Model-based 3D human shape estimation from silhouettes for virtual fitting

Shunta Saito, Makiko Kouchi, Masaaki Mochimaru, Yoshimitsu Aoki

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

We propose a model-based 3D human shape reconstruction system from two silhouettes. Firstly, we synthesize a deformable body model from 3D human shape database consists of a hundred whole body mesh models. Each mesh model is homologous, so that it has the same topology and same number of vertices among all models. We perform principal component analysis (PCA) on the database and synthesize an Active Shape Model (ASM). ASM allows changing the body type of the model with a few parameters. The pose changing of our model can be achieved by reconstructing the skeleton structures from implanted joints of the model. By applying pose changing after body type deformation, our model can represents various body types and any pose. We apply the model to the problem of 3D human shape reconstruction from front and side silhouette. Our approach is simply comparing the contours between the model's and input silhouettes', we then use only torso part contour of the model to reconstruct whole shape. We optimize the model parameters by minimizing the difference between corresponding silhouettes by using a stochastic, derivative-free non-linear optimization method, CMA-ES.

Original languageEnglish
Title of host publicationProceedings of SPIE - The International Society for Optical Engineering
PublisherSPIE
Volume9013
ISBN (Print)9780819499301
DOIs
Publication statusPublished - 2014
EventThree-Dimensional Image Processing, Measurement (3DIPM), and Applications 2014 - San Francisco, CA, United States
Duration: 2014 Feb 52014 Feb 5

Other

OtherThree-Dimensional Image Processing, Measurement (3DIPM), and Applications 2014
CountryUnited States
CitySan Francisco, CA
Period14/2/514/2/5

Fingerprint

Silhouette
Model-based
Model
Active Shape Model
Shape Reconstruction
Human
Derivative-free Optimization
Mesh
mesh
torso
Nonlinear Optimization
Skeleton
Principal Component Analysis
Optimization Methods
principal components analysis
musculoskeletal system
Principal component analysis
Optimise

Keywords

  • 3D reconstruction
  • Active shape model
  • Anatomical landmarks
  • Digital Human Model
  • Non-linear optimization
  • Principal component analysis
  • Shape-from-silhouettes
  • Skinning

ASJC Scopus subject areas

  • Applied Mathematics
  • Computer Science Applications
  • Electrical and Electronic Engineering
  • Electronic, Optical and Magnetic Materials
  • Condensed Matter Physics

Cite this

Saito, S., Kouchi, M., Mochimaru, M., & Aoki, Y. (2014). Model-based 3D human shape estimation from silhouettes for virtual fitting. In Proceedings of SPIE - The International Society for Optical Engineering (Vol. 9013). [901307] SPIE. https://doi.org/10.1117/12.2038457

Model-based 3D human shape estimation from silhouettes for virtual fitting. / Saito, Shunta; Kouchi, Makiko; Mochimaru, Masaaki; Aoki, Yoshimitsu.

