Pith. sign in

REVIEW 4 major objections 6 minor 35 references

Decision PCR: Decision version of the Point Cloud Registration task

T0 review · 4 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Judging whether a 3D alignment is correct can be learned, pushing recall to 86.97 percent.

desk verdict The Decision PCR idea is real and the dataset/classifier work is credible, but the headline SOTA gain is confounded by a change in both candidate pool and selection metric, so the paper needs one decisive ablation before the central claim holds. read the letter →

arxiv 2507.14965 v1 pith:HMQ7XR3Z submitted 2025-07-20 cs.CV

Yaojie Zhang , Tianlun Huang , Weijun Wang , Wei Feng This is my paper · ORCID
classification cs.CV
keywords DecisionPCRpointcloudregistrationevaluationlow-overlapKPConv3DLoMatchgeneralizationclassification
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

This paper argues that the hardest part of low-overlap point cloud registration is not generating alignment hypotheses but deciding which of them are actually correct. It isolates that judgment as a standalone task, called Decision PCR, and trains a binary point-cloud classifier to look at a merged cloud and say whether the alignment that produced it is correct or wrong. The classifier is then used as an evaluation layer inside established registration pipelines, replacing handcrafted scores such as Maximum Inlier Count when selecting among candidate transformations. The authors report that this raises registration recall on the hard 3DLoMatch benchmark to 86.97% when paired with GeoTransformer, and that the same trained model transfers to the unseen outdoor ETH dataset with an average recall of 98.88% for GeoTransformer features. If the claim holds, correctness judgment in 3D alignment is a learnable, plug-in capability rather than a metric that has to be redesigned for each pipeline.

What carries the argument

The load-bearing object is a learned judge: a binary point-cloud classifier that inputs the merged cloud $\{P', Q\}$ produced by applying a candidate transformation to the source cloud, with every point tagged source or target, and outputs a softmax score for the probability that the transformation is correct. The paper builds the dataset by intentionally generating hard negatives with FPFH and SC2 on 3DMatch pairs with over 10% overlap, so the judge is trained on near-miss failures rather than random misalignments. In deployment, the same judge scores every candidate transformation that passes the sight-view constraint, selects the highest-scoring one, and supports early truncation when a score exceeds a threshold, yielding both a choice and an uncertainty-aware confidence.

What would settle it

Run the reported pipeline on ETH with the input scale factor fixed without looking at the ETH test set—for example, taken from the average point spacing of the indoor training scenes—and check whether the 98.88% average registration recall with GeoTransformer features survives; if it drops sharply, the claimed cross-domain generalization depends on test-set tuning. A complementary check is to retrain the judge on negatives generated by GeoTransformer itself and compare its accuracy on GeoTransformer-derived hypotheses against the FPFH-trained judge.

Watch

Extended reading notes

Core claim

The paper's central claim is that the decision version of point cloud registration—given a candidate transformation, decide whether the merged point cloud is correctly aligned—is a well-posed, learnable problem and that solving it data-drivenly gives a stronger and more transferable selection criterion than existing rule-based evaluation metrics. To support this, the authors construct a training set from 3DMatch pairs by running the weak FPFH descriptor through the SC2 matcher to generate hundreds of deliberately challenging wrong transformations, classified into four failure types, and then train a KPConv classifier on the merged point clouds with per-point source/target tags. Integrated into the SC2 pipeline as a scoring layer, the learned judge improves registration recall on both 3DMatch and 3DLoMatch across descriptors, with the largest gains on low-overlap pairs; combined with GeoTransformer it reaches a new state-of-the-art 86.97% recall on 3DLoMatch. On the outdoor ETH dataset, applied without retraining but with a rescaled input, the same classifier attains an average registration recall of 98.88% with GeoTransformer features, which the paper reads as evidence that correctness judgment learned indoors generalizes across domains.

