T
E
L
KO
M
N
I
KA
T
e
lec
om
m
u
n
icat
ion
,
Com
p
u
t
i
n
g,
E
lec
t
r
on
ics
an
d
Cont
r
ol
Vol.
18
,
No.
2
,
Apr
il
2020
,
pp.
7
40
~
75
1
I
S
S
N:
1693
-
6930
,
a
c
c
r
e
dit
e
d
F
ir
s
t
G
r
a
de
by
Ke
me
nr
is
tekdikti
,
De
c
r
e
e
No:
21/E
/KP
T
/2018
DO
I
:
10.
12928/
T
E
L
KO
M
NI
KA
.
v18i2.
10714
740
Jou
r
n
al
h
omepage
:
ht
tp:
//
jour
nal.
uad
.
ac
.
id/
index
.
php/T
E
L
K
OM
N
I
K
A
Fe
at
u
r
e
e
xt
r
ac
t
io
n
of
J
a
b
on
(
A
n
t
h
oc
e
p
h
al
u
s sp
)
l
e
af
d
is
e
as
e
u
si
n
g
d
is
c
r
e
t
e
w
av
e
le
t
t
r
a
n
sf
o
r
m
F
e
ll
ik
s
F
e
i
t
e
r
s
T
am
p
in
on
gk
ol
1
,
Ye
n
i
Her
d
iyeni
2
,
E
li
s
Nin
a
Her
li
yan
a
3
1,
2
D
e
p
art
me
n
t
o
f
C
o
mp
u
t
er
Sci
e
n
ce,
Bo
g
o
r
A
g
ri
c
u
l
t
u
ra
l
U
n
i
v
er
s
i
t
y
,
In
d
o
n
es
i
a
3
D
ep
ar
t
men
t
o
f
Si
l
v
i
cu
l
t
u
re,
Bo
g
o
r
A
g
r
i
cu
l
t
u
ral
U
n
i
v
er
s
i
t
y
,
I
n
d
o
n
e
s
i
a
Ar
t
icle
I
n
f
o
AB
S
T
RA
CT
A
r
ti
c
le
h
is
tor
y
:
R
e
c
e
ived
J
ul
31
,
2019
R
e
vis
e
d
J
a
n
17
,
2020
Ac
c
e
pted
F
e
b
2
1
,
2020
J
ab
o
n
(A
n
t
h
o
ce
p
h
al
u
s
c
ad
am
b
a
(Ro
x
b
.
)
M
i
q
)
i
s
o
n
e
t
y
p
e
o
f
fo
res
t
p
l
an
t
s
t
h
a
t
h
av
e
v
ery
rap
i
d
g
r
o
w
t
h
u
n
t
i
l
t
h
e
p
r
o
ces
s
o
f
t
h
e
h
arv
es
t
.
O
n
e
i
n
h
i
b
i
t
o
r
i
s
a
d
i
s
ea
s
e
t
h
at
a
t
t
ac
k
s
t
h
e
l
eav
e
s
i
n
t
h
e
f
o
rm
o
f
s
p
o
t
s
a
n
d
b
l
i
g
h
t
t
h
at
ca
n
cau
s
e
d
eat
h
d
u
r
i
n
g
t
h
e
g
r
o
w
t
h
p
r
o
ces
s
o
f
t
h
i
s
t
ree.
T
h
e
p
u
rp
o
s
e
o
f
t
h
i
s
p
r
o
ce
s
s
i
s
t
o
d
et
ec
t
t
h
e
o
b
j
ec
t
o
f
d
i
s
eas
e
s
t
h
a
t
at
t
ac
k
t
h
e
l
ea
v
es
o
f
j
ab
o
n
at
t
h
e
t
i
me
i
n
t
h
e
n
u
rs
er
y
.
Imag
es
o
f
affec
t
ed
j
a
b
o
n
l
eaf
d
i
s
ea
s
e
s
e
g
men
t
ed
b
y
red
u
c
i
n
g
t
h
e
RG
B
co
l
o
r
cy
l
i
n
d
e
rs
t
o
s
e
p
arat
e
t
h
e
d
i
s
ea
s
e
o
b
j
ect
f
ro
m
t
h
e
b
ack
g
ro
u
n
d
.
Red
u
ce
d
ch
a
n
n
e
l
G
-
R
p
ro
v
i
d
es
i
n
f
o
rmat
i
o
n
i
n
t
h
e
f
o
rm
o
f
d
i
s
ea
s
e
areas
co
n
t
ai
n
ed
i
n
t
h
e
i
mag
e
o
f
J
a
b
o
n
l
eaf.
Fu
rt
h
erm
o
re,
t
h
e
ch
aract
er
i
s
t
i
c
s
o
f
l
eaf
d
i
s
eas
e
can
b
e
d
et
ec
t
ed
w
el
l
u
s
i
n
g
D
W
T
i
n
t
h
e
3
-
l
ev
e
l
d
eco
mp
o
s
i
t
i
o
n
p
ro
ce
s
s
w
i
t
h
SV
M
cl
as
s
i
fi
cat
i
o
n
res
u
l
t
s
t
h
at
ca
n
s
ep
arat
e
b
o
t
h
c
l
as
s
es
o
f
s
p
o
t
s
a
n
d
b
l
i
g
h
t
b
y
8
4
.
