Planes are essential features to describe the shapes of buildings. The segmentation of a plane is significant when reconstructing a building in three dimensions. However, there is a concern about the accuracy in segmenting plane from point cloud data. The objective of this paper was to develop an effective segmentation algorithm for building planes that combines the region growing algorithm with the distance algorithm based on boundary points. The method was tested on point cloud data from a cottage and pantry as scanned using a Faro Focus 3D laser range scanner and Matterport Camera, respectively. A coarse extraction of the building plane was obtained from the region growing algorithm. The coplanar points where two planes intersect were obtained from the distance algorithm. The building plane’s optimal segmentation was then obtained by combining the coarse extraction plane points and the corresponding coplanar points. The results show that the proposed method successfully segmented the plane points of the cottage and pantry. The optimal distance thresholds using the proposed method from the uncoarse extraction plane points to each plane boundary point of cottage and pantry were 0.025 m and 0.030 m, respectively. The highest correct rate and the highest error rate of the cottage’s (pantry’s) plane segmentations using the proposed method under the optimal distance threshold were 99.93% and 2.30% (98.55% and 2.44%), respectively. The F1 score value of the cottage’s and pantry’s plane segmentations using the proposed method under the optimal distance threshold reached 97.56% and 95.75%, respectively. This method can segment different objects on the same plane, while the random sample consensus (RANSAC) algorithm causes the plane to become over-segmented. The proposed method can also extract the coplanar points at the intersection of two planes, which cannot be separated using the region growing algorithm. Although the RANSAC-RG method combining the RANSAC algorithm and the region growing algorithm can optimize the segmentation results of the RANSAC (region growing) algorithm and has little difference in segmentation effect (especially for cottage data) with the proposed method, the method still loses coplanar points at some intersection of the two planes.

The tree skeleton is one of the important parameters for building 3D model of the tree. The accurate extraction of tree skeleton is of great significance for tree visualization. This paper proposed an effective method for extracting tree skeleton of individual trees. The complexity of canopy structure was reduced by slicing. The leaf and wood components were separated by combining classification and segmentation methods. L 1 -median algorithm was used to extract the tree skeleton points accurately. The results show that the method can accurately extract the tree skeleton from terrestrial LiDAR data. The extracted skeleton had a good conformance with the point cloud of the tree in morphology.

Semantic segmentation for 3D point clouds plays a critical role in the construction of 3D models. Due to the sparse and disordered natures of the point clouds, semantic segmentation of such unstructured data yields technical challenges. A recently proposed deep neural network, PointNet, delivers attractive semantic segmentation performance, but it only exploits the global features of point clouds without incorporating any local features, limiting its ability to recognize fine-grained patterns. For that, this paper proposes a deeper hierarchical structure called the high precision range search (HPRS) network, which can learn local features with increasing contextual scales. We develop an adaptive ball query algorithm that designs a comprehensive set of grouping strategies. It can gather detailed local feature points in comparison to the common ball query algorithm, especially when there are not enough feature points within the ball range. Furthermore, compared to the sole use of either the max pooling or the mean pooling, our network combining the two can aggregate point features of the local regions from hierarchy structure while resolving the disorder of points and minimizing the information loss of features. The network achieves superior performance on the S3DIS dataset, with a mIoU declined by 0.26% compared to the state-of-the-art DPFA network.

Canopy gaps affect the spatial distribution of radiation in the canopy. The estimation of gap fraction (GF) is important for the study of leaf area index (LAI). Terrestrial laser scanning (TLS) has been widely used for retrieving canopy structure parameters through massive high-resolution spatial samples in the form of 3-D point cloud data sets. Monte Carlo simulations can be used to obtain approximate solutions to quantitative problems through a large number of random sampling while avoiding complicated mathematical calculations. Monte Carlo simulations are suitable for analyzing TLS data and could help overcome resolution reductions inherent to current point cloud processing approaches. However, few studies have applied the Monte Carlo method to capture the canopy structure features implicit in high-resolution point clouds. This letter proposes a method for estimating the GF based on Monte Carlo simulations. A large number of randomly simulated laser beams were emitted to identify the canopy and gaps according to the discrimination distance, and the results were compared with the GFs derived from digital hemispherical photography (DHP). In general, the proposed TLS method estimated smaller GFs than the DHP method since DHP was vulnerable to exposure conditions and complex canopy structure. The GF estimated by the two methods is consistent (R² = 0.7522), indicating the effectiveness of the Monte Carlo method.

As one of the most important components of plants, the reconstruction of high-precision leaf models is a critical step for building tree models. According to the morphological structure of leaves, this paper proposes a leaf reconstruction method based on the laser point cloud data. The moving least squares method is used to fit the leaf surface to extract the complex profile information of the leaf, and the delaunay triangulation algorithm is used to reconstruct the three-dimensional model of the fitted leaf point cloud data. The results show that the proposed method can not only realize the three-dimensional reconstruction of the leaf, but also reflect the morphological characteristics of real leaf.