US 7,876,964 B2Granted

Method for associating a digital image with a class of a classification system

AssigneeBetriebsforschungsinstitut VDEH-Institut Für Angewandte Forshung GmbH

InventorGünther Coen

Life of this patent

Estimating the term…

Loading patent…

Abstract

Disclosed is a method for assigning the content of a digital image to a class of a classification system. Said method comprises the following steps: —a predetermined number of F numerical shape characteristics ψm are determined; —the value of each shape characteristic of the F numerical shape characteristics determined for the image is compared to the value filed in a table for the respective shape characteristic, values for the individual numerical shape characteristics being allocated to each class in the table; —the class in which the F numerical shape characteristics determined for said image correspond best to the values of the numerical shape characteristics indicated in the table for said class is output as the class into which the image that is to be recognized is classified.

A method for associating the content of a digital image with a class of a classification system, wherein the image is represented by N pixels, each pixel being located at the position (xi, yj) in a predetermined coordinate system, said image extending from the coordinates (0, 0) to (ximax, yjmax) and imax being the maximum number of pixels in the direction of the x-coordinate and jmax is the maximum number of pixels in the direction of the y-coordinate and wherein at least one numerical content attribute data [j, i] is associated with each pixel, said method comprising the steps of:

determining at least one group of a predetermined number of F numerical shape attributes ψm with m as a running index, wherein ψm is a transformed expression of the moment ρm, and ρm is derived from
ρ
m
_
=
k
m
⁢
R
m
_
with
⁢
⁢
k
m
=
(
m
+
2
)
2
⁢
(
π

A

)
m
2
A
=
m
0
,
0
=
Δ
⁢
⁢
a
⋆
Δ
⁢
⁢
b
⋆
∑
j
=
1
j
⁢
⁢
max
⁢
∑
i
=
1
i
⁢
⁢
max
⁢
data
⁡
[
j
,
i
]
R
m
_
=
υ
m
υ
0
=
υ
m
m
0
,
0
υ
m
=
Δ
⁢
⁢
a
⋆
Δ
⁢
⁢
b
⋆
∑
j
=
1
j
⁢
⁢
max
⁢
∑
i
=
1
i
⁢
⁢
max
⁢
(
R
⁡
(
j
,
i
)
)
m
⁢
data
⁡
[
j
,
i
]
R
⁡
(
j
,
i
)
=
(
(
i
-
0
,
5
)
⋆
Δ
⁢
⁢
a
-
x
_
)
2
+
(
(
j
-
0
,
5
)
⋆
Δ
⁢
⁢
b
-
y
_
)
2
x
_
=
m
1
,
0
m
0
,
0
y
_
=
m
0
,
1
m
0
,
0
m
1
,
0
=
(
Δ
⁢
⁢
a
)
2
⋆
Δ
⁢
⁢
b
⋆
∑
j
=
1
j
⁢
⁢
max
⁢
∑
i
=
1
i
⁢
⁢
max
⁢
(
i
-
0
,
5
)
⋆
data
[
j
,
i
]
m
0
,
1
=
(
Δ
⁢
⁢
b
)
2
⋆
Δ
⁢
⁢
a
⋆
∑
j
=
1
j
⁢
⁢
max
⁢
∑
i
=
1
i
⁢
⁢
max
⁢
(
j
-
0
,
5
)
⋆
data
[
j
,
i
]
wherein
Δa=width of the pixel in the x-coordinate direction,
Δb=width of the pixel in the y-coordinate direction,
data [j, i]=content attribute of the pixel at the position (yj, xi)
m=a number running from 1 to F as a counter of the shape attributes;
comparing the value of each shape attribute of the F numeric shape attributes intended for the picture in the at least one group with the value saved in a table for the respective shape attribute of this group, wherein values for the individual numerical shape attributes of this group in the table are associated with each class;
and
outputting the class as an association class in which the image to be recognized is classified, in which the F numeric shape attributes intended for the picture best correspond to the values of the numeric shape attributes recorded in the table for this class.

310 more elements

Abstract

Claims22