Researchon The Single Tree Skeleton Modeling Based On Terrestrial Lidar Point Clouds

The three-dimensional model of trees is of great significance in virtual three-dimensional landscape,road navigation,forestry resource planning and design.The single tree point cloud data has a large amount of data,and there is a large amount of redundant data,and the single tree branches and leaves are occluded and disturbed by the wind,which may easily lead to the lack of single-tree branch data and excessive noise points,which can not accurately identify the branches.Therefore,it is difficult to directly carry out three-dimensional reconstruction of single single.The skeleton line of the tree can well represent the three-dimensional shape and geometric structure of the tree.It is widely used in the point cloud compression and three-dimensional modeling of trees.Therefore,this paper mainly extracts the single-tree skeleton based on the point cloud model of single-tree branches..The research contents of this paper are as follows:Because of the large amount of single-tree point cloud data,in order to quickly query and retrieve the neighborhood of single-tree point cloud,this paper constructs the KD tree structure of single-tree point cloud data to improve the single tree point cloud data processing efficiency.In this paper,the single tree with leaves is collected.Therefore,this paper uses the scanned single-tree three-dimensional point cloud data to extract the spatial characteristics,reflection intensity,RGB color features and other multi-dimensional features of the single-wood point cloud,in order to improve the efficiency of classification,through randomization.The forest algorithm is sorted according to its characteristic importance,removing redundant features,retaining RGB color,reflection intensity,and normal distribution characteristics as the basis for segmentation.The Ertrewe learning Machine(ELM)is used to classify and identify the single tree branches and leaves.In this paper,the single-tree branch point cloud is clustered by K-Means,and the clustering results are analyzed to identify the case where the point clouds of different branches belonging to the same upper-level branch are clustered into one class.Secondary clustering was performed using the K-Means algorithm.The single-tree branching point cloud is clustered completely according to the direction of its branch growth.After clustering,the point cloud is not completely perpendicular to the XOY plane,and it is inclined at a certain angle.Therefore,according to the different parts of the cluster,according to the characteristics that the branehes of the trees are approximately eylindrical,I calculate the rotation coordinates of the point cloud of the branch by using the Rodrigues rotation formula.As a rotation axis of the set of points to be rotated,a certain angle is rotated,so that the main normal direction vector of the rotated point set coincides with the z direction vector.The Rodrigues rotation matrix can be obtained from the rotation axis and the rotation angle,and the point set coordinates are calculated by calculating the point set.The most rotated point set is projected onto the XOY plane for iterative variable weight least squares method for circle fitting,and the fitting circle center transformation to the three-dimensional space is the skeleton point of the point set.For all the skeleton points extracted,constructing the adjacency graph matrix,and constructing a skeleton model by using a level set and a minimum spanning tree to connect skeleton points.The skeletal model can be obtained by connecting the geodesic distance to the source point and connecting the shortest path points.Comparing the effects of the skeletal models obtained by the two methods,it can be seen that the minimum spanning tree can avoid the generation of the closed loops of the skeleton points.Therefore,the minimum spanning tree is used to form the single-tree stem skeleton.Connect unconnected skeleton points of consecutive branches according to certain judgment conditions.After the trimming,the final skeleton model can be obtained.In the obtained single-tree branch skeleton,the linear points are connected between the skeleton points.In order to be closer to the real tree branch model,this paper uses the cubic B-spline function to interpolate the skeleton points and fit the branch skeleton curve.The skeleton after the cubic B-spline curve fitting is smoother,making it realistic.