pith. sign in

arxiv: 2602.17517 · v2 · pith:RTDJHAMWnew · submitted 2026-02-19 · 💻 cs.CV

Depth Augmented and FE Free 3D/2D Liver Registration for Laparoscopic Liver AR

Hanyuan Zhang , Lucas He , Runlong He , Weixi Yi , Abdolrahim Kadkhodamohammadi , Danail Stoyanov , Brian R. Davidson , Evangelos B. Mazomenos
show 1 more author
Matthew J. Clarkson
This is my paper

Pith reviewed 2026-05-22 10:53 UTC · model grok-4.3

classification 💻 cs.CV
keywords 3D-2D registrationlaparoscopic liver surgeryaugmented realitystatistical deformation modelmonocular depthnon-rigid ICPsurgical navigation
0
0 comments X

The pith

A depth-augmented pipeline registers 3D liver models to 2D laparoscopic video at 14.73 mm mean error without finite-element models.

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

The paper presents a registration method that first refines rigid pose using monocular depth and multi-class contours, then applies a patient-specific statistical deformation model for non-rigid adjustment. This combination targets the challenges of partial views and tissue movement in laparoscopic liver surgery while avoiding the engineering demands of finite-element simulations. Results on a public clinical dataset show a mean target registration error of 14.73 mm under controlled manual contour conditions. Ablation tests confirm that depth data strengthens the initial alignment step.

Core claim

The pipeline adapts RefineNet for rigid alignment with contour maps and monocular depth, then builds a statistical deformation model from non-rigid ICP correspondences on patient data and optimizes pose plus shape parameters through coarse-to-fine L-BFGS-B to achieve a mean target registration error of 14.73 mm.

What carries the argument

Patient-specific statistical deformation model built from non-rigid ICP correspondences and optimized with L-BFGS-B, augmented by monocular depth for rigid initialization.

If this is right

  • Monocular depth input improves rigid pose estimates compared with contour-only inputs.
  • High-quality surface alignment does not automatically reduce error in locating internal targets such as tumors.
  • The approach supplies an alternative to finite-element pipelines for controlled 3D-2D registration in surgical AR.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same depth-plus-statistical-model pattern could be tested on other soft-tissue organs where patient-specific deformation data can be collected preoperatively.
  • Integration into real-time surgical video streams would require checking whether the coarse-to-fine optimization runs fast enough for live guidance.
  • Collecting deformation statistics across multiple patients might reduce the need for per-case NICP modeling in future versions.

Load-bearing premise

The statistical deformation model built from the patient's NICP data captures enough of the actual intraoperative tissue changes for the optimizer to reach useful alignments.

What would settle it

On new cases where measured tissue deformations fall outside the span of the statistical model, the method would produce target registration errors well above 20 mm or visibly misalign internal structures such as tumors.

Figures

Figures reproduced from arXiv: 2602.17517 by Abdolrahim Kadkhodamohammadi, Brian R. Davidson, Danail Stoyanov, Evangelos B. Mazomenos, Hanyuan Zhang, Lucas He, Matthew J. Clarkson, Runlong He, Weixi Yi.

Figure 1
Figure 1. Figure 1: The registration workflow diagram: The first stage adopts the Refine Net component from PoseFoundation by Wen et al., with the inputs replaced [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: PCA-based liver deformation model. A reference liver mesh [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The registration results of the first-frame images for Patient 1, 3, and 4. The leftmost column shows the input intra-operative image. The second [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
read the original abstract

