Parametric body shape model of standing children aged 3–11 years

Abstract

A statistical body shape model (SBSM) for children was developed for generating a child body shape with desired anthropometric parameters. A standardised template mesh was fit to whole-body laser scan data from 137 children aged 3-11 years. The mesh coordinates along with a set of surface landmarks and 27 manually measured anthropometric variables were analysed using principal component (PC) analysis. PC scores were associated with anthropometric predictors such as stature, body mass index (BMI) and ratio of erect sitting height to stature (SHS) using a regression model. When the original scan data were compared with the predictions of the SBSM using each subject's stature, BMI and SHS, the mean absolute error was 10.4 ± 5.8 mm, and 95th percentile error was 24.0 ± 18.5 mm. The model, publicly available online, will have utility for a wide range of applications. Practitioner Summary: A statistical body shape model for children helps to account for inter-individual variability in body shapes as well as anthropometric dimensions. This parametric modelling approach is useful for reliable prediction of the body shape of a specific child with a few given predictors such as stature, body mass index and age.

Full text

Parametric Child Body Shape Model
1
Parametric Body Shape Model of Standing Children Ages 3 to 11 Years
Byoung-Keon Park1 and Matthew P. Reed1*
*Corresponding author: Matthew P. Reed, Ph.D (mreed@umich.edu)
1 Biosciences Group, University of Michigan Transportation Research Institute, Ann Arbor, MI, U.S.A

Parametric Child Body Shape Model
2
Parametric Body Shape Model of Standing Children Ages 3 to 11 Years
Byoung-Keon Park and Matthew P. Reed
Abstract
A statistical body shape model (SBSM) for children was developed for generating a child body
shape with desired anthropometric parameters. A standardized template mesh was fit to whole-
body laser scan data from 137 children ages 3 to 11. The mesh coordinates along with a set of
surface landmarks and 27 manually measured anthropometric variables were analyzed using
principal component (PC) analysis. PC scores were associated with anthropometric predictors
such as stature, body mass index (BMI), and ratio of erect sitting height to stature (SHS) using a
regression model. When the original scan data were compared with the predictions of the SBSM
using each subject’s stature, BMI, and SHS, the mean absolute error was 10.4±5.8 mm, and 95th
percentile error was 24.0±18.5 mm. The model, publicly accessible at childshape.org, will have
utility for a wide range of applications.
Practitioner Summary: A statistical body shape model (SBSM) for children helps to account
for inter-individual variability in body shapes as well as anthropometrics. This parametric
modeling approach is useful for reliable prediction of the body shape of a specific child with a
few given predictors such as stature, body mass index, and age.
Keywords: statistical body shape model; parametric modeling; anthropometry; body shape
prediction; child body shape measurement
Introduction
Three-dimensional, whole-body surface measurement has revolutionized applied anthropometry
(Daanen and Ter Haar 2013, Lu et al. 2008, Stančić et al. 2013). In the U.S. and Europe, the
Civilian American and European Surface Anthropometry Resource (CAESAR) study provided
the first large-scale database of human body shapes (Robinette et al. 1999). Since that time,
numerous 3D studies have been conducted, increasing the availability of data on body shape,

Parametric Child Body Shape Model
3
particularly in standing postures (Allen et al. 2003, Ben Azouz et al. 2006, Wells et al. 2008, Han
et al. 2010, Yu et al. 2010)
The utility of raw body scan data is limited, however. Often, traditional anthropometric
dimensions are estimated based on the scan data, while the scans themselves are not widely used.
In recent years, two methods for making greater use of the three-dimensional information have
become more common. First, individual scans can be selected from a database, using information
such as a target stature, body weight, or even body shape (Allen et al. 2004, Yu et al. 2010).
Second, statistical models of body size and shape are created by analysis of a large number of
scans (Allen et al. 2003, Mochimaru and Kouchi 2000, Hasler et al. 2009). A statistical body
shape model (SBSM) can be designed to predict the mean expected body shape as a function of a
wide range of possible predictors, including standard anthropometric descriptors, such as stature
and body weight, or other variables such as the locations of body landmarks.
Previous studies reporting the development of SBSMs have focused on adults. The current paper
presents an SBSM developed from body scans of children ages 3 to 11 years. The model
reported here will be publicly accessible at childshape.org following publication of this paper.
Methods
Subject Pool
The protocol and recruitment methods used in this study were approved by a University of
Michigan Institutional Review board. A total of 137 children ages 3 to 11 years old were
recruited using flyers and word-of-mouth. The age limits were determined based on (1)
developmental maturity sufficient to stand still for 15 seconds at a time and (2) minimal evidence
of puberty. The selected children were free of evident skeletal deformities or musculoskeletal
injuries or disabilities. Figure 1 shows the distribution of stature and body weight versus age in
months in the subject pool. Growth curves from the U.S. National Center for Health Statistics for
weight and stature (5th, 25th, 75th and 95th percentile) are overlaid on the charts.
Written informed consent was obtained from a parent or guardian and oral assent was obtained
from the participant. Each participant changed into a close-fitting swimsuit provided by the

