Fusion of anthropometric data and principal component analysis of the bones for generating a personalized skeleton: case of the lower limb - Université Claude Bernard Lyon 1 Accéder directement au contenu
Communication Dans Un Congrès Année : 2017

Fusion of anthropometric data and principal component analysis of the bones for generating a personalized skeleton: case of the lower limb

Résumé

An approach is presented and evaluated here to provide realistic internal and external dimensions of Human Body Models for a broad range of predictor values, such as, for example, to estimate the dimensions of a slightly overweight 11 year old of height 1m40. The difficulty due to lack of sufficient internal and external data is alleviated by combining widely available anthropometric measurements with more sparse databases of surface meshes of bones. Assuming that the external anthropometric measurements are estimated from the predictors through a regression, the link with the internal shape dimensions is made through the subset of target ?skeletal? measurements that can be closely estimated from skeletal landmarks. The matching of internal dimensions then proceeds as follows. First the set of surface meshes of bones is aligned and a Principal Component Analysis (PCA) model of their variance compared to their mean shape is evaluated. This is done by aligning the bones, body part by body part, in such a way that the effect of posture is removed, and by evaluating the variance for the whole skeleton, so that the covariance of shapes between different bones is captured. Any subject sampled from this PCA model consists in a set of bones loosely positioned compared to each other, for particular values of the Principal Component (PC) modal magnitudes or scores. In a second step, values of these modal magnitudes are estimated in such a way that the corresponding sampled subject skeleton has dimensions that estimate closely the target skeletal anthropometric measurements. In this last step, a direct kinematic approach and a sequential quadratic programming method are used to reposition the sampled bones in an appropriate posture and to evaluate the most appropriate PC scores. The whole approach is successfully applied and evaluated on a three segment model of the lower limb.
Fichier principal
Vignette du fichier
ris00000739.pdf (307.13 Ko) Télécharger le fichier
Origine : Fichiers éditeurs autorisés sur une archive ouverte

Dates et versions

hal-01769913 , version 1 (18-04-2018)

Identifiants

  • HAL Id : hal-01769913 , version 1

Citer

Christophe Lecomte, Xuguang Wang. Fusion of anthropometric data and principal component analysis of the bones for generating a personalized skeleton: case of the lower limb. 5th International Digital Human Modeling Symposium, Jun 2017, BONN, Germany. 14 p. ⟨hal-01769913⟩
42 Consultations
227 Téléchargements

Partager

Gmail Facebook X LinkedIn More