Clothing design similarity determination system and method
A clothing design similarity determination system and method are provided. The clothing design similarity determination method includes: a first step (S 100 ) of collecting image data about clothing using a device ( 1 ); a second step (S 110 ) of removing unnecessary parts from the collected image data using a pre-processing unit; a third step (S 120 ) of image-processing features of the clothing by calculating an extracted target image using a feature calculation unit ( 9 ); a fourth step (S 130 ) of storing the data of the first, second, and third steps (S 100 , S 110 , S 120 ) in a database ( 13 ); and a fifth step (S 140 ) of calculating similarity of clothing design using artificial intelligence. This work was supported by project for Smart Manufacturing Innovation R&D funded Korea Ministry of SMEs and Startups in 2023 . (Project No. RS- 2022 - 00141433 )
1 . A clothing design similarity determination method comprising:
a first step (S 100 ) of collecting image data about clothing using a device ( 1 );
a second step (S 110 ) of removing unnecessary parts from the collected image data using a pre-processing unit;
a third step (S 120 ) of image-processing features of the clothing by calculating an extracted target image using a feature calculation unit ( 9 );
a fourth step (S 130 ) of storing the data of the first, second, and third steps (S 100 , S 110 , S 120 ) in a database ( 13 ); and
a fifth step (S 140 ) of calculating similarity of clothing design using artificial intelligence,
wherein a data processing process in the third step (S 120 ) is performed by the feature calculation unit ( 9 ) including first and second calculation modules ( 15 , 17 ), and
the first calculation module ( 15 ) grasps and identifies features of a clothing image on the basis of Equation 1 below:
R
(
t
)
=
F
(
C
x
(
t
)
,
C
y
(
t
)
,
C
z
(
t
)
,
x
anchor
,
y
anchor
,
z
anchor
)
[
Equation
1
]
wherein x anchor , y anchor , z anchor ; anchor points on X, Y, and Z axes of each clothing,
C x (t), C y (t), C z (t): contour functions in the X, Y, and Z axes, which are obtained from edges of an extracted image, and
F: a function that converts and regresses pieces of input information into an eigenfunction of the clothing.
2 . The clothing design similarity determination method of claim 1 , wherein, in the third step (S 120 ), the second calculation module ( 17 ) calculates features of each part constituting the clothing image calculated by the first calculation module ( 15 ) on the basis of Equation 2 below:
R
s
e
t
=
{
R
1
(
t
)
,
R
2
(
t
)
,
…
,
R
N
(
t
)
}
[
Equation
2
]
K
i
=
{
R
id
×
1
(
t
)
,
R
id
×
2
(
t
)
,
…
,
R
id
×
M
(
t
)
}
,
i
=
1
,
…
,
U
wherein R i (t): equation obtained from Equation 1,
R set : a combination thereof, and
K i : clothing feature with i-th feature.
3 . A clothing design similarity determination system comprising:
a device ( 1 ) configured to obtain and control clothing images; and
a management server ( 3 ) configured to determine similarity by processing clothing design data transmitted from the device ( 1 ) through a network,
wherein the management server ( 3 ) comprises:
an input unit ( 5 ) configured to receive clothing image data from the device ( 1 );
an image extraction unit ( 7 ) configured to extract target image data through pre-processing performed by removing unnecessary parts from the collected clothing image data;
a feature calculation unit ( 9 ) configured to calculate features of clothing using an equation from the extracted target image data and process the calculated features into a curved shape;
a similarity calculation unit ( 11 ) configured to calculate similarity of clothing data calculated in the feature calculation unit ( 9 ) with respect to other clothing designs using artificial intelligence; and
a database ( 13 ) in which data is stored,
wherein the feature calculation unit ( 9 ) comprises:
a first calculation module ( 15 ) configured to grasp and identify the features of the clothing image on the basis of Equation 1 below; and
a second calculation module ( 17 ) configured to calculate the features of each part constituting the clothing image calculated by the first calculation module ( 15 ):
R
(
t
)
=
F
(
C
x
(
t
)
,
C
y
(
t
)
,
C
z
(
t
)
,
x
anchor
,
y
anchor
,
z
anchor
)
[
Equation
1
]
wherein x anchor , y anchor , z anchor ; anchor points on X, Y, and Z axes of each clothing,
C x (t), C y (t), C z (t): contour functions in the X, Y, and Z axes, which are obtained from edges of an extracted image, and
F: a function that converts and regresses pieces of input information into an eigenfunction of the clothing.
4 . The clothing design similarity determination system of claim 3 , wherein the second calculation module ( 17 ) calculates features of each part constituting the clothing image calculated by the first calculation module ( 15 ) on the basis of Equation 2 below:
R
s
e
t
=
{
R
1
(
t
)
,
R
2
(
t
)
,
…
,
R
N
(
t
)
}
[
Equation
2
]
K
i
=
{
R
id
×
1
(
t
)
,
R
id
×
2
(
t
)
,
…
,
R
id
×
M
(
t
)
}
,
i
=
1
,
…
,
U
wherein R i (t): equation obtained from Equation 1,
R set : a combination thereof, and
K i : clothing feature with i-th feature.