Parametric Child Body Shape Model
4
investigators. Standard anthropometric measures were obtained from each participant using
methods identical to those in Gordon et al. (1989). A set of surface landmarks listed in Appendix
I were marked on the child’s skin using water-soluble pens and body paint.
Scanning
Surface measurement was conducted using a VITUS XXL laser scanner and reconstruction
software ScanWorX (Human Solutions). Hardware and software system performance was
verified prior to each day of testing by scanning a 100-mm diameter pole and verifying the
circumference measurement. In general, the system accuracy is approximately 2 mm, depending
on the location within the scan volume.
Figure 2 shows the scan posture used for the current analysis. The children wore tight-fitting
swimsuits and a swim cap. During the scan, the subjects were asked to stay still in a standing
posture with the arms abducted from the torso about 30 degrees and the legs slightly spread.
Surface landmark locations were extracted from the scans using Meshlab version 1.3 software
(meshlab.org). A set of internal joint centers was estimated using methods reported in Reed et al.
(1999).
Template Fitting
To facilitate statistical analysis, a standardized template model with 30k vertices and 60k
polygons was fit to the data from each scan. As described in Figure 3, the template fitting
procedure comprises two morphing steps: initial morphing to match the body pose and
orientation based on synthetic landmarks, and fine morphing to fit the template vertices to the
corresponding positions on the target data surface.
In the initial morphing step, the 20 automatically generated synthetic landmarks shown in Figure
4 were used to roughly fit the body orientation and the posture of the template model to a target
scan. The landmarks were automatically found based on local geometric features from both of
the template and target scan data (Lu et al. 2008, Zhong et al. 2006). An interpolated based on
radial basis function (RBF) techniques was then built using the landmarks to morph the template
model. In general, an RBF interpolator is a function of the form

Parametric Child Body Shape Model
5
)()()( 1xxxx pws i
N
ii 


, (1)
where p(x) = a1+a2x+a3y+a4z is a polynomial defined in the null space of the differential
operator, the coefficients wi and ai are real numbers,

is the Euclidean norm, and
)log()( 2rrr 

is a basis function known as the thin-plate spline. The coefficients of the
interpolating function in equation (1) can be found by solving a linear system of equations given
by

























O
L
a
w
OL LB trg
T
tpl
tpl
, (2)
where
)(
,jiji xxB 

, and O is a zero matrix. Ltpl is the matrix of the landmark points of
the template with ith row (1, xi , yi , zi), Ltrg is the matrix of target landmark points. Once the
coefficients of the interpolation function (w, a)T are solved from equation (2), all the vertices of
the template model are moved by equation (1).
For the fine morphing, the surface fitting problem of the morphed template model was
formulated using an implicit function constructed from the target scan data (Carr et al. 2001).
The fine fitting process used here is as follows:
1. Partition the target data into boxes that contain 200 to 300 points sized using kd-tree.
2. Within each box:
a. Assign each surface point a scalar value of zero.
b. For a subset of the surface points, construct new points that are “inside” and
“outside” the surface. Inside and outside points are created by moving along the
normal by a margin.
c. Assign each inside point a value of -1, each outside point a value of 1.
d. Compute an interpolation function (Input: 3D coordinates, output: 1D scalar) using
RBF presented in equation (1), such that the function has a value of zero on the
surface.