Augmented reality (AR) guidance in laparoscopic liver surgery requires accurate registration of preoperative 3D models to intraoperative 2D video, but remains challenging due to partial visibility, specularities, and tissue deformation. Existing methods often rely on contour-based rigid initialization and finite-element (FE) models for deformable registration, increasing modeling and engineering complexity. We present a depth-augmented, FE-free 3D--2D registration pipeline that combines robust rigid initialization with patient-specific non-rigid refinement. For rigid alignment, we adapt the RefineNet module of FoundationPose to laparoscopic liver scenes by using multi-class contour maps and monocular depth for relative pose refinement. For deformable alignment, we construct a patient-specific statistical deformation model from non-rigid ICP (NICP) correspondences and optimize pose and shape parameters using a coarse-to-fine L-BFGS-B strategy. On a public clinical laparoscopic liver dataset, the proposed method achieves a mean target registration error (TRE) of 14.73\,mm under a controlled manual-contour setting designed to isolate registration performance. Ablation studies show that monocular depth improves rigid initialization over contour-only inputs, while tumor-mapping analysis indicates that good surface alignment does not necessarily translate into lower target localization error. On an external dataset without ground truth, the method produces visually plausible overlays for qualitative assessment. These results suggest that depth-augmented pose refinement and FE-free statistical deformation modeling provide a promising alternative to FE-based pipelines for controlled 3D--2D liver registration in surgical AR.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper proposes a depth-augmented, finite-element-free pipeline for 3D/2D registration in laparoscopic liver AR. Rigid initialization adapts the RefineNet module from FoundationPose using multi-class contour maps and monocular depth. Deformable alignment builds a patient-specific statistical deformation model from NICP correspondences on the same data and optimizes pose plus shape parameters via coarse-to-fine L-BFGS-B. On a public clinical laparoscopic liver dataset under controlled manual-contour conditions, the method reports a mean target registration error of 14.73 mm, with ablations showing benefit from depth and a tumor-mapping analysis indicating that surface alignment does not guarantee lower target error. Qualitative results are shown on an external dataset without ground truth.

Significance. If the reported TRE holds under more realistic automatic contour extraction, the approach would provide a lower-complexity alternative to FE-based deformable registration for surgical AR. The use of an external FoundationPose module for rigid stage and independent TRE metric on clinical data are strengths; however, the 14.73 mm mean error and lack of statistical tests or error bars limit immediate clinical claims. The patient-specific NICP-derived model and ablation results on depth are concrete contributions that could be built upon if the deformation basis is shown to generalize.

major comments (2)
  1. [Abstract and Methods (statistical deformation model)] Abstract (deformable alignment paragraph) and Methods section on statistical model construction: the patient-specific statistical deformation model is derived directly from NICP correspondences on the same patient data used for testing and evaluation. No information is given on the number of input shapes, percentage of variance explained by the principal modes, regularization strength, or any held-out reconstruction validation. This is load-bearing for the central claim that L-BFGS-B optimization of pose and shape parameters yields the reported 14.73 mm mean TRE, because an incomplete or biased basis could produce low surface error while leaving target error high, consistent with the tumor-mapping observations already noted in the abstract.
  2. [Experiments (evaluation and ablation)] Experiments section (evaluation protocol): the quantitative results are obtained under a controlled manual-contour setting that isolates registration performance. The manuscript does not report how performance degrades when contours are extracted automatically, nor does it provide error bars or statistical significance tests on the 14.73 mm TRE. This weakens the strength of the claim that the pipeline is ready for laparoscopic AR under realistic intraoperative conditions.
minor comments (2)
  1. [Abstract and Experiments (ablations)] The abstract states that monocular depth improves rigid initialization over contour-only inputs, but the corresponding ablation table or figure should explicitly report the rigid-stage TRE or rotation/translation errors for both conditions to allow direct comparison.
  2. [Methods] Notation for the statistical model parameters (pose and shape coefficients) and the L-BFGS-B objective function should be introduced with equations in the Methods section for reproducibility.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the constructive feedback on our manuscript. We address each major comment below and indicate where revisions will be incorporated.

read point-by-point responses

  1. Referee: [Abstract and Methods (statistical deformation model)] Abstract (deformable alignment paragraph) and Methods section on statistical model construction: the patient-specific statistical deformation model is derived directly from NICP correspondences on the same patient data used for testing and evaluation. No information is given on the number of input shapes, percentage of variance explained by the principal modes, regularization strength, or any held-out reconstruction validation. This is load-bearing for the central claim that L-BFGS-B optimization of pose and shape parameters yields the reported 14.73 mm mean TRE, because an incomplete or biased basis could produce low surface error while leaving target error high, consistent with the tumor-mapping observations already noted in the abstract.

