The study of pattern auto generation system based on silk fabric properties

1. The Study of Pattern Auto-generation System Based
on Silk Fabric Properties
Yanzhen Wang
College of Fashon
Donghua University&Shanghai University of Engineering
Science
Shanghai, China
Wangyanzhen1227@126.com
JianPing Wang
College of Fashon
Donghua University
Shanghai, China
wangjp@dhu.edu.cn
Abstract—To study pattern Auto-generation system based on
silk fabric properties, it is proposed a concept about rapid
generation of individualize pattern. By carrying out instrumental
tests on the 3-silk fabric properties and sample wear trials, the
regression mathematical models between fabric properties and
parameter pattern (the easing of sleeve cap) were established in
this dissertation. Using the mathematical models of the
parametric pattern and taking silk clothing as examples, it
described the realization process of pattern ato-generation
system , which was based on physical properties of the silk fabric.
Index Terms—silk fabric, intelligentize, pattern,
parameterization.
I. INTRODUCTION
Compared with existing clothing CAD systems, intelligent
fashion system is a kind of breakthrough and innovation, a kind
of attempt and explore, too [1]. However, there are some
bottlenecks that design is isolated from pattern making and
intelligent realization of pattern is not completely which have
influenced and restricted efficiency and quality of garment’s
production [2].
It has attained to some degree of the achievements in China
and abroad about the field of intelligent garments system,
which are belong to the input system of made to measure and
can choose the most appropriate pattern according to the
selected style[3]. The research in Pattern Auto-generation
Expert System is to establish expert knowledge base, inference
mechanism, intelligent algorithm and mathematical model to
optimize the 2D structure and pattern of the garment on the
basis of this[4,5]. Meanwhile former researches mainly
western-style clothes and shirt and no research to the silk style.
According to the different relation of suitable between the
armhole and sleeve cap of silk garment, the paper first tests 3-
representative silk fabric properties and analyze the sleeve
pattern to establish two mathematical model (fabric property
and the easing of sleeve, sleeve top height and cap). Take the
advantage of pattern auto-generation and interactive technology,
combined with the silk garments’ characteristic to complete
parametric analyze of the armhole and cap, build the
corresponding mathematical mode to realize the intelligent
operation for the silk garment.
II. SILK FABRIC PROPERTY AND PATTERN REGRESSION
MODEL
A. Choose and Property of Silk Fabric
The silk was divided into fourteen categories according to
silk fabric weave, meridian and parallel’s combination,
processing technic and expression appearance. Experiments for
fabric property are carried out by selecting three sets of light,
middle and heavy fabrics, such as organza, Sateen and double
Crepe de Chine. The experiments utilized the Siro-Fast fabric
style tester to obtain four physical performances, such as fabric
weight, thickness, bending rigidity and extension. Table1
shows three silk fabric’s physical properties.
TABLE I. SILK FABRIC PHYSICAL PROPERTIES TABLE
Proper
ty
Values
Fabric Varieties
Organza Sateen
Double Crepe de
Chine
Weight
(g/100c
m2
)
0.3581 0.6535 0.8186
Thickn
ess(m
m)
0.122 0.212 0.22
Bendin
g
rigidity
(
N.m)
Ac
ros
s
wa
rp
right
side
758.7996
10-6
1628.5638
10-6
2955.9591 10-
6
wron
g
side
584.4689
10-6
801.3544
10-6
24481.8794
10-6
W
arp
-
wi
se
right
side 10-6
10-6
10-6
wron
g
side 10-6
10-6
10-6
45
-
de
gre
e
right
side 10-6
10-6
10-6
wron
g
side 10-6
10-6
10-6
Extensi
on
Ac
ros
s
wa
rp
5 0 0.466667 0
20 0.033333 2.2 0.733333
100 1.033333 5.633333 1.833333
- 419 -