Load-bearing premise

The method assumes that the near-miss failures produced by weak FPFH matching on indoor 3DMatch pairs look enough like the failure modes of stronger learned descriptors and of outdoor laser-scan data that a judge trained on the former can reliably score the latter.

Editorial extensions

If this is right

  • Every tested registration pipeline improves when the learned score replaces a rule-based one; on 3DLoMatch with FCGF, recall rises from 58.62% for the SC2 baseline to 71.59% with the decision layer.
  • The decision layer plus GeoTransformer reaches 86.97% registration recall on 3DLoMatch, exceeding prior methods and rivaling a MAC variant that is given ground-truth overlap information.
  • Because the score is computed from the merged geometry itself, the evaluation no longer depends on the quality of the initial correspondences that generated the hypothesis.
  • The output score gives each registration an uncertainty value, not just a final classification, so downstream systems can separate high-confidence alignments from guesses.
  • The same indoor-trained model transfers to the unseen outdoor ETH dataset, reaching up to 98.88% average recall with GeoTransformer features after rescaling the input, indicating that learned correctness judgment is not specific to indoor geometry.

Reading between the lines

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

  • An untested extension follows from the paper's own logic: training the judge on negative examples produced by the exact descriptor and matcher used in deployment should improve selection further, since the paper only ever trains on FPFH/SC2 failures.
  • The same 'is this merged scene coherent?' verdict could serve adjacent problems such as loop-closure verification in SLAM or temporal consistency checks in multi-view reconstruction, where a learned correctness score could replace handcrafted pairwise thresholds.
  • The ETH experiment leaves an open question about scale: the scale factor 0.4 is selected by scanning the test set itself, so a stricter generalization test would fix that factor from training data alone and check whether the 98.88% average recall holds.
  • A falsifiable prediction implicit in the approach is that classifier accuracy tracks the visual distinctiveness of the scene: near-symmetric rooms or repetitive geometry should produce systematic wrong-alignment positives that the current training set may under-sample.
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 / 6 minor

Summary. The paper proposes the 'Decision PCR' task: given two point clouds and a set of transformation hypotheses, decide whether each transformation is correct. The authors construct a training dataset from 3DMatch by using FPFH features and the SC2 registration method to generate 'challenging wrong' transformations, then train a KPConv-based binary classifier that scores a merged point cloud. The classifier is inserted into a generation-evaluation pipeline: SC2 produces up to m=100 hypotheses, the Sight View Constraint (SVC) filters obviously wrong ones, and the learned scorer selects the final transformation with an early-truncation threshold. The paper reports improved registration recall on 3DMatch and 3DLoMatch (e.g., 86.97% with GeoTransformer on 3DLoMatch) and strong results on the outdoor ETH dataset, and argues that a learned correctness evaluator is more reliable than rule-based metrics such as Maximum Inlier Count.

Significance. The conceptual reframing of registration evaluation as a binary 'Decision PCR' classification problem is a useful and, to my knowledge, novel viewpoint. The dataset construction procedure and the integration of a learned scorer as an evaluation layer are sensible and potentially impactful, especially for low-overlap scenarios where inlier-count-based metrics degrade. If the reported gains were cleanly attributable to the learned Decision PCR model, this would be a solid contribution to the 3D registration literature. The paper also conducts experiments across multiple descriptors and an outdoor dataset, which strengthens the scope of the evaluation. However, the current experiments do not yet isolate the contribution of the learned scorer from the enlarged hypothesis pool, and the ETH 'generalization' experiment involves test-set tuning of a scale parameter. These issues must be addressed before the headline performance claims can be accepted.

