Abstract:Laser point cloud provides a high-precision data source for the 3D model reconstruction of existing trees. However, due to the complexity of tree structure and the data missing of point cloud, it is difficult to reconstruct complete, detailed and accurate 3D tree models from discrete points, especially the points are incomplete due to single-view scanning. To this end, a new point cloud driven method based on skeleton refined extraction is proposed in this paper. First, as the focus of refined modeling, the main (thick) branches are separated from the raw tree points. Then, the main branch points are over-segmented into some whorl-segments, and the corresponding raw points on the branch surfaces of each whorl-segment are shrunk to generate smooth, continuous and local detail preserved potential skeleton nodes of the main branches. Next, a new shortest path metric meeting the natural growth characteristics of trees is adopted to extract the complete, detailed and correctly connected tree skeleton from both of the shrunk main branch points and the non-main-branch points. After that, to ensure the skeleton centralization, especially that extracted from single-view scanned data, the extracted skeleton is translated by a combination of cylinder fitting and allometric theory. Finally, the complete, detailed and accurate 3D tree model is expressed by a series of cylinders based on the extracted refined skeleton. The proposed method is tested with 7 different tree point clouds, some of which are obtained by single-view scanning and suffer from serious data missing. Both qualitative and quantitative evaluations prove the effectiveness of this method for the refined reconstruction of trees from point clouds. Overall, the average modeling accuracy and its standard deviation of the used different trees are 2.58 cm and 3.20 cm respectively.

Reconstruction Method of Tree Geometric Structures from Point Clouds Based on Angle-constrained Space Colonization Algorithm

Improved Method for 3D Reconstruction of Tree Model Based on Point Cloud Data

Improved Fast 3D Reconstruction Algorithm Based on Laser Scanning Point Cloud

3D Modeling of Laser-Scanned Trees Based on Skeleton Refined Extraction.

Research on 3D Reconstruction Algorithm of Citrus Seedlings Based on Multi-View Geometry

Automatic Reconstruction of Tree Skeletal Structures from Point Clouds

3D Tree Modeling from Incomplete Point Clouds Via Optimization and L1-MST.

Tree branching reconstruction from unilateral point clouds

Reconstruction of Single Tree with Leaves Based on Terrestrial LiDAR Point Cloud Data.

Outdoor Scene 3D Reconstruction from Multiple Point Cloud

3D Reconstruction of a Single Tree from Terrestrial LiDAR Data

A Local Structure and Direction-Aware Optimization Approach for Three-Dimensional Tree Modeling.

Tree Geometrical 3D Modeling from Terrestrial Laser Scanned Point Clouds: A Review

Efficient Tree Modeling from Airborne LiDAR Point Clouds.

Point Cloud Registration Based on Fast Point Feature Histogram Descriptors for 3D Reconstruction of Trees

Restoration of Individual Tree Missing Point Cloud Based on Local Features of Point Cloud

A Structure-Aware Global Optimization Method for Reconstructing 3-D Tree Models from Terrestrial Laser Scanning Data

A 3D individual tree modeling technique based on terrestrial LiDAR point cloud data

Three-Dimensional Reconstruction of Leaves Based on Laser Point Cloud Data

Development of a Precise Tree Structure from LiDAR Point Clouds

Reconstruction Of Individual Trees Based On Lidar And In Situ Data