Pith. sign in

REVIEW 4 major objections 5 minor 28 references

Automatic marker-free registration based on similar tetrahedras for single-tree point clouds

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper proposes AMRST, a marker-free method that registers single-tree laser scans by matching similar tetrahedra built from branch skeletons, and reports substantially higher accuracy and speed than ICP and NDT.

desk verdict Plausible skeleton-based tetrahedron registration for single-tree point clouds, but the accuracy evaluation is unverifiable as reported. read the letter →

arxiv 2411.13069 v1 pith:O2JE53NV submitted 2024-11-20 cs.CV

Jing Ren , Pei Wang , Hanlong Li , Yuhan Wu , Yuhang Gao , Wenxin Chen , Mingtai Zhang , Lingyun Zhang This is my paper · ORCID
classification cs.CV
keywords pointcloudregistrationsingle-treeterrestriallaserscanningmarker-freetetrahedramatchingskeletonkeypointswood-leafseparationLM-ICP
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper introduces AMRST, a fully automatic, marker-free method for registering two terrestrial laser scans of the same tree. It separates each scan into wood and leaves, extracts a minimum-spanning-tree skeleton from the wood points, and selects key points where branches fork or end. It then searches both scans for pairs of similar tetrahedra whose vertices are those key points; matching four corresponding vertices supplies the coarse transformation, and a Levenberg-Marquardt ICP on the leaf points refines it. On eight trees of different species the method reports RMSE and Hausdorff distances far below those of ICP and NDT, with total runtimes of about a second to a few seconds per tree (up to 593x faster than ICP and 113x faster than NDT). If these results hold, complete single-tree point clouds for forestry measurements can be assembled without placing artificial targets in the field.

What carries the argument

The load-bearing object is the congruent tetrahedron pair defined on skeleton key points. The pipeline extracts an MST skeleton from the wood point cloud, labels branch points (nodes with more than two children) and endpoints (branch tips and the root), and keeps a depth-filtered set of the first five nodes along each root-to-tip path. It then forms tetrahedra from these key points, compares edge lengths with a relative tolerance epsilon, ranks tetrahedra by volume, checks that volume ratios are within beta, and verifies congruence by SVD with threshold delta. The four matched vertices of a congruent pair provide the coarse transformation; the fine stage then uses LM-ICP on leaf point clouds, minimizing point-to-point distance with a Levenberg-Marquardt update.

What would settle it

On a tree with dense foliage, scan the same tree from two positions and run the method: if coarse-registration RMSE jumps to the meter level (the range where ICP and NDT fail) instead of staying near the decimeter level reported for the eight test trees, the skeleton-consistency assumption is not met. Scanning the same tree in leaf and leaf-off would isolate how much of the registration quality depends on visible branch structure.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that branch skeletons give viewpoint-invariant key points for single-tree registration, and four such points define tetrahedra that are congruent across scans up to a small tolerance. The method builds all tetrahedra from each key-point set, filters edges by relative length error, sorts them by volume, and accepts a match when volume ratio is near one, sorted edges agree, and an SVD reconstruction error stays below threshold; the matched vertices then give the rigid transformation for coarse registration. The paper further claims that refining with leaf points rather than trunk points avoids the tight-fit problem of trunk geometry and converges quickly because the coarse alignment is already good. The reported outcome is coarse registration already more accurate than the full ICP and NDT baselines, and fine registration reducing error further (e.g., RMSE 0.06 m and Hausdorff distance 0.03 m on Tree01).

Load-bearing premise

The method assumes the two scans' branch skeletons are topologically similar enough that branch points and endpoints can be matched; when dense foliage hides branches or branching is ambiguous, those key points and the tetrahedra built from them lose correspondence.