major comments (4)
  1. [Sec. 3.4, Algorithm 1, and Table 5] The central SOTA claim is confounded: the comparison between 'SC2' and 'Ours' changes both the selection metric and the size of the candidate pool. Algorithm 1 searches up to m=100 SC2 hypotheses after SVC filtering, whereas the SC2 baseline returns the single transformation selected by SC2's own criterion. Table 5 shows that with GeoTransformer and SVC the Top-100 RR is 92.31%, against an SC2 baseline of 78.11%; hence the expanded pool alone provides a ceiling large enough to explain most of the 8.86-point gap, and the reported Ours result (86.97%) lies below that ceiling. Please add an ablation that applies the original SC2/MIC selection criterion to the exact same SVC-filtered Top-100 hypothesis list. Without this control, the abstract's statement that combining the framework with GeoTransformer 'achieves a new SOTA registration recall' is not attributable to the learned Decision PCR classifier.
  2. [Sec. 4.3.1 and Fig. 6] The ETH generalization experiment tunes a free parameter on the test set. The paper states that the scale factor is varied and the value giving optimal RR (scale=0.4) is adopted for Table 4; this means the 'unseen outdoor ETH dataset' claim does not come from a fully untuned model. Please report ETH results with a scale factor fixed without reference to ETH test RR (e.g., chosen on a validation split or by a principled rule), or at minimum clearly label scale=0.4 as a test-set-tuned hyperparameter and provide sensitivity curves. Note also that Sec. 4.3.1 sets m=30 while Sec. 4.3.2 uses m=100; the reason for this inconsistency should be clarified.
  3. [Sec. 3.2 and Sec. 4.2] The training negatives are generated by SC2 with FPFH features on 3DMatch, while evaluation uses SC2 hypotheses produced from FCGF, GeoTransformer, PREDATOR, and PEAL features. The classifier may therefore learn SC2-specific failure signatures rather than a general correctness model, and the reported gains may not transfer to other hypothesis generators. Please provide an experiment in which the learned scorer is applied to hypotheses produced by a different generator (e.g., RANSAC or MAC) without retraining, or an analysis of whether the score transfers across failure modes. This is important for the paper's claim that Decision PCR is a foundational evaluation task rather than an SC2-specific post-processor.
  4. [Sec. 4 and Tables 2-4] All reported RR values are point estimates with no error bars, repeated runs, or statistical significance tests. Because the pipeline includes stochastic training of the classifier and threshold-based selection, it would be useful to report variance across at least a few seeds or to state that all experiments are deterministic. This is particularly relevant for the small differences over MAC-OP on some 3DLoMatch rows, where the claimed superiority could be within run-to-run variation.
minor comments (6)
  1. [Table 5] The header 'T op10 T op50 T op100 RR RR RR RR' is confusing: it is not clear which columns are Top-10/Top-50/Top-100 and which is the final RR, especially because some rows appear non-monotonic under a natural reading. Please reformat the table with clearly separated columns and a caption explaining the metric definitions.
  2. [Table 3 caption] The caption reads 'Comparison of different PCR Methods on ETH dataset,' but the table reports 3DLoMatch results. This appears to be a copy-paste error and should be corrected.
  3. [Sec. 4.2.3] The sentence 'As shown in Table 2' should refer to Table 3, since the described results (PREDATOR, GeoTransformer, PEAL on 3DLoMatch) are in Table 3.
  4. [Eq. (4)] Please define the operator 'acos' and ensure the argument is clamped to [-1, 1] numerically; also clarify that RE and TE are the standard evaluation metrics rather than part of the training loss.
  5. [Sec. 4.4.1] The definition of Top-m RR in Eq. (6) uses '1(top m hypotheses)', but the notation is ambiguous about whether the top-m set is ordered by the method's own score or by ground-truth error. Please define the ordering explicitly.
  6. [Throughout] There are several typos and grammatical issues, e.g., 'To be consistency with the indoor setting' in Sec. 4.3.2 and 'the decision version of the PCR problem' inconsistencies in capitalization. A careful proofreading pass is recommended.

Circularity Check

1 steps flagged · score 3.0 of 10

Partial circularity in the ETH generalization claim via test-set-tuned scale parameter; core 3DLoMatch result is not circular.