Parametric Child Body Shape Model
6
3. For each vertex in the template, move it onto the surface (zero-valued position) using
the gradient in the implicit surface function.
Statistical Analysis
A principal component analysis (PCA) was conducted using the methods described in Reed and
Parkinson (2008), which are based on methods described by Allen et al. (2003). In brief, the
coordinates of the mesh vertices are flattened to create a 3×30k geometry vector for each fitted
template. The 27 standard anthropometric variable values were appended to each vector along
with the 3D coordinates of 138 surface landmarks and estimated joint center locations, resulting
in a 168×90447 geometry matrix. PCA was conducted on the geometry matrix. To facilitate
computation, the matrix was partitioned into 20 segments and the first 60 PC scores were
retained for each segment, representing 99% of the variance in the geometric data.
Following the methods described in Reed and Parkinson (2008), a linear regression analysis was
conducted to predict principal component scores using stature, body mass index (body mass in
kilograms divided by stature in meters squared), and the ratio of erect sitting height to stature
(SHS) as predictors. Because the interpretation of the regression coefficients is not meaningful,
statistical significance tests were not conducted and all coefficients were retained for the 60 PC
scores.
Results
Figure 5 shows qualitative comparisons between the template model (a), results of each template
fitting process (b, c), and target scan data (d) of a subject.
The SBSM model was built based on the fitted models. A range of body sizes and shapes
generated using the resulting SBSM model is shown in Figure 6. Predictors such as stature (100 ~
160 cm) and BMI (11 ~ 27 kg/m2) vary over approximately the range in the subject pool along the
vertical and horizontal axes of the plot, respectively. The created models readily capture the
different effects of each predictor. Note that the change in body proportions was also observed
along with stature, which is a predominant visual cue to age.

Parametric Child Body Shape Model
7
Figure 7 shows a comparison of template fit for 10 subjects with the reconstruction using SBSM
with 60 PCs. The absolute distances between the two surfaces at each vertex were coded with a
“heat map” corresponding to 0 to 20 mm to assess the error due to limiting the number of PC
scores. Mean (±standard deviation) absolute error was 2.2±0.28 mm, and 95th percentile error
was 4.51±0.57 mm. In most cases, the maximum errors were found in parts of the back side of
head (where hair was covered with a swim cap) and in the crotch area, which was shadowed
from the scanner.
The explanatory power of the SBSM’s anthropometric predictors such as stature, BMI, and SHS
was validated by comparing the scan data with created models based on actual anthropometric
data of the same subjects. Figure 8 shows the result of this comparison. Errors were coded with a
heat map (ranged 0 mm to 50 mm) and were evaluated only for the torso to remove the effect of
limb posture variance. The residual variance of the regression model across subjects was
computed displayed in Figure 9 on the mean figure. Mean error was 10.4±5.8 mm, and 95th
percentile error was 24.0±18.5 mm. Error variance was largest in the shoulders, upper hips, and
crotch, and lowest in the abdomen.
Discussion
Model Generation
This paper presents the first statistical model of body shape in children ages 3 to 11 years as a
function of stature, body weight, and erect sitting height based on high-resolution 3D surface
measurements. A total of 137 subjects’ body shape and a set of surface landmarks and internal
joint centers were measured and standardized to conduct a PCA model. 27 anthropometric
variables were associated with PC scores using a regression model to create models
representative of people with desired body dimensions.
Applications
The SBSM presented here is broadly applicable and can be readily extended to a wide range of
applications. We have applied the model to fitting low-resolution data from Microsoft Kinect,
demonstrating the rapid generation of avatars from Kinect data along with accurate prediction of