Editorial extensions

If this is right

  • Forestry workflows can register single-tree scans without physical markers, removing a time-consuming field step.
  • The reported sub-0.1 m RMSE after fine registration is in the range useful for diameter-at-breast-height and biomass estimation.
  • Because total runtime is on the order of seconds per tree, the method can be applied to plot-level data in reasonable time.
  • The coarse registration's independence from the final ICP initialization means the pipeline does not need a user-provided initial alignment.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The tetrahedron-matching core is not tree-specific: any object scanned from two viewpoints that yields a stable skeleton (root systems, river deltas, piping) could be registered by the same congruent-tetrahedra search.
  • Because the matching stage is deliberately decoupled from wood-leaf separation, improvements in foliage removal should translate directly into better registration for dense canopies without redesigning the matching step.
  • The method leaves the question of heavy leaf occlusion open; a direct test would compare registration of the same tree scanned in leaf and after leaf-off, isolating the contribution of visible branching structure to registration success.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper proposes AMRST, a marker-free automatic registration method for single-tree terrestrial laser scanning point clouds. The pipeline separates wood and leaves, generates a tree skeleton from the branch points, extracts key points (branch points and endpoints) from the skeleton with a depth-filtering mechanism, and then finds pairs of similar tetrahedra in the two key-point sets to compute a coarse transformation. Fine registration is performed by applying Levenberg-Marquardt ICP to the leaf point clouds. The method is evaluated on eight trees of different species against ICP and NDT, reporting RMSE and Hausdorff distance as accuracy metrics and runtime as the efficiency metric. The paper claims that AMRST significantly outperforms both baselines in accuracy and is up to 593x and 113x faster than ICP and NDT, respectively.

Significance. If the reported results are reliable, the method would offer a practical, automatic, marker-free alternative for single-tree TLS registration, a task that is challenging due to the lack of distinct planar features and the structural complexity of tree canopies. The paper's main strengths are its coherent use of skeletal topology to create repeatable key points, the tetrahedron-based similarity matching for coarse alignment, and the explicit use of leaf point clouds for efficient fine registration. The speed advantages, if reproducible, would be valuable for forestry applications. However, the significance is substantially tempered by weaknesses in the evaluation: the accuracy metrics are undefined with respect to ground truth or correspondence, the method's free parameters are not reported, and the runtime comparison does not clarify which preprocessing steps are included. These issues currently prevent the central claims from being verified, so the paper's practical impact remains a promise rather than a demonstrated result.

major comments (4)
  1. [Section III.B, Table 2] The central accuracy claim that AMRST 'significantly outperforms both ICP and NDT' is not verifiable because the RMSE and Hausdorff distance are not defined with respect to any ground truth or correspondence protocol. The RMSE formula in Section II.F describes y_i as 'actual observation' and y_i_hat as 'predicted value', but in the context of point-cloud registration these terms have no operational meaning unless the text specifies which point sets are compared, how correspondences are established, and whether a known transformation (e.g., from scanner poses or artificial markers) was used as reference. If the RMSE is computed via nearest-neighbor distances after alignment, it may simply reflect initialization quality rather than true accuracy; if it is computed from the correspondences obtained by the tetrahedron matching itself, the metric is circular and biased in favor of AMRST. The Hausdorff distance is similarly ambiguous because no statement clarifies whether it is measured before or after fine registration and whether it uses the full point clouds or only the leaf/wood subsets. Without this protocol, the numerical values in Table 2 cannot be reproduced or interpreted, and the claimed superiority over ICP and NDT is unsupported.
  2. [Section II.D, Tables 2 and 3] The free parameters of the method are not reported, making the experiments non-reproducible and the claimed robustness questionable. The edge-length tolerance epsilon, the volume-ratio tolerance beta, the SVD reconstruction threshold delta, and the depth threshold of five nodes for key-point selection are all introduced in Section II.D, and the paper states recommended ranges (epsilon between 0.01 and 0.1, beta less than 10%, delta less than 0.1) but never states the actual values used for the eight trees. If these parameters were tuned on the test data, as the phrasing 'based on experimental testing' suggests for the leaf-point-cloud choice, then the reported accuracy may reflect overfitting rather than generalizable performance. At minimum, the paper should report the parameter values and, ideally, a sensitivity analysis showing how the RMSE and HD vary with each parameter.
  3. [Section III.B, Table 3] The runtime comparison lacks a clear accounting of which computational stages are included, so the speed advantage (up to 593x faster than ICP) may be inflated. The AMRST pipeline includes wood-leaf separation, skeleton generation, key-point extraction, tetrahedron matching, and fine registration; the table reports 'coarse registration time', 'fine registration time', and 'total time' for AMRST, but it is unclear whether the total time for ICP and NDT includes any preprocessing or is just the pure registration time on the full point cloud. Since AMRST operates on the skeleton and leaf point clouds, which are much smaller than the raw scans, the comparison is only meaningful if the cost of generating these reduced representations is included and if the baseline methods are given the same starting conditions. The paper should clarify the exact time breakdown and define the hardware/implementation details for all three methods.
  4. [Section IV] The paper's own discussion acknowledges that the method 'may struggle with trees whose branch structure is not clearly defined' and that dense leaves can obscure branches, but the experiments do not quantify the sensitivity to these factors. Given that the coarse registration is entirely dependent on the topological consistency of the skeleton between the two scans, and the wood-leaf separation method is only referenced (Section II.C, reference [34]) rather than described or evaluated, the paper should include a sensitivity analysis or at least report the number of key points extracted and matched per tree, the success/failure of tetrahedron matching, and the quality of the recovered transformation against a known ground truth. Without this information, the claimed robustness is not established by the presented evidence.
minor comments (5)
  1. [Abstract and Title] The phrase 'similar tetrahedras' uses an irregular plural; it should be 'similar tetrahedra'.
  2. [Section II.C] The reference to the wood-leaf separation method appears as '[34]' but the reference list only contains 29 entries, and item [28] corresponds to the Sun et al. paper on wood-leaf classification; the citation should be corrected.
  3. [Section III.A, Figures] The figure numbering is inconsistent: Figure 3 is described both as the demonstration of Tree04 preprocessing and as the skeleton generation, and Figure 7 is referenced twice in the text; the captions do not match the descriptions. The paper should be carefully proofread to fix figure references.
  4. [Section II.D] The equations for edge-length comparison, volume ratio, and the LM-ICP update are not numbered, making them awkward to reference; consider adding equation numbers for clarity.
  5. [Section III.B] The text states that for Tree01, the AMRST coarse registration RMSE is 0.08 while NDT is 3.83 and ICP is 0.65, and after fine registration RMSE is 0.06 and HD is 0.03; these values should be cross-checked against Table 2 because the table is not shown in the manuscript text and appears to have some formatting issues.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the registration pipeline and its evaluation are not derived by construction from their inputs.

