
255Formal Design of an Eye Disease Classification System
Id k
IxyIxky
NM kIxy
y
N
x
Mk
2
11
0
()
=
()
−+
()
−
()
−
()
=
==
−
,,
,count of
(()
∀
()
≠
()
=
()
−++
()
−
()
−
=
−
=
−
∑∑
,,
,,
Ixy
Id k
IxyIxkyk
MkN
y
Nk
x
Mk
0
3
11
kkIxy
Ixy
Id k
Ixny Ix kn
()
−
()
=
()
∀
()
≠
()
=
−+
()
−+
count of ,
,,
,,
0
0
1
4
++− +
()
−
()
−
()
−
()
=
()
∀
=
−
=
−
∑∑
1
0
11
yk
MkNk Ixy
I
y
Nk
x
Mk
count of ,
,
xxy
ID k
Id k
kmn
j
j
,
,,,,
≠
()
=
()
∀= …×
()
−
=
∑
0
01 1
1
4
(14.16)
where all pixel pairs calculated in the absolute difference (Id
i
(k) with i ∈ {1, 2, 3, 4}) are along
horizontal, vertical, diagonal, and asymmetric-diagonal directions, respectively. A func-
tion f(k) is dened, which describes normalized intensity difference vectors. These vectors
are normalized ...