Parametric Child Body Shape Model
8
standard anthropometry (Park and Reed 2014). We expect that the model will be useful for
generating anthropometric specifications for human surrogates used for safety applications, such
as crash test dummies and finite-element models used for crash simulation. We expect the model
to be integrated into commercial ergonomics software, such as the Jack human modeling
software, in the same manner as we have demonstrated using adult SBSMs (Reed et al. 2014).
We also anticipate that this or similar SBSMs may be useful for stochastic investigation of the
performance of protective equipment, which currently is performed using a small number of
body forms. Likewise, clothing fitting simulations might be enhanced by the availability of a
well-validated parametric model. By making the model freely available, we hope to stimulate
further interest in accurate parametric representations of human body shape.
Limitations and Future Work
The model is limited by the characteristics of the subject pool. The children were chosen as a
convenience sample distributed by age, and hence body size as shown in Figure 1, but the sample
lacks a large percentage of children with high body mass index. The sample is predominantly
U.S. children with European ancestry and hence may not accurately represent the body shapes of
children from other countries or groups.
The PCA approach is widely used to reduce the dimension of body shape data, but some
variance is lost by retaining only 60 PC scores at each segment. The effect is that of a low-pass
filter that removes high-resolution spatial content. While detailed features in idiosyncratic
aspects of individual body shapes such as facial features and sharp edges were smoothed, the
overall dimensions of the body are well preserved, as demonstrated in Figure 7.
The SBSM presented here includes prediction of selected body landmarks and joint center
locations, but the underlying model does not include a parameterization of pose based on a
kinematic linkage (e.g., Hasler et al. 2009). Although building the model on a single pose
simplifies both the creation and application of the model, we are exploring pose
parameterizations that will be easily applied across human modeling environments.
The underlying statistical methodology is inherently linear; as with other body shape models
created using similar techniques, changes in the values of the predictors move the mesh vertices

Parametric Child Body Shape Model
9
linearly in Cartesian space. Nonlinearity in the predictors can be introduced in the regression
step, but we have not found that nonlinear terms improve the prediction of the PC scores.
Moreover, we have not found that the distribution of prediction error changes in nonlinear ways
for subjects far from the mean. Nonetheless, we are continuing to explore whether methods other
than the PCA and linear regression would produce better predictions of unusual body types.
Acknowledgement
This project was supported in part by the U.S. National Highway Traffic Safety Administration
under a cooperative agreement with the University of Michigan.

Parametric Child Body Shape Model
10
References
Allen, B., Curless, B., & Popović, Z. (2003). The space of human body shapes: reconstruction
and parameterization from range scans. In ACM Transactions on Graphics (TOG) (Vol. 22, No.
3, pp. 587-594). ACM.
Allen, B., Curless, B., & Popovic, Z. (2004). Exploring the space of human body shapes: Data-
driven synthesis under anthropometric control. In Proceedings of Conference on Digital Human
Modeling for Design and Engineering. SEA International.
Ben Azouz, Z., Shu, C, Mantel, A. (2006). Automatic locating of anthropometric landmarks on
3D human models. In: Third International Symposium on 3D Data Processing, Visualization and
Transmission. 750–757.
Carr, J. C., Beatson, R. K., Cherrie, J. B., Mitchell, T. J., Fright, W. R., McCallum, B. C., &
Evans, T. R. (2001, August). Reconstruction and representation of 3D objects with radial basis
functions. In Proceedings of the 28th annual conference on Computer graphics and interactive
techniques (pp. 67-76). ACM.
Daanen, H. A. M., & Ter Haar, F. B. (2013). 3D whole body scanners revisited. Displays, 34(4),
270-275.
Gordon, C. C., Churchill, T., Clauser, C. E., Bradtmiller, B., McConville, J. T., Tebbetts, I., &
Walker, R. A. (1989). 1988 Anthropometric Survey of U.S. Army Personnel: Methods and
Summary Statistics. Final Report. (NATICK/TR-89/027). Natick, Massachusetts: U.S. Army
Natick Research, Development and Engineering Center.
Han, H., Nam, Y., & Choi, K. (2010). Comparative analysis of 3D body scan measurements and
manual measurements of size Korea adult females. International Journal of Industrial
Ergonomics, 40(5), 530-540.
Hasler, N., Stoll, C., Sunkel, M., Rosenhahn, B., & Seidel, H. P. (2009). A statistical model of
human pose and body shape. In Computer Graphics Forum (Vol. 28, No. 2, pp. 337-346).
Blackwell Publishing Ltd.
Lu, J. M., & Wang, M. J. J. (2008). Automated anthropometric data collection using 3D whole
body scanners. Expert Systems with Applications, 35(1), 407-414.
Mochimaru, M., & Kouchi, M. (2000). Statistics for 3D human body forms. In Proceedings of
the Human Factors and Ergonomics Society Annual Meeting (Vol. 44, No. 38, pp. 852-855).
SAGE Publications.
Park, B-K. and Reed, M.P. (2014). Rapid generation of custom avatars using depth cameras.
Proc. 3rd International Digital Human Modeling Conference. Tokyo, Japan.
Reed, M. P., & Parkinson, M. B. (2008). Modeling variability in torso shape for chair and seat
design. In ASME 2008 International Design Engineering Technical Conferences and Computers