  1. fitted input called prediction [Sec. 4.3.1 (Eq. 5, Fig. 6) and Sec. 4.3.2 (Tab. 4)]
    "We resolve this by simply scaling the merged point clouds while preserving outdoor sampling sizes. Under this condition, the Score can be computed as: Score = F (scale ∗ {P′, Q}) (5) We set the m = 30and threshold = 0.6 for the Alg.1. By varying the scale parameter, we establish the relationship between Registration Recall (RR) and scale, as depicted in the Fig.6. Baseline performance (without Decision PCR) serves as reference. The optimal RR occurs at scale = 0.4, which is adopted for the following experiment."

    Eq. (5) makes 'scale' an input to the Decision PCR scorer. The scale is chosen by sweeping RR on the ETH test set (Fig. 6) and taking the argmax ('The optimal RR occurs at scale = 0.4'), and the resulting ETH RRs are then presented as generalization (Tab. 4, 'unseen outdoor ETH dataset'). The same dataset is therefore used both to select the parameter and to produce the reported prediction, so the ETH generalization number is a fitted value, not an independent forecast. The 3DLoMatch SOTA claim (Tab. 3) is not affected by this tuning and remains an external-benchmark result.

full rationale

The Decision PCR classifier itself is not equation-level circular: labels are defined from ground-truth transformation error (Eq. 4), the network is a standard supervised classifier on merged point clouds, and test-time correctness is measured against the same external RR criterion; no output is recycled as an input. The main 3DLoMatch improvement is benchmarked against an external test set and does not reduce to the training labels. The most defensible circularity flag is the ETH generalization claim, where the scale parameter in Eq. (5) is tuned on the ETH test set and then ETH RR is reported as evidence of generalization. A separate, non-circular validity threat is that the classifier is trained on negatives produced by the same SC2 generator used at inference, and Table 5 shows the Top-100 hypothesis pool alone explains part of the SOTA gap; this is an experimental-design/attribution concern rather than a definitional circularity. Self-citations (SVC [34]) are used as an ablated module, not as an imported uniqueness theorem.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The method adds no new physical entities; its free parameters are mostly algorithm hyperparameters, the most concerning being the ETH scale chosen on the test set. The key domain assumptions are the representativeness of FPFH/SC2-generated negatives and the adequacy of geometric plausibility as a correctness signal.

free parameters (6)
  • score threshold = 0.6
    Used in Alg.1 to accept a transformation; effect on RR shown in Fig.8 but the value is set by hand and the RR measurements are on the same test benchmarks.
  • m (max hypotheses checked) = 100 (30 for scale experiment)
    Truncation length in Alg.1; RR stabilizes for m>60 on 3DLoMatch (Fig.8), and 30 is used in the ETH scale sweep.
  • scale factor on ETH = 0.4
    Applied to merged outdoor point clouds (Eq.5); chosen by scanning RR vs scale on the ETH test set itself (Fig.6), i.e., fitted to the test data.
  • voxel downsampling size = 5 cm
    Point cloud resolution for training/inference in indoor settings; correctness of the learned score is sensitive to it (Sec.4.3.1).
  • dataset overlap threshold = >10%
    Pairs with over 10% overlap are used to construct the Decision PCR dataset (Sec.3.2); this cutoff shapes the difficulty of the negatives.
  • negative generator descriptor = FPFH
    Wrong transformations are generated with the FPFH descriptor (Sec.3.2); distribution of negatives depends on this choice.
assumptions (4)
  • domain assumption Merged-point-cloud plausibility is a valid cue for alignment correctness
    The whole Decision PCR model assumes that visual/geometric consistency of the merged cloud carries enough signal to separate correct from wrong transformations; repetitive or featureless geometry can violate this (implicit throughout Sec.3.3).
  • domain assumption SC2 hypotheses with FPFH features are representative hard negatives for other descriptors and datasets
    The training set is built exclusively from FPFH+SC2 hypotheses on 3DMatch pairs; generalization to FCGF/GeoTransformer features and to ETH is assumed (Sec.3.2, Sec.4.3).
  • domain assumption Correctness threshold (15 deg, 30 cm) is the right binary label
    Labels in the dataset are based on the standard registration error threshold (Sec.4.1.1); the classifier inherits any ambiguity near this boundary.
  • domain assumption Standard 3DMatch train/test scene split is respected
    The classifier is trained on the 3DMatch train set and evaluated on the 3DMatch test set; the paper does not state that no training-scene point clouds leak into the test hypotheses via pretrained descriptors.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Decision PCR: Decision version of the Point Cloud Registration task." pith.science (2026). https://pith.science/paper/HMQ7XR3Z