    Authors: We appreciate the referee pointing out the need for additional details on the statistical deformation model. The model is constructed in a patient-specific manner from NICP correspondences across multiple surface reconstructions of the same patient to capture observed deformations. In the revised manuscript we will explicitly report the number of input shapes, the percentage of variance explained by the retained principal modes, the regularization strength used during construction, and note that held-out reconstruction validation was not performed because the shape parameters are optimized jointly with pose for each individual registration case. The tumor-mapping analysis already included in the paper demonstrates awareness that surface alignment does not guarantee minimal target error, which directly relates to the concern about basis completeness. revision: yes

  2. Referee: [Experiments (evaluation and ablation)] Experiments section (evaluation protocol): the quantitative results are obtained under a controlled manual-contour setting that isolates registration performance. The manuscript does not report how performance degrades when contours are extracted automatically, nor does it provide error bars or statistical significance tests on the 14.73 mm TRE. This weakens the strength of the claim that the pipeline is ready for laparoscopic AR under realistic intraoperative conditions.

    Authors: We agree that the quantitative results are reported under a controlled manual-contour protocol, as stated in the Experiments section, to isolate the contribution of the registration pipeline itself. In the revision we will add error bars to the TRE values and include statistical significance tests. The manuscript does not evaluate automatic contour extraction because the focus is on the registration method; assessing degradation under automatic contours would require implementing and testing a complete automatic contour pipeline, which lies outside the current scope. revision: partial

standing simulated objections not resolved
  • Performance degradation when contours are extracted automatically, as this requires new experiments and integration of an automatic contour detector not present in the current study.

Circularity Check

0 steps flagged

No significant circularity in claimed registration pipeline

full rationale

The paper describes an empirical 3D/2D registration method for laparoscopic liver AR, evaluated via independent target registration error (TRE) on a public clinical dataset under manual-contour conditions. The patient-specific statistical deformation model is constructed from NICP correspondences as an explicit modeling choice within the deformable stage, then optimized via L-BFGS-B; this is a standard patient-specific fitting step rather than a first-principles derivation or renamed prediction. No equations, uniqueness theorems, or self-citations are invoked to force the result. The rigid stage relies on an external FoundationPose module, and tumor-mapping analysis explicitly separates surface alignment from target error. The pipeline is self-contained against external benchmarks with no load-bearing step that reduces by construction to its own inputs.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The pipeline rests on the assumption that monocular depth estimates are sufficiently accurate for pose refinement and that NICP correspondences produce a representative low-dimensional shape space for intraoperative deformations.

free parameters (1)
  • L-BFGS-B optimization hyperparameters
    Coarse-to-fine strategy parameters for pose and shape optimization are chosen but not detailed in the abstract.
axioms (1)
  • domain assumption Monocular depth from the laparoscopic scene provides useful relative pose cues when combined with multi-class contours
    Invoked in the rigid alignment description to justify depth augmentation over contour-only inputs.

pith-pipeline@v0.9.0 · 5852 in / 1370 out tokens · 28638 ms · 2026-05-22T10:53:36.208218+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Warm-Started Reinforcement Learning for Iterative 3D/2D Liver Registration

    cs.CV 2026-04 unverdicted novelty 5.0

    A warm-started discrete-action RL framework for CT-to-video liver registration achieves 15.70 mm average TRE with faster convergence than supervised methods plus optimization.

  2. Warm-Started Reinforcement Learning for Iterative 3D/2D Liver Registration

    cs.CV 2026-04 unverdicted novelty 4.0

    Warm-started RL policy performs iterative rigid registration of CT to laparoscopic video, reaching 15.70 mm average TRE with quicker convergence than hybrid supervised-optimization methods.

