Integrating 3D Posture Neutralization with Statistical Shape Modeling for Head-Neck-Shoulder Ergonomic Design

Jinbo Luo; Hang Ren; Yadie Yang; Jie Zhang

cs.GROct 13, 2026

Integrating 3D Posture Neutralization with Statistical Shape Modeling for Head-Neck-Shoulder Ergonomic Design

Authors: Jinbo Luo, Hang Ren, Yadie Yang, Jie Zhang

Organizations: School of Fashion and Textiles, The Hong Kong Polytechnic University, Hong Kong SAR, China · Shenzhen Pingle Orthopedic Hospital, Shenzhen, China · Faculty of Applied Sciences, Macao Polytechnic University, Macau SAR, China

Abstract

The design of ergonomic products and workspaces requires a deep understanding of human anatomy and posture. In particular, the head-neck-shoulder region is critical for many applications such as cervical orthoses, helmets, and other head-worn or neck-worn devices. This paper presents a framework that integrates posture neutralization with statistical shape modeling (SSM) for the head-neck-shoulder (HNS) region. We acquire a single three-dimensional (3D) scan of the head-neck-shoulder region from each of 200 Chinese adults in a natural posture for each subject. Every scan is then posture-neutralized before SSM. We fit a parametric head-neck model to each scan to separate cervical pose from anatomy, and reset the cervical pose to a common neutral configuration, so that the resulting model captures anatomical shape variation rather than posture-induced deformation. The resulting shape space yields a standardized template and population-wide modes of variation, providing baselines for parametric ergonomic design. We further discuss how the model supports population coverage analysis and the design of products that accommodate anatomical diversity while maintaining neutral cervical posture.

Figures & tables

Figure 1: Typical applications of 3D head-neck-shoulder morphological data. (This figure was created by using Procreate software.)

Study	Group	Age	Sample size	Region	Modeling	Model information			Usage
						Shape	Size	Pose
Wang et al. (2017) [ 34 ]	Chinese Male	25–30	200	Neck	$\times$	$\checkmark$	$\checkmark$	$\checkmark$	Garment design
FLAME (2017) [ 35 ]	US/European	18–65	3,800	Head-Neck	$\checkmark$	$\checkmark$	$\times$	$\checkmark$	Animation
LYHM (2020) [ 23 ]	Mostly White	1–89	1,212	Head	$\checkmark$	$\checkmark$	$\times$	$\times$	Head generation
Wang and Gu (2021) [ 32 ]	Chinese Female	18–25	188	Neck-Shoulder	$\times$	$\checkmark$	$\checkmark$	$\checkmark$	Garment design, anthropometry
Zhang et al. (2022a) [ 31 ]	Chinese Male	18–25	174	Neck-Shoulder	$\times$	$\checkmark$	$\checkmark$	$\checkmark$	Garment design, anthropometry

Table 1: Comparison of head, neck and shoulder shape studies.

Figure 2: Overview of the head-neck-shoulder shape modelling pipeline. (1) Raw three-dimensional scans are collected in natural posture. (2) Anatomical landmarks and non-rigid iterative closest point (NICP) registration establish vertex correspondence. (3) Model-derived per-vertex transformations map each registered head-neck surface to a common neutral posture while retaining individual surface detail. (4) Generalized Procrustes analysis (GPA) aligns the meshes by rotation and translation. (5) Principal component analysis (PCA) represents population shape variation by the mean shape and principal components. The cervical bones [ 43 ] and joint chains are shown for illustration only and do not represent experimentally measured skeletal data.

Figure 3: Distributions of participant characteristics in the study cohort ( $n=200$ ). (a) Age. (b) Height. (c) Weight. (d) Body mass index (BMI).

Panel of Figure 4 — Figure 4: Our predefined $73$ landmarks on the template mesh, including 22 ear-neck-shoulder landmarks and 51 facial landmarks. (a) Ear-Neck-Shoulder landmarks (1-13): shown in front (a1), side (a2), and back (a3) views. Indices 1-4 are bilateral pairs along the anterior helix (1, helix apex, 2, upper helix, 3, lower anterior helix, 4, pre-auricular notch), and 5-7 are bilateral pairs along the posterior helix (5, posterior helix near antitragus, 6, mid posterior helix, 7, posterior helix apex). Midline points 8 and 9 are the thyroid notch and the suprasternal (jugular) notch. Indices 10a/10b are the supraclavicular fossae. Posterior midline points 11 and 12 are the high cervical point near the C1 level and the C7 vertebra prominens. Indices 13a/13b are the acromion tips. (b) Facial landmarks (14-43) are shown in three zoomed panels. The nose (b1) includes points 25-29 along the midline from nasion to subnasale, and pairs 30a/30b and 31a/31b at the inner and outer nostril edges. The eyebrows and eyes (b2) include brow points 14-18 in five bilateral pairs from lateral to medial, and eye points 19-24 in six pairs running from outer canthus around the eye to lower lid lateral. The mouth (b3) includes outer lip points 32-38 from mouth corner to lower-lip midline, and inner lip points 39-43 from paramedian to midline. Facial indices follow the iBUG 300-W convention [ 47 ] , reusing points 17-67 of the 68-point scheme. Throughout, suffixes a / b denote the left/right side of the image.

Figure 6: Neck posture neutralization. (a) The HACK reference model is fitted to the registered head-neck surface to estimate individual shape, acquired posture and the global similarity transform between model and registered coordinates. (b) The fitted shape is held fixed while cervical posture is reset to the common neutral configuration. Joint transformations and fixed skinning weights define the per-vertex neutralization. $R$ : Rotation Matrix $\in R^{3\times 3}$ . (c) The model-derived neutralization is transferred through vertex correspondence to update the original registered surface, with coordinate changes determined by the fitted similarity transform. The cervical bones and joint chains are shown for illustration only and do not represent experimentally measured skeletal data.