@misc{pith2026250714965,
  author       = {Pith},
  title        = {Pith review of: Decision PCR: Decision version of the Point Cloud Registration task},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HMQ7XR3Z}},
  note         = {Machine review of arXiv:2507.14965}
}
read the original abstract

Low-overlap point cloud registration (PCR) remains a significant challenge in 3D vision. Traditional evaluation metrics, such as Maximum Inlier Count, become ineffective under extremely low inlier ratios. In this paper, we revisit the registration result evaluation problem and identify the Decision version of the PCR task as the fundamental problem. To address this Decision PCR task, we propose a data-driven approach. First, we construct a corresponding dataset based on the 3DMatch dataset. Then, a deep learning-based classifier is trained to reliably assess registration quality, overcoming the limitations of traditional metrics. To our knowledge, this is the first comprehensive study to address this task through a deep learning framework. We incorporate this classifier into standard PCR pipelines. When integrated with our approach, existing state-of-the-art PCR methods exhibit significantly enhanced registration performance. For example, combining our framework with GeoTransformer achieves a new SOTA registration recall of 86.97\% on the challenging 3DLoMatch benchmark. Our method also demonstrates strong generalization capabilities on the unseen outdoor ETH dataset.

Figures

Figures reproduced from arXiv: 2507.14965 by the authors.

Figure 1
Figure 1. The decision version of the PCR task aims to evaluate [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The pipeline of dataset construction. Existing PCR method [ [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The pipeline of model training [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: PCR pipeline based on Decision PCR model. (1) Hypotheses and input point clouds are used to generate merged point clouds. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Confusion matrix of FCGF registration result on 3DLo [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: RR vs scale on ETH dataset with FCGF [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Registration Recall under different inlier ratio [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Parameter experiments. For the Fig.8a, the m is set as 30. For the Fig.8b, the Score threshold is set as 0.6. 5. Conclusion This work establishes the Decision PCR task as the foundational basis for robust point cloud registration [PITH_FULL_IMAGE:figures/full_fig_p008…
Figure 9
Figure 9. Figure 9: Visualization Results on 3DLoMatch. Red and green colors indicate failed and successful registrations, respectively. The [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]
Figure 10
Figure 10. Figure 10: Visualization Results on ETH dataset. Red and green colors indicate failed and successful registrations, respectively. The [PITH_FULL_IMAGE:figures/full_fig_p012_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 35 canonical work pages

  1. [1]

    Spinnet: Learning a gen- eral surface descriptor for 3d point cloud registration

    Sheng Ao, Qingyong Hu, Bo Yang, Andrew Markham, and Yulan Guo. Spinnet: Learning a gen- eral surface descriptor for 3d point cloud registration. In Proceedings of the IEEE/CVF conference on com- puter vision and pattern recognition , pages 11753– 11762, 2021. 2, 7, 8

    work page 2021

  2. [2]

    Pointnetlk: Robust & ef- ficient point cloud registration using pointnet

    Yasuhiro Aoki, Hunter Goforth, Rangaprasad Arun Srivatsan, and Simon Lucey. Pointnetlk: Robust & ef- ficient point cloud registration using pointnet. In Pro- ceedings of the IEEE/CVF conference on computer vi- sion and pattern recognition, pages 7163–7172, 2019. 4

    work page 2019

  3. [3]