Reference graph

Works this paper leans on

30 extracted references · 30 canonical work pages · cited by 1 Pith paper · 1 internal anchor

  1. [1]

    The value of augmented reality in surgery—a usability study on laparoscopic liver surgery,

    J. Ramalhinho, S. Yoo, T. Dowrick, B. Koo, M. Somasundaram, K. Gurusamy, D. J. Hawkes, B. Davidson, A. Blandford, and M. J. Clarkson, “The value of augmented reality in surgery—a usability study on laparoscopic liver surgery,”Medical Image Analysis, vol. 90, p. 102943, 2023

    work page 2023

  2. [2]

    Comparison of manual and semi-automatic registration in augmented reality image-guided liver surgery: a clinical feasibility study,

    C. Schneider, S. Thompson, J. Totz, Y . Song, M. Allam, M. Soder- gren, A. Desjardins, D. Barratt, S. Ourselin, K. Gurusamyet al., “Comparison of manual and semi-automatic registration in augmented reality image-guided liver surgery: a clinical feasibility study,”Surgical endoscopy, vol. 34, no. 10, pp. 4702–4711, 2020

    work page 2020

  3. [3]

    An objective comparison of methods for augmented reality in laparoscopic liver resection by preoperative-to-intraoperative image fusion from the miccai2022 challenge,

    S. Ali, Y . Espinel, Y . Jin, P. Liu, B. G ¨uttner, X. Zhang, L. Zhang, T. Dowrick, M. J. Clarkson, S. Xiaoet al., “An objective comparison of methods for augmented reality in laparoscopic liver resection by preoperative-to-intraoperative image fusion from the miccai2022 challenge,”Medical image analysis, vol. 99, p. 103371, 2025

    work page 2025

  4. [4]

    Au- tomatic, global registration in laparoscopic liver surgery,

    B. Koo, M. R. Robu, M. Allam, M. Pfeiffer, S. Thompson, K. Gu- rusamy, B. Davidson, S. Speidel, D. Hawkes, D. Stoyanovet al., “Au- tomatic, global registration in laparoscopic liver surgery,”International Journal of Computer Assisted Radiology and Surgery, vol. 17, no. 1, pp. 167–176, 2022

    work page 2022

  5. [5]

    Automatic 3d/2d deformable registration in minimally invasive liver resection using a mesh recovery network

    M. Labrunie, D. Pizarro, C. Tilmant, and A. Bartoli, “Automatic 3d/2d deformable registration in minimally invasive liver resection using a mesh recovery network.” inMIDL, 2023, pp. 1104–1123

    work page 2023

  6. [6]

    Depth- driven geometric prompt learning for laparoscopic liver landmark detection,

    J. Pei, R. Cui, Y . Li, W. Si, J. Qin, and P.-A. Heng, “Depth- driven geometric prompt learning for laparoscopic liver landmark detection,” inInternational Conference on Medical Image Computing and Computer-Assisted Intervention. Springer, 2024, pp. 154–164

    work page 2024

  7. [7]

    Comparative analysis of non-rigid registration techniques for liver surface registration,

    B. Kundu, Z. Yang, R. Simon, and C. Linte, “Comparative analysis of non-rigid registration techniques for liver surface registration,” inMed- ical Imaging 2024: Image-Guided Procedures, Robotic Interventions, and Modeling, vol. 12928. SPIE, 2024, pp. 520–526

    work page 2024

  8. [8]

    Mechanical properties of human hepatic tissues to develop liver- mimicking phantoms for medical applications,

    A. S. Lemine, Z. Ahmad, N. J. Al-Thani, A. Hasan, and J. Bhadra, “Mechanical properties of human hepatic tissues to develop liver- mimicking phantoms for medical applications,”Biomechanics and Modeling in Mechanobiology, vol. 23, no. 2, pp. 373–396, 2024

    work page 2024

  9. [9]

    A clinically applicable laser-based image-guided system for laparoscopic liver procedures,

    M. Fusaglia, H. Hess, M. Schwalbe, M. Peterhans, P. Tinguely, S. Weber, and H. Lu, “A clinically applicable laser-based image-guided system for laparoscopic liver procedures,”International journal of computer assisted radiology and surgery, vol. 11, no. 8, pp. 1499– 1513, 2016

    work page 2016

  10. [10]

    Evaluation of a novel navigation platform for laparoscopic liver surgery with organ deformation compensation using injected fiducials,

