Individual Tree Segmentation Based on Mean Shift and Crown Shape Model for Temperate Forest

Tusa, Eduardo; Monnet, Jean‐Matthieu; Barré, Jean-Baptiste; Mura, Mauro Dalla; Dalponte, Michele; Chanussot, Jocelyn

doi:10.1109/lgrs.2020.3012718

Cited by 6 publications

(3 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Finally, the mean shift algorithm on the LiDAR point cloud did not perform as strongly as the other approaches for two main reasons; low point density and the influence of forest composition. First, the mean shift algorithm benefits from high point density (20 and 60 pulses per m2, Tusa et al 2020) providing abundant information on the spatial distribution of the neighbors of each point (Wu, Yao & Polewski, 2018). A low point density does not provide enough definition among clusters, which leads to undersegmentation, as was seen in the results.…”

Section: Delineationmentioning

confidence: 98%

See 1 more Smart Citation

Data science competition for cross-site delineation and classification of individual trees from airborne remote sensing data

Graves

Marconi

Stewart

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Delineating and classifying individual trees in remote sensing data is challenging. Many tree crown delineation methods have difficulty in closed-canopy forests and do not leverage multiple datasets. Methods to classify individual species are often accurate for common species, but perform poorly for less common species and when applied to new sites. We ran a data science competition to help identify effective methods for delineation of individual crowns and classification to determine species identity. This competition included data from multiple sites to assess the methods' ability to generalize learning across multiple sites simultaneously, and transfer learning to novel sites where the methods were not trained. Six teams, representing 4 countries and 9 individual participants, submitted predictions. Methods from a previous competition were also applied and used as the baseline to understand whether the methods are changing and improving over time. The best delineation method was based on an instance segmentation pipeline, closely followed by a Faster R-CNN pipeline, both of which outperformed the baseline method. However, the baseline (based on a growing region algorithm) still performed well as did the Faster R-CNN. All delineation methods generalized well and transferred to novel forests effectively. The best species classification method was based on a two-stage fully connected neural network, which significantly outperformed the baseline (a random forest and Gradient boosting ensemble). The classification methods generalized well, with all teams training their models using multiple sites simultaneously, but the predictions from these trained models generally failed to transfer effectively to a novel site. Classification performance was strongly influenced by the number of field-based species IDs available for training the models, with most methods predicting common species well at the training sites. Classification errors (i.e., species misidentification) were most common between similar species in the same genus and different species that occur in the same habitat. The best methods handled class imbalance well and learned unique spectral features even with limited data. Most methods performed better than baseline in detecting new (untrained) species, especially in the site with no training data. Our experience further shows that data science competitions are useful for comparing different methods through the use of a standardized dataset and set of evaluation criteria, which highlights promising approaches and common challenges, and therefore advances the ecological and remote sensing field as a whole.

show abstract

Section: Delineationmentioning

confidence: 98%

“…A test-time augmentation was implemented to improve the delineations. The INRAE-GISPA team implemented an adaptive 3D mean shift method using the LiDAR point cloud data (Tusa et al, 2020). Critical to this approach was the use of a probability density function (PDF) to identify clusters of points that define an ITC.…”

Section: Delineation Algorithmsmentioning

confidence: 99%

Data science competition for cross-site delineation and classification of individual trees from airborne remote sensing data

Graves

Marconi

Stewart

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Similar methods include integrating multispectral image information with MLS point clouds for tree identification [20]. Most existing methods can effectively segment and extract trees [19][20][21][22]. However, these methods that rely on MLS systems require expensive high-precision GNSS/IMU positioning and survey-grade laser scanners, which limit their large-scale deployment.…”

Section: Related Workmentioning

confidence: 99%

Simultaneous Vehicle Localization and Roadside Tree Inventory Using Integrated LiDAR-Inertial-GNSS System