full rationale

After walking the derivation chain—skeleton construction, key-point extraction, tetrahedron matching, coarse alignment, LM-ICP fine registration, and error reporting—I find no step in which a claimed prediction or first-principles result is equivalent by construction to its inputs. The AMRST pipeline computes its transformation from matched tetrahedron vertices and then refines with LM-ICP; this is a fitting algorithm, but the paper does not present the resulting RMSE/HD values as predictions derived from those fits. The tolerance parameters ε, β, δ and the depth threshold are hand-set and affect matching quality, yet they are not fitted to the reported accuracy numbers, so the 'hand-tuned on test data' concern is a generalization or overfitting risk, not circularity. The RMSE and Hausdorff formulas in Section III.B are generic and underspecified—the paper never states which point sets or correspondences are used, making the accuracy claim hard to reproduce—but underspecification is not circularity unless the metric is shown to be computed from the algorithm's own correspondences, which the text does not establish. Citation [28] (wood–leaf separation) is self-citation by a co-author, but it is used as a preprocessing component with visual results, not as the proof of the central registration claim; therefore it is not load-bearing in the circular sense. No uniqueness theorem, renamed-known-result, or ansatz-smuggled-in-via-citation pattern appears. Verdict: no significant circularity; score 0. The main validity risk is external (missing ground-truth registration protocol and parameter tuning), not circular derivation.

Assumptions & free parameters 4 free parameters · 7 assumptions · 0 invented entities

The method introduces several hand-tuned thresholds (epsilon, beta, delta, depth threshold) and relies on prior wood-leaf separation and skeleton extraction without providing code or data. No new physical entities are introduced.