Parametric Child Body Shape Model
11
and Information in Engineering Conference (pp. 561-569). American Society of Mechanical
Engineers.
Reed, M.P., Manary, M.A., and Schneider, L.W. (1999). Methods for measuring and
representing automobile occupant posture. Technical Paper 990959. Proceedings of the SAE
International Congress and Exposition, Society of Automotive Engineers, Warrendale, PA.
Reed, M.P., Raschke, U., Tirumali, R., and Parkinson, M.B. (2014). Developing and
implementing parametric human body shape models in ergonomics software. Proc. 3rd
International Digital Human Modeling Conference. Tokyo, Japan.
Robinette, K. M., Daanen, H., & Paquet, E. (1999). The CAESAR project: a 3-D surface
anthropometry survey. In 3-D Digital Imaging and Modeling, 1999. Proceedings. Second
International Conference on (pp. 380-386). IEEE.
Stančić, I., Musić, J., & Zanchi, V. (2013). Improved structured light 3D scanner with
application to anthropometric parameter estimation. Measurement, 46(1), 716-726.
Wells, J. C. K., Cole, T. J., Bruner, D., & Treleaven, P. (2008). Body shape in American and
British adults: between-country and inter-ethnic comparisons. International journal of obesity,
32(1), 152-159.
Yu, C. Y., Lin, C. H., & Yang, Y. H. (2010). Human body surface area database and estimation
formula. Burns, 36(5), 616-629.