    D3feat: Joint learning of dense detection and description of 3d local fea- tures

    Xuyang Bai, Zixin Luo, Lei Zhou, Hongbo Fu, Long Quan, and Chiew-Lan Tai. D3feat: Joint learning of dense detection and description of 3d local fea- tures. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 6359– 6367, 2020. 2

    work page 2020

  4. [4]

    Pointdsc: Robust point cloud registration using deep spatial consistency

    Xuyang Bai, Zixin Luo, Lei Zhou, Hongkai Chen, Lei Li, Zeyu Hu, Hongbo Fu, and Chiew-Lan Tai. Pointdsc: Robust point cloud registration using deep spatial consistency. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion, pages 15859–15869, 2021. 2, 5, 6

    work page 2021

  5. [5]

    Graph-cut ransac

    Daniel Barath and Ji ˇr´ı Matas. Graph-cut ransac. In Proceedings of the IEEE conference on computer vi- sion and pattern recognition, pages 6733–6741, 2018. 2, 5, 6

    work page 2018

  6. [6]

    Magsac: marginalizing sample consensus

    Daniel Barath, Jiri Matas, and Jana Noskova. Magsac: marginalizing sample consensus. In Proceedings of the IEEE/CVF conference on computer vision and pat- tern recognition, pages 10197–10205, 2019. 2

    work page 2019

  7. [7]

    Sc2- pcr: A second order spatial compatibility for efficient and robust point cloud registration

    Zhi Chen, Kun Sun, Fan Yang, and Wenbing Tao. Sc2- pcr: A second order spatial compatibility for efficient and robust point cloud registration. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 13221–13231, 2022. 1, 2, 3, 4, 5, 6

    work page 2022

  8. [8]

    SC 2-PCR++: Rethinking the generation and se- lection for efficient and robust point cloud registration

    Zhi Chen, Kun Sun, Fan Yang, Lin Guo, and Wenbing Tao. SC 2-PCR++: Rethinking the generation and se- lection for efficient and robust point cloud registration. IEEE Transactions on pattern analysis and machine intelligence, 2023. 1, 2, 5, 6

    work page 2023

Show all 35 references
  1. [9]

    Fully convolutional geometric features

    Christopher Choy, Jaesik Park, and Vladlen Koltun. Fully convolutional geometric features. In Proceed- ings of the IEEE/CVF international conference on computer vision, pages 8958–8966, 2019. 2, 5, 7, 8

    2019

  2. [10]

    Deep global registration

    Christopher Choy, Wei Dong, and Vladlen Koltun. Deep global registration. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 2514–2523, 2020. 5, 6

    2020

  3. [11]

    Matching with prosac- progressive sample consensus

    Ondrej Chum and Jiri Matas. Matching with prosac- progressive sample consensus. In 2005 IEEE com- puter society conference on computer vision and pat- tern recognition (CVPR’05) , pages 220–226. IEEE,

    2005

  4. [12]

    Random sam- ple consensus: a paradigm for model fitting with ap- plications to image analysis and automated cartogra- phy

    Martin A Fischler and Robert C Bolles. Random sam- ple consensus: a paradigm for model fitting with ap- plications to image analysis and automated cartogra- phy. Communications of the ACM , 24(6):381–395,

  5. [13]

    Vision meets robotics: The kitti dataset

    Andreas Geiger, Philip Lenz, Christoph Stiller, and Raquel Urtasun. Vision meets robotics: The kitti dataset. The International Journal of Robotics Re- search, 32(11):1231–1237, 2013. 3

    2013

  6. [14]

    Predator: Reg- istration of 3d point clouds with low overlap

    Shengyu Huang, Zan Gojcic, Mikhail Usvyatsov, An- dreas Wieser, and Konrad Schindler. Predator: Reg- istration of 3d point clouds with low overlap. In Proceedings of the IEEE/CVF Conference on com- puter vision and pattern recognition , pages 4267– 4276, 2021. 1, 2, 4, 6

