method based on canopy feature points,the terrestrial-airborne LiDAR point cloud registration method proposed in this paper had the advantages of simple feature point extraction steps,low computation demanding,low time complexity and more automatic;in terms of registration accuracy,the residuals of the plots registration distance were reduced by 0.04 cm,0.42 cm,0.43 cm and 0.42 cm,which indicated that the proposed registration method of canopy gap shape feature point was better than the registration method of canopy feature point.(3)Terrestrial-airborne LiDAR point cloud fusion can improve the accuracy of tree height parameter extraction effectively.In order to verify the effectiveness of fused point clouds in forest parameter extraction,parameters of DBH and tree height were extracted and comparatively analyzed from TLS point clouds and fused point clouds.The results showed that there was no significant difference in the extraction accuracy of DBH between TLS point clouds and fused point clouds,and the extraction accuracy can reach more than 95%;after the fusion of TLS and ALS point cloud,the extraction accuracy of tree height was significantly improved,with the highest Coefficient of Determination R2 improved by 25.4%,the Root Mean Square Error reduced by 0.45 m,and the relative Root Mean Square Error reduced by 4.9 percentage points.Based on the above conclusions,the novel terrestrial and airborne LiDAR point cloud fusion method proposed in this study can improve the efficiency and automation of feature point extraction and registration accuracy;the terrestrial and airborne LiDAR point cloud fusion can better quantify the structure of trees and forests,which is more conducive to monitoring and studying forest ecosystems. |
PDF Full Text Request