基于点云数据的树木骨架提取算法OA

Tree point cloud models are of vital importance for applications in tree conservation,growth monitoring,and digital twin construction.Traditional point cloud skeleton extraction algorithms often encounter difficulties in accurately capturing the bifurcation structures and terminal branch extremities of tree point clouds.To address these limitations,this paper presents an improved point cloud-based skeleton extraction algorithm for trees.The proposed method first employs a modified L1-median skeleton extraction algorithm incorporating K-means clustering to identify branch center points.Subsequently,convex hull algorithms are utilized to extract the outer boundary points of the tree structure.Furthermore,a novel nearest-point search algorithm based on these boundary points is introduced to accurately detect bifurcation points.The final skeletal point cloud is obtained by integrating these extracted center points,bifurcation points and boundary points.The results demonstrate that the skeleton points extracted by this algorithm exhibit superior correspondence with the original point cloud,as evidenced by quantitative evaluations using Hausdorff distance and chamfer distance metrics.Compared with traditional methods,the skeleton points extracted by this algorithm,after normalization processing,have a Hausdorff distance and chamfer distance of less than 0.5,and overall performance is superior to the comparison algorithm.Additionally,the implementation of cubic Bézier curves for connecting the skeletal points effectively preserves the original topological characteristics of the tree structure,thereby significantly improving both the accuracy and completeness of the skeleton extraction process.