Figure 7: Cervical skinning weights on a representative neutralized registered head-neck surface. Weight fields for the C7–T1, C6–C7, C5–C6, C4–C5, C3–C4, C2–C3, C1–C2 and occiput–C1 (Oc–C1) joints, respectively.

Figure 8: Average posture-neutralized head-neck-shoulder shape. The same mean mesh is shown in (a) frontal and (b) lateral views.

Panel of Figure 9 — Figure 9: First five principal components of the posture-neutralized statistical shape model. The first and second rows show the $+3$ SD and $-3$ SD endpoints in the left lateral view. The third row overlays both endpoints. Slate blue denotes $+3$ SD, and orange denotes $-3$ SD.

Figure 11: Anthropometric changes along the first five principal components of the posture-neutralized model. (a) Endpoint changes in nine dimensions. (b) Absolute endpoint measurements for head circumference, neck circumference, neck height, and shoulder width.

Figure 12: Predicted head-neck shape across gender and BMI. (a) Mean surfaces from gender-specific BMI regressions at BMI 18.5, 23, 25 and 30 kg/m 2 . (b) Surface displacement between successive BMI levels. Red and blue indicate outward and inward normal displacement, respectively.

Panel of Figure 13 — Figure 13: Enlarged cervical-chain Cobb angle comparison for the first five PC endpoints. Rows show (a) the neutralized model imposed and (b) the non-neutralized model with natural poses. Blue solid chains denote $+3$ SD, and orange dashed chains denote $-3$ SD.

Figure 15: Evaluation of four SSM representations. (a) Compactness for the first 50 components of the full dataset. (b) Generalization for the first 20 components. (c) Specificity for the first 20 components on a logarithmic vertical axis.

Statistical Shape Modeling

Statistical shape modeling represents and analyzes population-level geometric variation to support shape comparison, reconstruction, and synthesis.

See all 23 papers

MorphoFlow: Sparse-Supervised Generative Shape Modeling with Adaptive Latent Relevance
Apr 13, 2026Mokshagna Sai Teja Karanam, Tushar Kataria, Shireen ElhabianStatistical Shape Modeling 3D Shape Generation
Kahlert School of Computing, Scientific Computing and Imaging Institute, University of Utah, Utah, USA
TRACE: Artifact-Robust Statistical Shape Modeling from Imperfect Surface Scans - A Case Study in Craniosynostosis 3D Photography
Aug 22, 2026Sanjay Bhandari, Nawazish Khan, Alzbeta Novotna +7Statistical Shape Modeling Deformable Shape Reconstruction
Scientific Computing and Imaging Institute, University of Utah, SLC, UT, USA · Kahlert School of Computing, University of Utah, Salt Lake City, UT, USA · Division of Pediatric Plastic Surgery, UPMC Children’s Hospital of Pittsburgh, PA, USA
SCALP: Semi-Supervised Statistical Shape Modeling from Imperfect 3D Photogrammetry via Landmark-Anchored Spectral Warp
Jul 31, 2026Nawazish Khan, Sanjay Bhandari, Sarang Joshi +8Point Cloud Learning Statistical Shape Modeling
Scientific Computing and Imaging Institute, University of Utah, Salt Lake City, UT 84112, USA · Kahlert School of Computing, University of Utah, Salt Lake City, UT 84112, USA · Department of Biomedical Engineering, University of Utah, Salt Lake City, UT 84112, USA +1

Apr 13, 2026J/K move · Enter open · S save

MorphoFlow: Sparse-Supervised Generative Shape Modeling with Adaptive Latent Relevance

Mokshagna Sai Teja Karanam, Tushar Kataria, Shireen Elhabian

Kahlert School of Computing, Scientific Computing and Imaging Institute, University of Utah, Utah, USA

Statistical Shape Modeling 3D Shape Generation Flow-Based Generative Modeling

Open paper AskPDF arXiv

Statistical shape modeling (SSM) is central to population level analysis of anatomical variability, yet most existing approaches rely on densely annotated segmentations and fixed latent representations. These requirements limit scalability and reduce flexibility when modeling complex anatomical variation. We introduce MorphoFlow, a sparse supervised generative shape modeling framework that learns compact probabilistic shape representations directly from sparse surface annotations. MorphoFlow integrates neural implicit shape representations with an autodecoder formulation and autoregressive normalizing flows to learn an expressive probabilistic density over the latent shape space. The neural implicit representation enables resolution-agnostic modeling of 3D anatomy, while the autodecoder formulation supports direct optimization of per-instance latent codes under sparse supervision. The autoregressive flow captures the distribution of latent anatomical variability providing a tractable, likelihood-based generative model of shapes. To promote compact and structured latent representations, we incorporate adaptive latent relevance weighting through sparsity-inducing priors, enabling the model to regulate the contribution of individual latent dimensions according to their relevance to the underlying anatomical variation while preserving generative expressivity. The resulting latent space supports uncertainty quantification and anatomically plausible shape synthesis without manual latent dimensionality tuning. Evaluation on publicly available lumbar vertebrae and femur datasets demonstrates accurate high-resolution reconstruction from sparse inputs and recovery of structured modes of anatomical variation consistent with population level trends.

Abstract

Statistical Shape Modeling

Explore similar work

MorphoFlow: Sparse-Supervised Generative Shape Modeling with Adaptive Latent Relevance

TRACE: Artifact-Robust Statistical Shape Modeling from Imperfect Surface Scans - A Case Study in Craniosynostosis 3D Photography

SCALP: Semi-Supervised Statistical Shape Modeling from Imperfect 3D Photogrammetry via Landmark-Anchored Spectral Warp