Parametric Child Body Shape Model
12
Appendix I
Surface Landmarks and Internal Joint Center Estimates in mm for Mean Figure
Landmark name
x
y
z
Landmark name
x
y
z
Acromion_Ant_Lt_H
-3.7
-136.5
1071.1
LegLower1_Rt_M
8.4
111.9
263.5
Acromion_Ant_Rt_H
-5.1
132.5
1072.3
LegLower2_Lt_M
39.6
-105.4
168.9
AcromionOffsetLt
-6.2
-132.8
1068.3
LegLower2_Rt_M
36.9
112.3
173.1
AcromionOffsetRt
-11.5
127.0
1067.5
LegUpper1_Lt_M
-11.8
-121.8
593.4
AnkleJntLt
30.3
-68.9
72.1
LegUpper1_Rt_M
-19.4
129.4
603.8
AnkleJntRt
30.9
77.1
70.2
LegUpper2_Lt_M
-32.1
-111.5
510.6
ArmLower1_Lt_M
26.7
-271.5
767.4
LegUpper2_Rt_M
-36.8
108.8
520.8
ArmLower1_Rt_M
12.2
267.0
767.7
Malleolus_Lat_Lt_M
35.4
-96.9
70.7
ArmUpper1_Lt_M
18.3
-187.8
980.2
Malleolus_Lat_Rt_M
34.8
105.2
70.3
ArmUpper1_Rt_M
8.4
182.2
985.9
Malleolus_Med_Lt_M
25.5
-40.7
73.7
ArmUpper2_Lt_M
52.3
-181.8
938.6
Malleolus_Med_Rt_M
26.5
49.0
70.2
ArmUpper2_Rt_M
34.7
188.3
938.4
MetaTars1_Med_Lt_L
-85.5
-43.7
19.0
ArmUpper3_Lt_M
33.6
-207.2
906.8
MetaTars1_Med_Rt_L
-82.5
60.6
19.3
ArmUpper3_Rt_M
16.1
205.4
902.1
MetaTars5_Lat_Lt_L
-50.1
-116.3
14.1
ASISBoneLt_Faro
-40.2
-73.2
736.6
MetaTars5_Lat_Rt_L
-50.1
133.7
14.7
ASISBoneRt_Faro
-44.1
83.1
733.5
PSISBoneLt_Faro
62.8
-30.0
771.5
ASIS_Lt_L
-44.9
-73.1
735.0
PSISBoneRt_Faro
57.4
35.6
772.2
ASIS_Rt_L
-48.7
83.2
731.9
PSIS_Lt_L
67.4
-30.0
773.1
Axilla_Ant_Lt_L
-31.2
-109.2
978.7
PSIS_Rt_L
62.0
35.5
773.9
Axilla_Ant_Rt_L
-29.9
92.1
976.4
Rib10_Lt_M
-11.6
-109.5
871.8
Axilla_Pos_Lt_L
50.0
-126.1
969.5
Rib10_Rt_M
-20.7
112.4
876.4
Axilla_Pos_Rt_L
53.1
122.4
972.9
Scapula_Lat_Lt_M
30.1
-130.9
1062.9
BackOfHead_Ct_L
65.5
-0.7
1213.5
Scapula_Lat_Rt_M
23.2
129.9
1063.6
BackOfHeadOnMesh
-36.4
2.2
1294.9
Scapula_Med_Lt_M
69.7
-62.5
1057.5
BackOfHeadReHardSeat
49.5
5.2
1198.3
Scapula_Med_Rt_M
65.1
61.7
1054.7
C7T12Jnt
-7.8
-4.0
1078.5
ShoulderJntLt
-3.2
-132.8
1027.0
Clavicle_Lat_Lt_M
11.9
-123.4
1076.1
ShoulderJntRt
-8.7
127.1
1026.3
Clavicle_Lat_Rt_M
-5.2
80.3
1074.2
SpineC07_Ct_M
46.8
0.1
1086.0
Clavicle_Med_Lt_M
-45.2
-25.5
1064.0
SpineL01_Ct_M
47.8
-0.4
870.5
Clavicle_Med_Rt_M
-38.8
16.1
1065.0
SpineL02_Ct_M
45.4
-1.2
852.9
ElbowJntLt
52.6
-206.5
845.1
SpineL03_Ct_M
43.2
-0.1
834.5
ElbowJntRt
45.1
205.4
840.4
SpineL04_Ct_M
48.4
-0.5
817.7
EyeCenter_Lt_L
-76.5
-33.6
1194.4
SpineL05_Ct_M
53.8
-0.4
803.1
EyeCenter_Rt_L
-82.1
21.6
1191.2
SpineT04_Ct_M
71.6
-0.7
1035.3
EyeCorner_Lt_L
-87.1
-44.2
1197.1
SpineT08_Ct_M
71.0
-0.4
955.8
EyeCorner_Rt_L
-91.6
33.0
1200.0
SpineT12_Ct_M
54.5
0.4
887.5
FemoralEpiCon_Lat_Lt_M
-7.1
-97.8
391.0
Substernale_Ct_M
-105.6
-5.6
937.3