Fan,

Chen,

Liu

et al. 2023

Remote Sensing

View full text Add to dashboard Cite

Autonomous driving systems rely on a comprehensive understanding of the surrounding environment, and trees, as important roadside features, have a significant impact on vehicle positioning and safety analysis. Existing methods use mobile LiDAR systems (MLS) to collect environmental information and automatically generate tree inventories based on dense point clouds, providing accurate geometric parameters. However, the use of MLS systems requires expensive survey-grade laser scanners and high-precision GNSS/IMU systems, which limits their large-scale deployment and results in poor real-time performance. Although LiDAR-based simultaneous localization and mapping (SLAM) techniques have been widely applied in the navigation field, to the best of my knowledge, there has been no research conducted on simultaneous real-time localization and roadside tree inventory. This paper proposes an innovative approach that uses LiDAR technology to achieve vehicle positioning and a roadside tree inventory. Firstly, a front-end odometry based on an error-state Kalman filter (ESKF) and a back-end optimization framework based on factor graphs are employed. The updated poses from the back-end are used for establishing point-to-plane residual constraints for the front-end in the local map. Secondly, a two-stage approach is adopted to minimize global mapping errors, refining accumulated mapping errors through GNSS-assisted registration to enhance system robustness. Additionally, a method is proposed for creating a tree inventory that extracts line features from real-time LiDAR point cloud data and projects them onto a global map, providing an initial estimation of possible tree locations for further tree detection. This method uses shared feature extraction results and data pre-processing results from SLAM to reduce the computational load of simultaneous vehicle positioning and roadside tree inventory. Compared to methods that directly search for trees in the global map, this approach benefits from fast perception of the initial tree position, meeting real-time requirements. Finally, our system is extensively evaluated on real datasets covering various road scenarios, including urban and suburban areas. The evaluation metrics are divided into two parts: the positioning accuracy of the vehicle during operation and the detection accuracy of trees. The results demonstrate centimeter-level positioning accuracy and real-time automatic creation of a roadside tree inventory.

show abstract

Three-Dimensional Reconstruction of Forest Scenes with Tree–Shrub–Grass Structure Using Airborne LiDAR Point Cloud

Xu,

Yang,

Wang

et al. 2024

Forests

View full text Add to dashboard Cite

Fine three-dimensional (3D) reconstruction of real forest scenes can provide a reference for forestry digitization and forestry resource management applications. Airborne LiDAR technology can provide valuable data for large-area forest scene reconstruction. This paper proposes a 3D reconstruction method for complex forest scenes with trees, shrubs, and grass, based on airborne LiDAR point clouds. First, forest vertical distribution characteristics are used to segment tree, shrub, and ground–grass points from an airborne LiDAR point cloud. For ground–grass points, a ground–grass grid model is constructed. For tree points, a method based on hierarchical canopy point fitting is proposed to construct a trunk model, and a crown model is constructed with the 3D α-shape algorithm. For shrub points, a shrub model is directly constructed based on the 3D α-shape algorithm. Finally, tree, shrub, and ground–grass models are spatially combined to achieve the reconstruction of real forest scenes. Taking six forest plots located in Hebei, Yunnan, and Guangxi provinces in China and Baden-Württemberg in Germany as study areas, experimental results show that the accuracy of individual tree segmentation reaches 87.32%, the accuracy of shrub segmentation reaches 60.00%, the height accuracy of the grass model is evaluated with an RMSE < 0.15 m, the volume accuracy of shrub and tree models is assessed with R2 > 0.848 and R2 > 0.904, respectively. Furthermore, we compared the model constructed in this study with simplified point cloud and voxel models. The results demonstrate that the proposed modeling approach can meet the demand for the high-accuracy and lightweight modeling of large-area forest scenes.

show abstract

Individual Tree Segmentation Based on Mean Shift and Crown Shape Model for Temperate Forest

Cited by 6 publications

References 14 publications

Data science competition for cross-site delineation and classification of individual trees from airborne remote sensing data

Data science competition for cross-site delineation and classification of individual trees from airborne remote sensing data

Simultaneous Vehicle Localization and Roadside Tree Inventory Using Integrated LiDAR-Inertial-GNSS System

Three-Dimensional Reconstruction of Forest Scenes with Tree–Shrub–Grass Structure Using Airborne LiDAR Point Cloud

Contact Info

Product

Resources

About