2. Proper
ty
Values
Fabric Varieties
Organza Sateen
Double Crepe de
Chine
W
arp
-
wi
se
5 0.066667 0.1 0
20 1.133333 1.6 1.066667
100 4.266667 5.033333 3.7
45
-
de
gre
e
5 6.966667 7.1 8.266667
20 10.6 11.83333 13.56667
100 15.43333 17.6 19.06667
B. Silk samples experiment
In order to determine the most suitable sleeve casing value
for each silk fabric, and obtain the mathematical relationship
between sleeve cap curve length and sleeve cap height, so that
these data can be performed the regression model. The
experimental samples have one armhole and the sleeve casing
value from 0cm to 0.2cm, every fabric have 17samples. Table
II shows the sleeve cap height and sleeve cap curve length got
by experiments.
TABLE II. THE CORRESPONDING VALUES BETWEEN SLEEVE TOP HEIGHT
AND CAP LINE
Sleeve
Pattern
Number
Table Column Head
Sleeve
Pattern
Number
Table Column
Head
Cap
line
Sleeve top
height
Cap
line
Sleeve top
height
Sv01 46.93 14.51 Sv10 48.73 15.69
Sv02 47.13 14.64 Sv11 48.93 15.87
Sv03 47.13 14.77 Sv12 49.13 15.98
Sv04 47.53 14.92 Sv13 49.33 16.12
Sv05 47.73 15.05 Sv14 49.53 16.24
Sv06 47.93 15.20 Sv15 49.73 16.38
Sv07 48.13 15.32 Sv16 49.93 16.52
Sv08 48.33 15.46 Sv17 50.13 16.66
Sv09 48.53 15.56
C. The Regression Model of Silk Fabric Properties and Sleeve
Casing
This paper used the regeression model to express the
relationshhip between fabric properties and sleeve casing.
Equation 1 is the Linear regression equation. The regression
equation between dependent variable Y and independent
variable X1, X2, . . ., Xm is as follows.
HEED

3. D, β1…βm :estimating parameter
ε error of random variables
According to Equation 1 in the multivariate regression
model, combined with Table I shows the relationship between
fabric properties and casing value, compared by two physical
properties of the regression coefficient with significance test
results, using T test. From the results it can be seen, in models
of all independent variables significant values are less than 0.01,
the constant level value of 0.012, but less than 0.02, it shows
that in 0.02 significance level, constant significantly different
from 0, so the regression model is as following.
Sleeve Casing Value =-4.588+ 0.471 × 45 –degree extension.
Similarly, obtain the model of sleeve cap curve length and
sleeve cap height:
The sleeve cap curve length =25.238+1.495 × sleeve cap height.
III. CONSTRUCTION OF PARAMETRIC PATTERN MATHEMATIC
MODELS AND ALGORITHM
A. Construction of Parametric Pattern Mathematic Models
when in pattern drawing, it is the work for built the
relationship between any conventional numbers and pattern
sizes, if a new size of pattern is designed, just modify its data,
this method is parameterized design in the garment pattern
drawing system. curve is the difficult and key point in this
system, to obtain the automatic system and modify it is that the
computer display curve is similar with actual curve, and at the
same time the pattern modified by size is consistent with the
actual modification.
1) Point: Point A in the plane is defined by the coordinate A
(x y) (see Fig.1) .
2) Line: A1(x1, y1), A2(x2, y2) are given in the two
dimensional space. Introducing to parameter μ, then the
parametric equations of a line will be x=x1 (x2 x1) μ y
y1 (y2 y1) μ.
Fig. 1. Point and line.
3) Curve: There are many ways of establishing parametric
mathematical model by Curve such as Spline Curve, Bezier
Curve, B-Spline Curve and Gamma Curve, Two-Bezier curve
and Three-Bezier curve are mainly used in the research.
The definition of Bezier Curve[6]:
Consider N+1 control points Pi ( i =0, 1,…, n),called N-
Bezier curve and the N parametric curve
¦ 0 t 1 (2)
- 420 -