Parametric Child Body Shape Model
13
FemoralEpiCon_Lat_Rt_M
-10.0
100.0
398.5
Substernale_RB
-94.5
-4.9
939.4
FemoralEpiCon_Med_Lt_M
3.5
-16.6
380.2
Suprasternale_Ct_M
-47.7
-6.0
1060.1
FemoralEpiCon_Med_Rt_M
-1.0
22.0
384.2
Suprasternale_RB
-39.8
-2.6
1048.9
FemurTibiaJnt_Lat_Lt_M
-1.8
-98.3
372.0
T12L1Jnt_RB
9.0
-4.7
894.6
FemurTibiaJnt_Lat_Rt_M
-2.5
101.9
375.3
ThighJnct_Lat_Lt_L
-28.0
-121.8
695.1
FemurTibiaJnt_Med_Lt_M
14.0
-20.3
345.2
ThighJnct_Lat_Rt_L
-36.6
117.4
700.6
FemurTibiaJnt_Med_Rt_M
14.8
22.7
348.5
ThighJnct_Med_Lt_L
-49.5
-10.0
614.2
G_AcromionLt_M
-7.9
-134.7
1071.0
ThighJnct_Med_Rt_L
-41.9
7.4
611.8
G_AcromionRt_M
-14.5
128.2
1069.3
TibialHead_Lat_Lt_M
8.6
-104.5
351.8
Glabella_Ct_L
-109.7
-3.7
1214.1
TibialHead_Lat_Rt_M
10.3
108.9
354.0
HandMetCarp2_Med_Lt_L
-41.0
-315.1
596.7
Toe_Ct_Lt_L
-116.0
-53.5
12.8
HandMetCarp2_Med_Rt_L
-40.8
321.5
597.1
Toe_Ct_Rt_L
-122.1
60.8
13.8
HandMetCarp5_Lat_Lt_L
25.1
-309.8
602.1
TopOfHead_Ct_L
-6.6
-21.3
1278.7
HandMetCarp5_Lat_Rt_L
20.0
312.3
597.1
TopOfHeadOnMesh
-6.6
-2.2
1278.7
Head1_Rt_M
-50.4
64.7
1184.6
TopOfHeadReHardSeat
65.5
-4.4
1213.5
Head2_Lt_L
-47.1
-67.0
1180.4
Torso_Ct_Bot_M
-129.4
-4.2
755.3
Head3_Ct_M
-109.2
-3.6
1235.7
Torso_Ct_Mid_M
-133.7
-4.9
827.2
HeadNeckJnt
-21.6
-5.9
1171.2
Torso_Ct_Top_M
-124.8
-5.5
877.2
Heel_Lt_L
71.5
-63.9
84.4
Torso_Lat_Lt_M
-12.0
-110.9
835.5
Heel_Rt_L
74.3
77.1
24.4
Torso_Lat_Rt_M
-21.2
111.5
841.4
HipJntLt_Faro
6.3
-51.8
693.3
Tragion_Lt_L
-20.7
-69.8
1184.4
HipJntRt_Faro
3.6
59.9
691.1
Tragion_Rt_L
-19.9
69.1
1185.0
HumeralEpiCon_Lat_Lt_M
43.3
-225.8
851.7
WristJntLt
1.1
-293.1
669.7
HumeralEpiCon_Lat_Rt_M
34.2
224.9
846.5
WristJntRt
0.4
292.0
664.5
HumeralEpiCon_Med_Lt_M
50.9
-176.5
831.3
Wrist_Lat_Lt_M
29.8
-307.6
670.5
HumeralEpiCon_Med_Rt_M
47.2
172.5
823.8
Wrist_Lat_Rt_M
18.1
301.6
665.9
Iliocristal_Lt_M
-10.9
-122.2
797.9
Wrist_Med_Lt_M
-18.2
-290.5
670.1
Iliocristal_Rt_M
-18.8
118.0
801.9
Wrist_Med_Rt_M
-27.9
291.2
672.6
KneeJntLt
-3.4
-56.5
387.1
WristMidBot_Lt_M
16.7
-267.2
673.3
KneeJntRt
-5.9
60.1
392.0
WristMidBot_Rt_M
-0.8
261.7
661.7
L5S1Jnt_FitLower
-6.3
4.5
771.3
WristMidTop_Lt_M
1.2
-306.2
672.0
LegLower1_Lt_M
13.2
-104.7
259.2
WristMidTop_Rt_M
-1.2
308.4
673.6
APPENDIX II
Standard Anthropometrics for Mean Figure
ANTHROPOMETRIC NAMES
VALUE
AGEATTESTING
8.2
WEIGHTKG
35.3
STATURE
1300.4

Parametric Child Body Shape Model
14
ERECTSITTINGHEIGHT
660.6
SHS
0.5
BMI
20.0
EYEHEIGHT
611.7
ACROMIALHEIGHT
411.0
KNEEHEIGHT
419.3
TRAGIONTOTOPOFHEAD
143.0
HEADLENGTH
182.7
HEADBREADTH
147.1
SHOULDER-ELBOWLENGTH
281.4
ELBOW-HANDLENGTH
359.1
MAXHIPBREADTH
270.9
BUTTOCK-KNEELENGTH
461.9
BUTTOCK-POPLITEALLENGTH
384.4
BIACROMIALBREADTH
272.3
SHOULDERBREADTH
348.8
CHESTDEPTH(SCAPULA)
185.1
CHESTDEPTH(SPINE)
159.6
BIASISBREADTH
176.3
CHESTCIRCUMFERENCE
723.4
WAISTCIRCUMFERENCE
692.2
HIPCIRCUMFERENCE
737.2
UPPERTHIGHCIRCUMFERENCE
457.5