    E. Pelanis, A. Teatini, B. Eigl, A. Regensburger, A. Alzaga, R. P. Kumar, T. Rudolph, D. L. Aghayan, C. Riediger, N. Kvarnstr ¨om et al., “Evaluation of a novel navigation platform for laparoscopic liver surgery with organ deformation compensation using injected fiducials,” Medical image analysis, vol. 69, p. 101946, 2021

    work page 2021

  11. [11]

    Global rigid registration of ct to video in laparoscopic liver surgery,

    M. R. Robu, J. Ramalhinho, S. Thompson, K. Gurusamy, B. Davidson, D. Hawkes, D. Stoyanov, and M. J. Clarkson, “Global rigid registration of ct to video in laparoscopic liver surgery,”International Journal of Computer Assisted Radiology and Surgery, vol. 13, no. 6, pp. 947–956, 2018

    work page 2018

  12. [12]

    Augmented reality navigation for liver resection with a stereoscopic laparoscope,

    H. Luo, D. Yin, S. Zhang, D. Xiao, B. He, F. Meng, Y . Zhang, W. Cai, S. He, W. Zhanget al., “Augmented reality navigation for liver resection with a stereoscopic laparoscope,”Computer methods and programs in biomedicine, vol. 187, p. 105099, 2020

    work page 2020

  13. [13]

    Point cloud registration in la- paroscopic liver surgery using keypoint correspondence registration network,

    Y . Zhang, Y . Zou, and P. X. Liu, “Point cloud registration in la- paroscopic liver surgery using keypoint correspondence registration network,”IEEE Transactions on Medical Imaging, 2024

    work page 2024

  14. [14]

    Learning feature descriptors for pre-and intra-operative point cloud matching for laparoscopic liver registration,

    Z. Yang, R. Simon, and C. A. Linte, “Learning feature descriptors for pre-and intra-operative point cloud matching for laparoscopic liver registration,”International journal of computer assisted radiology and surgery, vol. 18, no. 6, pp. 1025–1032, 2023

    work page 2023

  15. [15]

    Automatic preoperative 3d model registration in laparoscopic liver resection,

    M. Labrunie, M. Ribeiro, F. Mourthadhoi, C. Tilmant, B. Le Roy, E. Buc, and A. Bartoli, “Automatic preoperative 3d model registration in laparoscopic liver resection,”International Journal of Computer Assisted Radiology and Surgery, vol. 17, no. 8, pp. 1429–1436, 2022

    work page 2022

  16. [16]

    A methodology and clinical dataset with ground-truth to evaluate registration accuracy quantitatively in computer-assisted laparoscopic liver resection,

    N. Rabbani, L. Calvet, Y . Espinel, B. Le Roy, M. Ribeiro, E. Buc, and A. Bartoli, “A methodology and clinical dataset with ground-truth to evaluate registration accuracy quantitatively in computer-assisted laparoscopic liver resection,”Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, vol. 10, no. 4, pp. 441–450, 2022

    work page 2022

  17. [17]

    Towards multi-modal self-supervised video and ultrasound pose estimation for laparoscopic liver surgery,

    N. Monta ˜na-Brown, J. Ramalhinho, B. Koo, M. Allam, B. Davidson, K. Gurusamy, Y . Hu, and M. J. Clarkson, “Towards multi-modal self-supervised video and ultrasound pose estimation for laparoscopic liver surgery,” inInternational Workshop on Advances in Simplifying Medical Ultrasound. Springer, 2022, pp. 183–192

    work page 2022

  18. [18]

    Deep hashing for global registration of preoperative ct and video images for laparoscopic liver surgery,

    H. Zhang, S. Bulathsinhala, B. R. Davidson, M. J. Clarkson, and J. Ramalhinho, “Deep hashing for global registration of preoperative ct and video images for laparoscopic liver surgery,”International Journal of Computer Assisted Radiology and Surgery, pp. 1–9, 2025

    work page 2025

  19. [19]

    A 3d-2d rigid liver registration method using pre-training and transfer learning with staged alignment of anatomical landmarks,