    2021

  7. [15]

    Robust outlier rejection for 3d registration with variational bayes

    Haobo Jiang, Zheng Dang, Zhen Wei, Jin Xie, Jian Yang, and Mathieu Salzmann. Robust outlier rejection for 3d registration with variational bayes. In Proceed- ings of the IEEE/CVF conference on computer vision and pattern recognition, pages 1148–1157, 2023. 5, 6

    2023

  8. [16]

    Deep hough voting for robust global registra- tion

    Junha Lee, Seungwook Kim, Minsu Cho, and Jaesik Park. Deep hough voting for robust global registra- tion. In Proceedings of the IEEE/CVF international conference on computer vision , pages 15994–16003,

  9. [17]

    A spectral technique for correspondence problems using pair- wise constraints

    Marius Leordeanu and Martial Hebert. A spectral technique for correspondence problems using pair- wise constraints. In Tenth IEEE International Confer- ence on Computer Vision (ICCV’05) Volume 1, pages 1482–1489. IEEE, 2005. 5

    2005

  10. [18]

    Challenging data sets for point cloud registration algorithms

    Franc ¸ois Pomerleau, Ming Liu, Francis Colas, and Roland Siegwart. Challenging data sets for point cloud registration algorithms. The International Journal of Robotics Research, 31(14):1705–1711, 2012. 4, 7

    2012

  11. [19]

    Geometric transformer for fast and robust point cloud registration

    Zheng Qin, Hao Yu, Changjian Wang, Yulan Guo, Yuxing Peng, and Kai Xu. Geometric transformer for fast and robust point cloud registration. In Proceed- ings of the IEEE/CVF conference on computer vision and pattern recognition, pages 11143–11152, 2022. 2, 6, 7, 8

    2022

  12. [20]

    Geotransformer: Fast and robust point cloud registra- tion with geometric transformer

    Zheng Qin, Hao Yu, Changjian Wang, Yulan Guo, Yuxing Peng, Slobodan Ilic, Dewen Hu, and Kai Xu. Geotransformer: Fast and robust point cloud registra- tion with geometric transformer. IEEE Transactions on Pattern Analysis and Machine Intelligence , 45(8): 9806–9821, 2023. 8

    2023

  13. [21]

    Compatibility-guided sampling consensus for 3-d point cloud registration

    Siwen Quan and Jiaqi Yang. Compatibility-guided sampling consensus for 3-d point cloud registration. IEEE Transactions on Geoscience and Remote Sens- ing, 58(10):7380–7392, 2020. 2

    2020

  14. [22]

    Fast point feature histograms (fpfh) for 3d registration

    Radu Bogdan Rusu, Nico Blodow, and Michael Beetz. Fast point feature histograms (fpfh) for 3d registration. In 2009 IEEE international conference on robotics and automation , pages 3212–3217. IEEE, 2009. 5, 7, 8

    2009

  15. [23]

    Kpconv: Flexible and deformable convolution for point clouds

    Hugues Thomas, Charles R Qi, Jean-Emmanuel De- schaud, Beatriz Marcotegui, Franc ¸ois Goulette, and Leonidas J Guibas. Kpconv: Flexible and deformable convolution for point clouds. In Proceedings of the IEEE/CVF international conference on computer vi- sion, pages 6411–6420, 2...

    2019

  16. [24]

    Roreg: Pairwise point cloud registration with oriented descriptors and local rota- tions

    Haiping Wang, Yuan Liu, Qingyong Hu, Bing Wang, Jianguo Chen, Zhen Dong, Yulan Guo, Wenping Wang, and Bisheng Yang. Roreg: Pairwise point cloud registration with oriented descriptors and local rota- tions. IEEE Transactions on Pattern Analysis and Ma- chine Intelligence, 45(8)...

