Image Processing Reference
In-Depth Information
From Equation 9.18, a wide class of operators can be defined with different
values of λ. These are
n
1.
λ
n
() (
x
=− ≥
1
x
) ,
n
1
(9.21)
This is equivalent to Zadeh's concentration operation on linguistic hedges:
1
11 1
−
x
nx
n
2.
λ
n
x
()
=
(( ))
,
≥
1
(9.22)
+
/
−
This is equivalent to Sugeno's λ complementation:
1
x
ln()
ln()
3
2
158
3.
λ
n
()
x
=
−
,
n
≥
=
.
(9.23)
n
1
+
x
2
n
1
x
e
−
1
ln( ln((
−
e
12
12
+
)
/
e
))
4.
λ
n
()
x
=
,
n
≥
=
1 396
.
(9.24)
e
−
1
ln(
/
)
Equations 9.21 and 9.22 are widely used. With different selections of λ(
x
) and
with different values of '
n
', different eroded and dilated images are obtained.
9.3.3 Fuzzy Morphology Using
t
-Norms and
t
-Conorms
by De Baets and Kerre and Bloch and Maitre
DAx
5
() sup(
=
i By xAy
⎣
−
), ()
⎦
yS
∈
(9.25)
(
)
EAx
5
() inf
=
u cByxAy
⎣
⋅
(
−
), ()
⎦
yS
∈
where
i
is any
t
-norm (fuzzy intersection)
u
is any
t
-conorm (fuzzy union)
c
is a complement
A
is an image
B
is a structuring element
Using different
t
-norms (
T
) and
t
-conorms (
T
*), different forms of erosion
and dilation can be generated:
1. Zadeh
t
-operators
*
max( ,)
T
=
x y
T
=
min( ,)
x y
Search WWH ::
Custom Search