Database system
The use of an on-line anthropometric database system for morphotype analysisand
sizing system adaptation for different world market apparel sportwear
WEAR Conference- Banff - Canada- July 31/August 1st, 2007
Régis Mollard
Paris Descartes UniversityBiomedical Research Center
Ergonomics-Behavior & Interactions (EA 4070)Laboratory of Applied Anthropology
45 rue des Saints-Pères75270 PARIS Cedex 06 - FRANCE
Local Area Networks
Web
Web
Wear End Users
Database 1
Database n
Database 3
Database 2
WEAR will bea distributed on-line Database system
Wear members
Database
Principle of a Database System
Synthesis sheets
Dictionary ofmeasurements
Aggregated Data(statistics)
Individual Data(raw data)
Bibliographical Data
1-D
3-D
Demographic data+
Quality Evaluation of
Anthropometric Data+ xxx xxxx
Data files
Additional files
Data processingDigital man-models Shape analysis Fit tests
+xx
SortingQuery
Pilotes Français
Jeune populationmilitaire
Années
(hommes)
Stature (cm)
1940 1950 1960 1970 1980 1990 2000 2010 2020
Evolution of the stature. Mean values for two french populations from 1940 to 1991.
Prediction up to 2020.
660 760 860 960460
560
660
Buttock - Knee Length (mm)
Eye - Seat Height (mm)
ANGLO-SAXON COUNTRIES
SCANDINAVIAN COUNTRIES
WESTERN EUROPE
ORIENTAL COUNTRIES
PROB. 05
Choice of well-adapted measurements
.
Eye height sittingButtock-knee length
Percentiles5%min.
5%max.
50%min.
50%max.
95%min.
95%max.
Buttock-knee (mm)
Eye seat(mm)
661.7
624.0
586.3
548.5
510.8
473.1
677.
9
728.
0
778.
0
828.
0
878.
1
928.
1
Choice of typical human body models using bivariate distributions.Example to create small, medium and large digital man models
Databases ApplicationsExamples of 1-D anthropometric data processing
using Databases of WEAR
Shape analysis
Survey 2
Survey 1
Survey n
//// Principal Component Analysis (PCA) &Hierarchical Classification
The use of an on-line anthropometric database system for :
Example 1 : Morphotypes analysis
Example 2 : Sizing system adaptation for different world market apparel sportwear(from France to USA & China for bathing suits, pants, jackets,…)
22,12,22,32,42,52,62,72,82,93
500 550 600 650 700 750 800
S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 out Totalout 1 6 3 13 14 18 55S9S12 1 2 8 20 32 16 6 2 87S8 1 2 7 37 93 70 31 9 3 253S7 1 2 9 23 101 226 208 73 28 4 675S6 2 5 34 142 316 445 162 37 8 1 1152S5 3 14 44 240 493 720 368 94 15 2 1 1994S4 2 7 33 137 509 767 453 169 25 5 5 0 2112S3 1 8 75 219 499 363 171 62 6 3 2 1 1410S2 3 12 69 161 217 117 88 17 8 2 1 695S1 0 12 47 96 157 69 21 16 4 1 423S min 1 6 37 49 21 6 2 2 1 125
7 48 278 713 1687 1983 1881 1351 627 241 105 37 23 8981
Coverage / FitBivariate distribution / sizing
Log values
• Choice of a survey
Step 1
• Principal Component Analysis (PCA)
Step 2
• Hierarchical Classification
Step 3
• Interpretation and Morphotypes
Step 4
Example 1 : Morphotypes analysis
Survey 2Survey 1 Survey n
////
1 - Shortlisting of Surveys
2 - Search subjects corresponding to criteria
3 - Statistical calculationsmean, standard deviation, min, maxcoefficient of variation (Pearson)percentilescorrelationsGauss testcontingency tablehistogramplot of individual databivariate distributionsmorphological profilecomparison…
4 - Save the query
Men, Women, Aged >20France,…...
All surveysNorth America + Europe
Example using ERGODATA
Database Query
Sample 1 : French Army 1990 - n = 150
Step 1 : Database
38 measurements
150 subjects
Step 2 : PCA
Eigenvalues and percentages of the variance of the axes.
The first two axes explain 66% of the total variance
Morphotypes analysis for Females
Individual Data
Step 2 : PCA
21 measurements explain 65.8% of the variance of the axis 1.
The coordinates of the measurements on axis 1 are all positive. Axis 1 is a « size factor »
Step 2 : PCA
16 measurements explain 65.2% of the variance of the axis 2.
In axis 2, there is an opposition between height and width measurements
Sample 1 : French Army 1990 - n = 150Morphotypes analysis for Females
Step 2 : PCA
28 subjects explain 66.2% of the variance of the axis 1.
In axis 1, there is an opposition between Small&Slim and Tall&Wide
Step 2 : PCA
35 subjects explain 69.1% of the variance of the axis 2.
In axis 2, there is an opposition between Small&Wide and Tall&Slim
Sample 1 : French Army 1990 - n = 150Morphotypes analysis for Females
Step 3 : Hierarchical classification
Partition in 8 groups explains 69.7% of the total variance.