free parameters (4)
  • epsilon (edge length tolerance) = not reported (recommended 0.01 to 0.1)
    Controls which edge lengths are treated as equal when filtering candidate tetrahedra. The paper gives a recommended interval but does not state the value used in the experiments, suggesting it was tuned on the test data.
  • beta (volume ratio tolerance) = not reported (upper limit 10%)
    Threshold for accepting tetrahedra with similar volumes before edge-length matching. Exact value not given; the paper says upper limit 10% but not the actual setting.
  • delta (SVD reconstruction error threshold) = not reported (upper limit 0.1)
    Threshold for validating tetrahedron congruence after SVD. Exact value not given, likely tuned to achieve the reported accuracy.
  • depth threshold for key point selection = 5 nodes
    The method selects the first five nodes along the path from root to each endpoint as key points. This number is chosen by the authors and is a free parameter affecting key point set size and robustness.
assumptions (7)
  • standard math Tetrahedron congruence theorem: two tetrahedra with equal corresponding edge lengths are congruent.
    Used in Section II-D.2 to justify matching tetrahedra based on edge lengths. Standard Euclidean geometry.
  • standard math SVD provides a unique rigid transformation between matched point pairs.
    Used in Section II-D.2 to compute the coarse transformation. Standard linear algebra.
  • domain assumption MST-based skeleton extraction (from [35]) yields a branching structure that represents the true tree topology.
    Skeleton generation is cited to [35]; the paper assumes this skeleton accurately reflects branch structure. If the skeleton is noisy, key points and tetrahedra matching fail.
  • domain assumption Wood-leaf separation (from [28]) reliably separates branch points from leaf points.
    Preprocessing relies on the cited method; misclassification would remove or add branch points, corrupting the skeleton and key points.
  • domain assumption The branch structure of the same tree is consistent across the two scans (viewpoint invariance).
    The coarse registration assumes that key points and tetrahedra appear in both scans with similar geometry. Occlusion or differing viewpoints could break this.
  • ad hoc to paper The depth threshold of five nodes yields stable and repeatable key points across scans.
    The number 5 is chosen by the authors without evidence; the paper states it 'helps reduce data noise' but gives no analysis. This is a paper-specific heuristic.
  • ad hoc to paper Using leaf point clouds for fine registration is more effective than using branch point clouds.
    The paper says 'based on experimental testing' (no details) that leaf point clouds are preferred. This is a design choice particular to this work.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Automatic marker-free registration based on similar tetrahedras for single-tree point clouds." pith.science (2026). https://pith.science/paper/O2JE53NV

@misc{pith2026241113069,
  author       = {Pith},
  title        = {Pith review of: Automatic marker-free registration based on similar tetrahedras for single-tree point clouds},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/O2JE53NV}},
  note         = {Machine review of arXiv:2411.13069}
}
read the original abstract

In recent years, terrestrial laser scanning technology has been widely used to collect tree point cloud data, aiding in measurements of diameter at breast height, biomass, and other forestry survey data. Since a single scan from terrestrial laser systems captures data from only one angle, multiple scans must be registered and fused to obtain complete tree point cloud data. This paper proposes a marker-free automatic registration method for single-tree point clouds based on similar tetrahedras. First, two point clouds from two scans of the same tree are used to generate tree skeletons, and key point sets are constructed from these skeletons. Tetrahedra are then filtered and matched according to similarity principles, with the vertices of these two matched tetrahedras selected as matching point pairs, thus completing the coarse registration of the point clouds from the two scans. Subsequently, the ICP method is applied to the coarse-registered leaf point clouds to obtain fine registration parameters, completing the precise registration of the two tree point clouds. Experiments were conducted using terrestrial laser scanning data from eight trees, each from different species and with varying shapes. The proposed method was evaluated using RMSE and Hausdorff distance, compared against the traditional ICP and NDT methods. The experimental results demonstrate that the proposed method significantly outperforms both ICP and NDT in registration accuracy, achieving speeds up to 593 times and 113 times faster than ICP and NDT, respectively. In summary, the proposed method shows good robustness in single-tree point cloud registration, with significant advantages in accuracy and speed compared to traditional ICP and NDT methods, indicating excellent application prospects in practical registration scenarios.

Figures

Figures reproduced from arXiv: 2411.13069 by the authors.

Figure 1
Figure 1. Demonstration of eight tree morphology B. Experimental Methods The detailed workflow of the proposed tree point cloud registration method is shown in [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. First, the original tree point cloud is subjected to wood-leaf separation, dividing it into branch point clouds and leaf point clouds. Using the separated branch point clouds, a skeleton point cloud of the tree is constructed, capturing the core structural features of the tree. Next, key branch points and end points are identified in the skeleton point cloud, and an initial transformation matrix between the two skel… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

28 extracted references · 28 canonical work pages

  1. [1]

    International benchmarking of terrestrial laser scanning approaches for forest inventories,

    X. Liang, J. Hyyppä , H. Kaartinen, et al., "International benchmarking of terrestrial laser scanning approaches for forest inventories," ISPRS J. Photogramm. Remote Sens., vol. 144, pp. 137-179, 2018

    work page 2018

  2. [2]