Proceedings of SPIE - The International Society for Optical Engineering. Vol. 9013 SPIE, 2014. 901307.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Saito, S, Kouchi, M, Mochimaru, M & Aoki, Y 2014, Model-based 3D human shape estimation from silhouettes for virtual fitting. in Proceedings of SPIE - The International Society for Optical Engineering. vol. 9013, 901307, SPIE, Three-Dimensional Image Processing, Measurement (3DIPM), and Applications 2014, San Francisco, CA, United States, 14/2/5. https://doi.org/10.1117/12.2038457
Saito S, Kouchi M, Mochimaru M, Aoki Y. Model-based 3D human shape estimation from silhouettes for virtual fitting. In Proceedings of SPIE - The International Society for Optical Engineering. Vol. 9013. SPIE. 2014. 901307 https://doi.org/10.1117/12.2038457
Saito, Shunta ; Kouchi, Makiko ; Mochimaru, Masaaki ; Aoki, Yoshimitsu. / Model-based 3D human shape estimation from silhouettes for virtual fitting. Proceedings of SPIE - The International Society for Optical Engineering. Vol. 9013 SPIE, 2014.
@inproceedings{58bf69973c5b4dc0a37ee1165902024f,
title = "Model-based 3D human shape estimation from silhouettes for virtual fitting",
abstract = "We propose a model-based 3D human shape reconstruction system from two silhouettes. Firstly, we synthesize a deformable body model from 3D human shape database consists of a hundred whole body mesh models. Each mesh model is homologous, so that it has the same topology and same number of vertices among all models. We perform principal component analysis (PCA) on the database and synthesize an Active Shape Model (ASM). ASM allows changing the body type of the model with a few parameters. The pose changing of our model can be achieved by reconstructing the skeleton structures from implanted joints of the model. By applying pose changing after body type deformation, our model can represents various body types and any pose. We apply the model to the problem of 3D human shape reconstruction from front and side silhouette. Our approach is simply comparing the contours between the model's and input silhouettes', we then use only torso part contour of the model to reconstruct whole shape. We optimize the model parameters by minimizing the difference between corresponding silhouettes by using a stochastic, derivative-free non-linear optimization method, CMA-ES.",
keywords = "3D reconstruction, Active shape model, Anatomical landmarks, Digital Human Model, Non-linear optimization, Principal component analysis, Shape-from-silhouettes, Skinning",
author = "Shunta Saito and Makiko Kouchi and Masaaki Mochimaru and Yoshimitsu Aoki",
year = "2014",
doi = "10.1117/12.2038457",
language = "English",
isbn = "9780819499301",
volume = "9013",
booktitle = "Proceedings of SPIE - The International Society for Optical Engineering",
publisher = "SPIE",

}

TY - GEN

T1 - Model-based 3D human shape estimation from silhouettes for virtual fitting

AU - Saito, Shunta

AU - Kouchi, Makiko

AU - Mochimaru, Masaaki

AU - Aoki, Yoshimitsu

PY - 2014

Y1 - 2014

N2 - We propose a model-based 3D human shape reconstruction system from two silhouettes. Firstly, we synthesize a deformable body model from 3D human shape database consists of a hundred whole body mesh models. Each mesh model is homologous, so that it has the same topology and same number of vertices among all models. We perform principal component analysis (PCA) on the database and synthesize an Active Shape Model (ASM). ASM allows changing the body type of the model with a few parameters. The pose changing of our model can be achieved by reconstructing the skeleton structures from implanted joints of the model. By applying pose changing after body type deformation, our model can represents various body types and any pose. We apply the model to the problem of 3D human shape reconstruction from front and side silhouette. Our approach is simply comparing the contours between the model's and input silhouettes', we then use only torso part contour of the model to reconstruct whole shape. We optimize the model parameters by minimizing the difference between corresponding silhouettes by using a stochastic, derivative-free non-linear optimization method, CMA-ES.

AB - We propose a model-based 3D human shape reconstruction system from two silhouettes. Firstly, we synthesize a deformable body model from 3D human shape database consists of a hundred whole body mesh models. Each mesh model is homologous, so that it has the same topology and same number of vertices among all models. We perform principal component analysis (PCA) on the database and synthesize an Active Shape Model (ASM). ASM allows changing the body type of the model with a few parameters. The pose changing of our model can be achieved by reconstructing the skeleton structures from implanted joints of the model. By applying pose changing after body type deformation, our model can represents various body types and any pose. We apply the model to the problem of 3D human shape reconstruction from front and side silhouette. Our approach is simply comparing the contours between the model's and input silhouettes', we then use only torso part contour of the model to reconstruct whole shape. We optimize the model parameters by minimizing the difference between corresponding silhouettes by using a stochastic, derivative-free non-linear optimization method, CMA-ES.

KW - 3D reconstruction

KW - Active shape model

KW - Anatomical landmarks

KW - Digital Human Model

KW - Non-linear optimization

KW - Principal component analysis

KW - Shape-from-silhouettes

KW - Skinning

UR - http://www.scopus.com/inward/record.url?scp=84901784394&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84901784394&partnerID=8YFLogxK

U2 - 10.1117/12.2038457

DO - 10.1117/12.2038457

M3 - Conference contribution

AN - SCOPUS:84901784394

SN - 9780819499301

VL - 9013

BT - Proceedings of SPIE - The International Society for Optical Engineering

PB - SPIE

ER -