4. P0,P1,…,Pn are the data points Pi( i=0,1,2,… n are the
coordinates of n+1 points which control the polygon. Control
Polygons is a Polygon which is formed by connecting n sides.
(3)
Equation 3 was Called Bernstein Basis Functions, Cn
1
is
combination number, C1
0
= C2
0
= C3
0
= C1
1
= C1
2
= C1
3
= 1, C2
1
=2,
C3
1
= C3
2
=3.
When establishing curve’s mathematic model, we also
used two-Bezier curve and three-Bezier curve, which are given
in Fig. 2.
Fig. 2. Curve.
B. The application of Mathematic Models
1) Body parametric design: The curve combined curvature
is the armhole curve, so that it is can directly using the method
of three spline curve combined with two cumulative chord
length parameters, and it is also can using the B spline curve.
it is consider that one line using one curve can be controlled
easier, so using B spline curve. Front and behind armhole are
both controlled by 4 points.Table III and Table IV show the
front and behind garment piece corresponding coordinates.
TABLE III. POINT COORDINATES PARAMETER OF FRONT PIECE
Point Coordinate and Front picture
coordinate Front Picture
A (0,14.07)
B (-7.4,22.17)
C (-19.03,17.5)
D (-18.4 -17)
E (-17.55,13.1)
F (-16.7,6.72)
G (-24,0)
H (-24.83,-3.61)
I (-24,-7.1)
J (-24,18.78)
K (0,-18.78)
2) Sleeve parametric design: In the sleeve module, it is
important how to decide the sleeve easing (sleeve easing refers
to the amount of sleeve cap curve and armhole curve length
difference), its size affects shoulder modeling directly. Fabric
diagonal extension affects sleeve easing, which was increased
by the fabric extension. sleeve casing=-4.588+0.471
×45°extension, sleeve cap curve length=25.238+1.495 × sleeve
cap height.
Sleeve structure line are major minor angle curve and large
angle curve, front and behind sleeve seam are belong to minor
angle curve, top and under sleeve’s sleeve cap curve is belongs
to large angle curve. Three spline curve algorithm was used to
draw first curve, at the same time two Bezier curve algorithm
was used to fit the latter curve. However, because the angle of
front and behind sleeve seam are not very small, the Beizier
curve was used to draw the sleeve cap curve, controlled by 5
points through many iterations. Table V shows the sleeve
corresponding coordinates.
TABLE IV. POINT COORDINATES PARAMETER OF BACK PIECE
Point Coordinate and Back picture
coordinate Back Picture
L 10,22.67
M
17.6,25.2
N 30,21.17
O
28.92,9.3
P 28,8
Q
30.5,2.05
R 34,0
S
34,-
15.33
T
10,-
15.33
TABLE V. POINT COORDINATES PARAMETER OF SLEEVE
Point Coordinate and Sleeve picture
coordinate Sleeve Picture
U -18,35
U
1
-18,45
V 16,35
V
1
16,45
X 0,59.77
Y 0,60.2
Z 0,60.69
IV. AUTO-GENERATION SYSTEM BASED ON SILK FABRIC
CHARACTERISTICS
A. System Framework
Auto-generation system was established in the cognition
and expression of design thought, intelligent interface
technology, knowledge base, intellective model conversion.
Fig3 shows the system framework.
- 421 -

5. Fig. 3. System Framework
Auto-generation pattern got body data by the non-contact
3D body scanner on the drawing principle, and integrated the
related data. System can obtain the pattern with corresponding
suitable size with the help of changing the parameter data, and
use coordinate variables x, y to achieve the whole graph line,
curves and points description.
B. The Realization of Auto- generation system based on the
Silk Fabric Characteristics
The system has three interfaces, Interface two support the
dynamic input sleeve values. A sleeve width is 34cm, sleeve
cap height is from 14.51cm to 20cm, sleeve casing is in the
range from 0cm to 3cm. When the input is over the preset
value, it will jump put the window unless the input value is
right. In the interface two, there were the mathematical
equation of sleeve casing and oblique extension 45°for users to
use compute easily. After users input the values, click the
mouse can drawing sleeve pattern.
Interface three clicks each fabric to jump its corresponding
pattern on the basis of three silk fabric pattern drawing.
V. CONCLUSION
A concept was proposed about parametric garment pattern
based on the parametric garment pattern mathematical model.
It built silk parameterized pattern and initially realized auto-
generation system based on the silk fabric characteristics. In
conclusion, it is a new problem of auto-generation pattern, and
not existing patterns. Overall, garment pattern automatic
generation is a new issue, no ready-made model and need to be
studied further. It can be seen from soft application that
garment pattern automatic generation system will be realized, it
will be not only completed and developed continuously but
also has wide application prospects with the rapid development
of garment CAD and computer technology.
ACKNOWLEDGMENT
This work was financially supported by the College
students' innovative activity plan in Shanghai University of
Engineering Science (Cs1109018).
REFERENCES
[1] Wu Jun. Setting up of a mathematical model for pattern design of
trousers[J].Shanghai Textile Science Technology 2004,32(2):45-46
[2] LiuYan, GengZhaofeng,LiuXiaogang.Applications of Artificial
Intelligence in Garment Industry[J]. Journal of Donghua University,
Natural Science; 2002 28(4):124-127
[3] Song Huanhuan.A Research on Pattern Making System for Men
[D].Shanghai.Shanghai University of Engineering Science ,2007:35-37.
[4] Zhang Qi;Zhang Wenbing;Zhang Weiyuan .Study on Pattern -
autogeneration Expert System of Garment PDS[J].JOURNAL OF
CHINA TEXTILE UNIVERSITY 2000 10 62-65.
[5] Zhang Heng ,Zhang Xin.A research on the block pattern automatic
creation method suitable for garment cut according to the
figure[J].Knitting Industries 2005 12 29-31
[6] Bidarra R, Bronsvoort WF. Semantic feature modelling. Comput Aided
Des 2000;32(3):201–25.
begin
sleeve cap height and casing choose your fabric
drawing the pattern automatically
move
pattern drawing
based on data
back
pattern drawing
based on fabric
back
- 422 -