    Individual tree biomass estimation using terrestrial laser scanning,

    V. Kankare, M. Holopainen, M. Vastaranta, et al., "Individual tree biomass estimation using terrestrial laser scanning," ISPRS J. Photogramm. Remote Sens., vol. 75, pp. 64-75, 2013

    work page 2013

  3. [3]

    Assessing branching structure for biomass and wood quality estimation using terrestrial laser scanning point clouds,

    J. Pyö rä lä , X. Liang, N. Saarinen, et al., "Assessing branching structure for biomass and wood quality estimation using terrestrial laser scanning point clouds," Can. J. Remote Sens., vol. 44, no. 5, pp. 462-475, 2018

    work page 2018

  4. [4]

    Deriving merchantable volume in poplar through a localized tapering function from non-destructive terrestrial laser scanning,

    Y. Sun, X. Liang, Z. Liang, et al., "Deriving merchantable volume in poplar through a localized tapering function from non-destructive terrestrial laser scanning," Forests, vol. 7, no. 4, p. 87, 2016

    work page 2016

  5. [5]

    Feasibility of terrestrial laser scanning for collecting stem volume information from single trees,

    N. Saarinen, V. Kankare, M. Vastaranta, et al., "Feasibility of terrestrial laser scanning for collecting stem volume information from single trees," ISPRS J. Photogramm. Remote Sens., vol. 123, pp. 140-158, 2017

    work page 2017

  6. [6]

    Automatic stem mapping using single -scan terrestrial laser scanning,

    X. Liang, P. Litkey, J. Hyyppa, et al., "Automatic stem mapping using single -scan terrestrial laser scanning," IEEE Trans. Geosci. Remote Sens. , vol. 50, no. 2, pp. 661-670, 2011

    work page 2011

  7. [7]

    Automatic forest mapping at individual tree levels from terrestrial laser scanning point clouds with a hierarchical minimum cut method,

    B. Yang, W. Dai, Z. Dong, et al., "Automatic forest mapping at individual tree levels from terrestrial laser scanning point clouds with a hierarchical minimum cut method," 2016

    work page 2016

  8. [8]

    A novel approach for the detection of standing tree stems from plot -level terrestrial laser scanning data,

    W. Zhang, P. Wan, T. Wang, et al., "A novel approach for the detection of standing tree stems from plot -level terrestrial laser scanning data," Remote Sens., vol. 11, no. 2, p. 211, 2019

    work page 2019

Show all 28 references
  1. [9]

    Estimating leaf area distribution in savanna trees from terrestrial LiDAR measurements,

    M. Bé land, J. L. Widlowski, R. A. Fournier, et al., "Estimating leaf area distribution in savanna trees from terrestrial LiDAR measurements," Agric. For. Meteorol., vol. 151, no. 9, pp. 1252-1266, 2011

    2011

  2. [10]

    A density- based approach for leaf area index assessment in a complex forest environment using a terrestrial laser scanner,

    R. Kargar, R. MacKenzie, G. P. Asner, et al., "A density- based approach for leaf area index assessment in a complex forest environment using a terrestrial laser scanner," Remote Sens., vol. 11, no. 15, p. 1791, 2019

    2019

  3. [12]

    Advances in 3D data acquisition and processing for industrial applications,

    Z. M. Bi and L. Wang, "Advances in 3D data acquisition and processing for industrial applications," Rob. Comput. Integr. Manuf., vol. 26, no. 5, pp. 403-413, 2010

    2010

  4. [13]

    Automatic marker -free registration of terrestrial laser scans using reflectance,

    J. Bö hm and S. Becker, "Automatic marker -free registration of terrestrial laser scans using reflectance," in Proc. 8th Conf. Opt. 3D Meas. Tech., Zurich, Switzerland, 2007, pp. 9-12

    2007

  5. [14]

    Accuracy assessment of building models created from laser scanning data,

    Borkowski and G. Jóźków, "Accuracy assessment of building models created from laser scanning data," Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci. , vol. 39, pp. 253-258, 2012

    2012

  6. [15]

    3D is here: Point cloud library (PCL),

    R. B. Rusu and S. Cousins, "3D is here: Point cloud library (PCL)," in Proc. 2011 IEEE Int. Conf. Robot. Autom., 2011, pp. 1-4

    2011

  7. [16]

    Unique signatures of histograms for local surface description,

    F. Tombari, S. Salti, and L. Di Stefano, "Unique signatures of histograms for local surface description," in Proc. ECCV, Heraklion, Greece, 2010, pp. 356-369