    J. Hao, B. He, Y . Dai, Y . Li, Y . Wang, R. Zhao, R. Lian, X. Zeng, H. Tao, J. Yanget al., “A 3d-2d rigid liver registration method using pre-training and transfer learning with staged alignment of anatomical landmarks,”International Journal of Imaging Systems and Technology, vol. 35, no. 4, p. e70124, 2025

    work page 2025

  20. [20]

    Silhouette-based pose estimation for deformable organs appli- cation to surgical augmented reality,

    Y . Adagolodjo, R. Trivisonne, N. Haouchine, S. Cotin, and H. Courte- cuisse, “Silhouette-based pose estimation for deformable organs appli- cation to surgical augmented reality,” in2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2017, pp. 539–544

    work page 2017

  21. [21]

    Fem simulation of 3d deformable solids: a practitioner’s guide to theory, discretization and model reduction,

    E. Sifakis and J. Barbic, “Fem simulation of 3d deformable solids: a practitioner’s guide to theory, discretization and model reduction,” in Acm siggraph 2012 courses, 2012, pp. 1–50

    work page 2012

  22. [22]

    As-rigid-as-possible surface modeling,

    O. Sorkine and M. Alexa, “As-rigid-as-possible surface modeling,” in Symposium on Geometry processing, vol. 4, 2007, pp. 109–116

    work page 2007

  23. [23]

    Neural patient-specific 3d–2d registration in laparoscopic liver resection,

    I. Mhiri, D. Pizarro, and A. Bartoli, “Neural patient-specific 3d–2d registration in laparoscopic liver resection,”International Journal of Computer Assisted Radiology and Surgery, vol. 20, no. 1, pp. 57–64, 2025

    work page 2025

  24. [24]

    Foundationpose: Unified 6d pose estimation and tracking of novel objects,

    B. Wen, W. Yang, J. Kautz, and S. Birchfield, “Foundationpose: Unified 6d pose estimation and tracking of novel objects,” inProceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 17 868–17 879

    work page 2024

  25. [25]

    Depth anything v2,

    L. Yang, B. Kang, Z. Huang, Z. Zhao, X. Xu, J. Feng, and H. Zhao, “Depth anything v2,”Advances in Neural Information Processing Systems, vol. 37, pp. 21 875–21 911, 2024

    work page 2024

  26. [26]

    Saramis: Simulation assets for robotic assisted and minimally invasive surgery,

    N. Montana-Brown, S. U. Saeed, A. Abdulaal, T. Dowrick, Y . Kilic, S. Wilkinson, J. Gao, M. Mashar, C. He, A. Stavropoulouet al., “Saramis: Simulation assets for robotic assisted and minimally invasive surgery,”Advances in Neural Information Processing Systems, vol. 36, pp. 26 121–26 134, 2023

    work page 2023

  27. [27]

    Intraoperative liver surface completion with graph convolutional vae,

    S. Foti, B. Koo, T. Dowrick, J. Ramalhinho, M. Allam, B. David- son, D. Stoyanov, and M. J. Clarkson, “Intraoperative liver surface completion with graph convolutional vae,” inInternational Workshop on Uncertainty for Safe Utilization of Machine Learning in Medical Imaging. Springer, 2020, pp. 198–207

    work page 2020

  28. [28]

    Meshpreprocessing: A library for mesh registration and non-rigid icp,

    S. Foti, “Meshpreprocessing: A library for mesh registration and non-rigid icp,” GitHub repository, 2020, available: https://github.com/simofoti/MeshPreprocessing [Accessed: Sep. 11, 2025]

    work page 2020

  29. [29]

    Optimal step nonrigid icp algorithms for surface registration,

    B. Amberg, S. Romdhani, and T. Vetter, “Optimal step nonrigid icp algorithms for surface registration,” in2007 IEEE conference on computer vision and pattern recognition. IEEE, 2007, pp. 1–8

    work page 2007

  30. [30]

    The CMA Evolution Strategy: A Tutorial

    H. Nikolaus, “The cma evolution strategy: A tutorial,”arXiv preprint arXiv: 1604.00772, 2016

    work page internal anchor Pith review Pith/arXiv arXiv 2016