    2023

  17. [25]

    3d shapenets: A deep representation for volumetric shapes

    Zhirong Wu, Shuran Song, Aditya Khosla, Fisher Yu, Linguang Zhang, Xiaoou Tang, and Jianxiong Xiao. 3d shapenets: A deep representation for volumetric shapes. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 1912– 1920, 2015. 3

    1912

  18. [26]

    Efficient single correspondence voting for point cloud registration

    Xuejun Xing, Zhengda Lu, Yiqun Wang, and Jun Xiao. Efficient single correspondence voting for point cloud registration. IEEE Transactions on Image Pro- cessing, 2024. 1, 2

    2024

  19. [27]

    Teaser: Fast and certifiable point cloud registration

    Heng Yang, Jingnan Shi, and Luca Carlone. Teaser: Fast and certifiable point cloud registration. IEEE Transactions on Robotics , 37(2):314–333, 2020. 1, 2, 5, 6

    2020

  20. [28]

    Go-icp: A globally optimal solution to 3d icp point-set registration

    Jiaolong Yang, Hongdong Li, Dylan Campbell, and Yunde Jia. Go-icp: A globally optimal solution to 3d icp point-set registration. IEEE transactions on pat- tern analysis and machine intelligence, 38(11):2241– 2254, 2015. 2

    2015

  21. [29]

    Mac: Maximal cliques for 3d registration

    Jiaqi Yang, Xiyu Zhang, Peng Wang, Yulan Guo, Kun Sun, Qiao Wu, Shikun Zhang, and Yanning Zhang. Mac: Maximal cliques for 3d registration. IEEE Transactions on Pattern Analysis and Machine Intelli- gence, 2024. 1, 2, 5, 6, 7

    2024

  22. [30]

    Cofinet: Reliable coarse-to-fine cor- respondences for robust pointcloud registration

    Hao Yu, Fu Li, Mahdi Saleh, Benjamin Busam, and Slobodan Ilic. Cofinet: Reliable coarse-to-fine cor- respondences for robust pointcloud registration. Ad- vances in Neural Information Processing Systems, 34: 23872–23884, 2021. 2

    2021

  23. [31]

    Peal: Prior-embedded explicit attention learning for low-overlap point cloud regis- tration

    Junle Yu, Luwei Ren, Yu Zhang, Wenhui Zhou, Lili Lin, and Guojun Dai. Peal: Prior-embedded explicit attention learning for low-overlap point cloud regis- tration. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 17702–17711, 2023. 2, 6, 8

    2023

  24. [32]

    3dmatch: Learning local geometric descriptors from rgb-d reconstructions

    Andy Zeng, Shuran Song, Matthias Nießner, Matthew Fisher, Jianxiong Xiao, and Thomas Funkhouser. 3dmatch: Learning local geometric descriptors from rgb-d reconstructions. In Proceedings of the IEEE conference on computer vision and pattern recogni- tion, pages 1802–1811, 2017. 3, 4

    2017

  25. [33]

    3d registration with maximal cliques

    Xiyu Zhang, Jiaqi Yang, Shikun Zhang, and Yanning Zhang. 3d registration with maximal cliques. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 17745–17754,

  26. [34]

    Svc: Sight view constraint for robust point cloud registration

    Yaojie Zhang, Weijun Wang, Tianlun Huang, Zhiy- ong Wang, and Wei Feng. Svc: Sight view constraint for robust point cloud registration. Image and Vision Computing, 152:105315, 2024. 1, 2, 4, 5, 8

    2024

  27. [35]

    Fast global registration

    Qian-Yi Zhou, Jaesik Park, and Vladlen Koltun. Fast global registration. In Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Nether- lands, October 11-14, 2016, Proceedings, Part II 14 , pages 766–782. Springer, 2016. 2, 5, 6 Figure 9. Visualization Results ...

    2016

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.