    2010

  8. [17]

    Fast point feature histograms (FPFH) for 3D registration,

    R. B. Rusu, N. Blodow, and M. Beetz, "Fast point feature histograms (FPFH) for 3D registration," in Proc. 2009 IEEE Int. Conf. Robot. Autom., 2009, pp. 3212-3217

    2009

  9. [18]

    Efficient variants of the ICP algorithm,

    S. Rusinkiewicz and M. Levoy, "Efficient variants of the ICP algorithm," in Proc. 3rd Int. Conf. 3 -D Digital Imaging and Modeling, 2001, pp. 145-152

    2001

  10. [19]

    Method for registration of 3-D shapes,

    P. J. Besl and N. D. McKay, "Method for registration of 3-D shapes," in Sensor Fusion IV: Control Paradigms and Data Structures, vol. 1611, SPIE, 1992, pp. 586-606

    1992

  11. [20]

    A combined texture-shape descriptor for enhanced 3D feature matching,

    F. Tombari, S. Salti, and L. Di Stefano, "A combined texture-shape descriptor for enhanced 3D feature matching," in Proc. 2011 18th IEEE Int. Conf. Image Process., 2011, pp. 809-812

    2011

  12. [21]

    Automatic stem mapping by merging several terrestrial laser scans at the feature and decision levels,

    X. Liang and J. Hyyppä , "Automatic stem mapping by merging several terrestrial laser scans at the feature and decision levels," Sensors, vol. 13, no. 2, pp. 1614 -1634, 2013

    2013

  13. [22]

    Automated matching of multiple terrestrial laser scans for stem mapping without the use of artificial references,

    J. Liu, X. Liang, J. Hyyppä , et al., "Automated matching of multiple terrestrial laser scans for stem mapping without the use of artificial references," Int. J. Appl. Earth Obs. Geoinf., vol. 56, pp. 13-23, 2017

    2017

  14. [23]

    Marker - free registration of forest terrestrial laser scanner data pairs with embedded confidence metrics,

    D. Kelbe, J. Van Aardt, P. Romanczyk, et al., "Marker - free registration of forest terrestrial laser scanner data pairs with embedded confidence metrics," IEEE Trans. Geosci. Remote Sens., vol. 54, no. 7, pp. 4314-4330, 2016

    2016

  15. [24]

    Global registration of multiview unordered forest point clouds guided by common subgraphs,

    X. Ge, Q. Zhu, L. Huang, et al., "Global registration of multiview unordered forest point clouds guided by common subgraphs," IEEE Trans. Geosci. Remote Sens., vol. 60, pp. 1-14, 2021

    2021

  16. [25]

    GlobalMatch: Registration of forest terrestrial point clouds by global matching of relative stem positions,

    X. Wang, Z. Yang, X. Cheng, et al., "GlobalMatch: Registration of forest terrestrial point clouds by global matching of relative stem positions," ISPRS J. Photogramm. Remote Sens., vol. 197, pp. 71-86, 2023

    2023

  17. [26]

    Localized registration of point clouds of botanic trees,

    Bucksch and K. Khoshelham, "Localized registration of point clouds of botanic trees," IEEE Geosci. Remote Sens. Lett., vol. 10, no. 3, pp. 631-635, 2012

    2012

  18. [27]

    Automatic registration of tree point clouds from terrestrial LiDAR scanning for reconstructing the ground scene of vegetated surfaces,

    G. Zhou, B. Wang, and J. Zhou, "Automatic registration of tree point clouds from terrestrial LiDAR scanning for reconstructing the ground scene of vegetated surfaces," IEEE Geosci. Remote Sens. Lett., vol. 11, no. 9, pp. 1654- 1658, 2014

    2014

  19. [28]

    Wood–leaf classification of tree point cloud based on intensity and geometric information,

    J. Sun, P. Wang, Z. Gao, et al., "Wood–leaf classification of tree point cloud based on intensity and geometric information," Remote Sens., vol. 13, no. 20, p. 4050, 2021

    2021

  20. [29]

    AdTree: Accurate, detailed, and automatic modelling of laser - scanned trees,

    S. Du, R. Lindenbergh, H. Ledoux, et al., "AdTree: Accurate, detailed, and automatic modelling of laser - scanned trees," Remote Sens., vol. 11, no. 18, p. 2074, 2019

    2019

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.