Sample 1 : French Army 1990 - n = 150Morphotypes analysis for Females
G1 G2 G3 G4 G5 G6 G7 G8
161.7 cm
64.7 kg
155.9 cm
58.4 kg
168.3 cm
64.4 kg
172.2 cm
73.2 kg
163.4 cm
56.7 kg
161.7 cm
50.7 kg
167.4 cm
53.2 kg
153.7 cm
47.1 kg
Step 4 : Morphotypes of the 8 groups
Sample 1 : French Army 1990 - n = 150
Small & Slim
Tall & Wide
Morphotypes analysis for Females
Medium & Large
Step 4 : Morphotypes of the 7 groups
Sample 2 : French Army 1990 - n = 275
Small & Slim
Tall & Wide
167.6 cm
60.9 kg
174.0 cm
58.8 kg
177.6 cm
66.1 kg
169.2 cm
71.8 kg
176.1 cm
73.0 kg
182.6 cm
77.4 kg
179.9 cm
94 kg
G1 G2 G3 G4 G5 G6 G7
Morphotypes analysis for Males
• Identify key measurements for garments and
convert value in log-value if necessary
Step 1
• Confirm and/or adjust the existing
french sizing for the updated data
Step 2
Step 3
Step 4
• Estimate the percentage of fit
• Adapt french sizing to other
populations - USA and China
Example 2 : Sizing system adaptationfor different world market apparel sportwear
22,12,22,32,42,52,62,72,82,93
500 550 600 650 700 750 800
S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 out Totalout 1 6 3 13 14 18 55S9S12 1 2 8 20 32 16 6 2 87S8 1 2 7 37 93 70 31 9 3 253S7 1 2 9 23 101 226 208 73 28 4 675S6 2 5 34 142 316 445 162 37 8 1 1152S5 3 14 44 240 493 720 368 94 15 2 1 1994S4 2 7 33 137 509 767 453 169 25 5 5 0 2112S3 1 8 75 219 499 363 171 62 6 3 2 1 1410S2 3 12 69 161 217 117 88 17 8 2 1 695S1 0 12 47 96 157 69 21 16 4 1 423S min 1 6 37 49 21 6 2 2 1 125
7 48 278 713 1687 1983 1881 1351 627 241 105 37 23 8981
Step 1 : Identify key measurements for garments and convert value in log-value if necessary
Sizing system adaptation
Univariate distribution
The Stature and lengths have a gaussian distribution
The Weight and some perimeters (waist circumference,…) have not a gaussian distribution
Log
Step 1 : Identify key measurements for garments and convert value in log-value if necessary
Sizing system adaptation
Bivariate distribution
Values are converted in log when the distribution is not gaussianMostly the case for weight and related measures (perimeters)
Step 2 : Confirm and/or adjust the existing french sizing for the updated data
Sizing system adaptation
France - Bathing suits - Women
2
2,1
2,2
2,3
2,4
2,5
2,6
2,7
2,8
2,9
3
500 550 600 650 700 750 800
Trunk height (mm)
Pel
vis
peri
met
er (
log
valu
es)
Sizes notadapted
Need to adjust
Step 2 : Confirm and/or adjust the existing french sizing for the updated data
Sizing system adaptation
France - Jackets - Women
Sizes notadapted
Creat new sizes
2,1
2,2
2,3
2,4
2,5
2,6
2,7
2,8
2,9
3
3,1
420 470 520 570 620 670
Arm length(mm)
Pel
vis
peri
met
er (
log
valu
es)
1,5
1,7
1,9
2,1
2,3
2,5
2,7
2,9
3,1
1,5 1,7 1,9 2,1 2,3 2,5 2,7 2,9 3,1
Pel
vis
peri
met
er (
log
valu
es)
Thorax perimeter (log values)
Creat new sizes
Step 4 : Adapt french sizing to other populations -Sizing system adaptation
USA - Bathing suits - Women
Need to adjust
2,1
2,2
2,3
2,4
2,5
2,6
2,7
2,8
2,9
3
3,1
2,1 2,2 2,3 2,4 2,5 2,6 2,7 2,8 2,9 3
Pel
vis
peri
met
er (
log
valu
es)
Thorax perimeter (log values)
Delete 6 sizes
Step 4 : Adapt french sizing to other populations -Sizing system adaptation
China - Bathing suits - Women
2 new sizes
600
700
800
900
1000
1100
1200
1300
1400
700 800 900 1000 1100 1200 1300 1400 1500
Thorax perimeter (mm)
Pel
vis
peri
met
er (
log
valu
es)
Delete 4 sizes
Step 4 : Adapt french sizing to other populations -Sizing system adaptation
China - Jacket - Men
Step 4 : Adapt french sizing to other populations -Sizing system adaptation
China - Pants - Women
1. Delete 4 sizes
2,1
2,2
2,3
2,4
2,5
2,6
2,7
2,8
2,9
3
1,3 1,5 1,7 1,9 2,1 2,3 2,5 2,7 2,9
Natural waist perimeter (log values)
Pel
vis
peri
met
er (
log
valu
es)
2. Adjust the others
3. To creat a new sizing system
Synthesis
- Projects from manufacturers or apparel industry can beimproved using an on-line database system as WEAR
- From classical 1-D values as well as 3-D surface data extracted from different surveys it is possible to identify thedifferences of morphology according to the needs expressed in projects : design of equipments (mask, helmet, goggles,…), garments or workplaces,…..
Using WEAR database system,these results will be obtained very quicklyand all the data, methods, ergonomic rules, …will have been validated by the WEAR group