Parametric Child Body Shape Model
15
Figure captions
Figure 1. Distribution of stature and body weight versus age in months in the subject pool.
Growth curves from the U.S. National Center for Health Statistics (2001) are shown for
reference.
0
20
40
60
80
10.0 30.0 50.0 70.0 90.0 110.0 130.0 150.0
Weight (kg)
Age in months
Male Subj
Female Subj
Male 5th
Male 25th
Male 75th
Male 95th
Female 5th
Female 25th
Female 75th
Female 95th
80
100
120
140
160
180
10.0 30.0 50.0 70.0 90.0 110.0 130.0 150.0
Stature (cm)
Age in months
Male Subj
Female Subj
Male 5th
Male 25th
Male 75th
Male 95th
Female 5th
Female 25th
Female 75th
Female 95th

Parametric Child Body Shape Model
16
Figure 2. Scan posture used for the current analysis.
Figure 3. Process to fit a template model to scan data of a specific subject using radial
basis function technique.
Find synthetic landmark points
Build RBF using landmarks
Template model
Landmarks
Vertices
Target scan data
Morph template using RBF
Segment the scan using kd-tree
Convert normal information to
implicit functions using RBF
Compute normal vectors of sampled
vertex in scan data
Push template vertices onto target
surface using implicit functions
1. Initial morphing 2. Fine morphing

Parametric Child Body Shape Model
17
Figure 4. Synthetic landmarks over the body shape
(a) (b) (c) (d) (e)
Figure 5. Results of each template fitting step of a person: (a) template model, (b) initially
morphed template model, (c) fine-morphed template, (d) target scan data, and (e) estimated
surface landmarks and joints.
Top of Head
Chin
Acromion
Bust points
Abdomen
Crotch
Toes
Elbow
Wrist
Hip
Knee
Ankle

Parametric Child Body Shape Model
18
Figure 6. Range of body sizes and shapes. BMI varies along the horizontal axis (11 ~ 27
kg/m2) and stature effects vary along vertical axis (100 ~ 160 cm).
- BODY MASS INDEX +
+ STATURE -

Parametric Child Body Shape Model
19
Figure 7. Comparisons between fitted template models and reconstructed data using 60 PC
scores of each segment. IDs above each image list gender, age at testing (YO), and height
(cm) of the subject. Mean and 95th percentile errors evaluated from the absolute distances
between corresponding vertices are stated below each image. Left is front side and right is
back side.
S001-F-10-145 S002-F-6-120 S005-F-6-117 S006-F-9-146 S007-M-7-128
2.08 mm (4.30 mm) 2.14 mm (4.14 mm) 2.33 mm (4.87 mm) 2.52 mm (4.96 mm) 2.32 mm (4.71 mm)
S009-F-8-124 S010-M-6_109 S011-F-11-149 S012-M-10-148 S013-M-6-120
2.13 mm (4.10 mm) 2.12 mm (4.21 mm) 2.46 mm (5.08 mm) 2.26 mm (4.63 mm) 2.00 mm (4.11 mm)
0 mm 5 mm 10 mm 15 mm 20 mm

Parametric Child Body Shape Model
20
Figure 8. Sampled results of comparisons between fitted template models and
reconstructed data using anthropometric predictors such as stature, BMI and SHS.
Figure 9. Residual variance of the SBSM model aggregated across subjects when only
using anthropometric predictors, illustrated on the mean figure.
S001-F-10-145 S002-F-6-120 S005-F-6-117 S006-F-9-146 S007-M-7-128
7.44 mm (14.2 mm) 10.6 mm (27.5 mm) 8.65 mm (18.4 mm) 14.4 mm (42.5 mm) 16.1 mm (36.4 mm)
S009-F-8-124 S010-M-6_109 S011-F-11-149 S012-M-10-148 S013-M-6-120
8.41 mm (19.2 mm) 12.0 mm (24.2 mm) 9.18 mm (17.3 mm) 9.37 mm (22.6 mm) 7.51 mm (17.6 mm)
0 mm 10 mm 20 mm 30 mm 40 mm 50 mm
50 mm
40 mm
30 mm
20 mm
10 mm
0 mm

Parametric Child Body Shape Model
21