6
7
2
%
.
K
e
y
w
o
r
d
s
:
F
e
a
tur
e
e
xtr
a
c
ti
on
J
a
bon
L
e
a
f
dis
e
a
s
e
W
a
ve
let
W
a
ve
let
e
xtr
a
c
ti
on
Th
i
s
i
s
a
n
o
p
en
a
c
ces
s
a
r
t
i
c
l
e
u
n
d
e
r
t
h
e
CC
B
Y
-
SA
l
i
ce
n
s
e
.
C
or
r
e
s
pon
din
g
A
u
th
or
:
F
e
ll
iks
F
e
it
e
r
s
T
a
mpi
nongkol
,
De
pa
r
tm
e
nt
of
C
omput
e
r
S
c
ienc
e
,
B
ogor
Agr
icultur
a
l
Unive
r
s
it
y,
J
l.
R
a
ya
Dr
a
maga
,
Dr
a
maga
C
a
mpus
,
B
ogor
1668
0.
E
mail:
f
e
ll
iks
_il
kom2016@apps
.
ipb.
a
c
.
id
1.
I
NT
RODU
C
T
I
ON
J
a
bon
(
Anthoc
e
pha
lus
c
a
da
mba
(
R
oxb.
)
M
iq)
is
one
of
the
s
pe
c
ies
that
gr
ow
in
I
ndone
s
ia's
f
or
e
s
t
a
nd
ther
e
a
r
e
two
types
of
J
a
bon
plants
na
mely
r
e
d
J
a
bon
(
Anthoc
e
pha
lus
M
a
c
r
ophyll
us
)
a
nd
white
J
a
bon
(
Anthoc
e
pha
lus
c
a
da
mba)
both
dis
ti
nguis
ha
ble
f
r
om
lea
f
bon
e
c
olor
,
a
dult
lea
f
width
(
r
e
d
J
a
bon
w
ider
than
white
J
a
bon)
a
nd
the
c
olo
r
o
f
r
e
d
J
a
bon
tr
e
e
b
r
ow
ne
r
[
1]
.
J
a
bon
is
a
s
pe
c
ies
that
tr
e
e
s
c
a
n
gr
ow
quic
kly,
thi
s
wood
c
a
n
be
us
e
d
f
or
the
manu
f
a
c
tur
e
of
f
ur
n
it
ur
e
,
plywood
a
nd
bui
ldi
ng
mate
r
ials
.
T
he
c
ult
ivati
on
a
nd
maintena
nc
e
of
thi
s
plant
a
r
e
ne
e
de
d
to
pr
oduc
e
he
a
lt
hy
s
e
e
ds
on
a
n
ongoing
ba
s
is
.
J
a
bon
s
e
e
ding
bus
ines
s
ha
s
be
e
n
done
f
r
o
m
s
mall
s
c
a
le
to
lar
ge
s
c
a
le
pe
r
mane
nt
nu
r
s
e
r
ies
.
H
owe
ve
r
,
thi
s
e
f
f
or
t
c
a
n
a
ls
o
f
ind
obs
tac
les
in
the
f
or
m
of
d
i
s
e
a
s
e
s
of
S
pot
s
a
nd
Ha
wa
r
lea
ve
s
that
c
a
n
a
tt
a
c
k
t
he
plant
J
a
bon
[
2]
.
I
n
the
li
ter
a
tur
e
,
the
ba
c
ter
ia
that
c
a
u
s
e
lea
f
dis
e
a
s
e
c
a
n
be
s
e
e
n
ba
s
e
d
on
c
ha
r
a
c
ter
is
t
ic
[
3
-
4]
.
T
he
pr
oc
e
s
s
of
making
a
r
e
d
J
a
bon
s
e
e
dli
ng
a
t
B
P
K
(
B
a
lai
P
e
ne
li
ti
a
n
Ke
hutana
n)
M
a
na
do
wa
s
c
ons
tr
a
ine
d
by
the
pr
e
s
e
nc
e
of
s
plas
h
a
nd
bli
ght
dis
e
a
s
e
in
r
e
d
J
a
bon
[
5]
a
nd
the
tec
hnique
of
dis
e
a
s
e
c
ontr
ol
wa
s
a
ls
o
pe
r
f
or
med.
T
he
p
r
oc
e
s
s
of
dis
e
a
s
e
identif
ica
ti
o
n
c
a
n
be
done
with
two
e
ve
nts
,
f
i
r
s
t
wi
th
mi
c
r
os
c
opic
inf
or
mation
that
is
to
diagnos
e
the
type
o
f
f
ungus
with
the
he
lp
o
f
a
mi
c
r
os
c
ope
a
nd
a
s
e
c
ond
mac
r
os
c
opic
inf
or
mation
that
c
a
n
e
xplain
the
s
ympt
oms
that
a
r
is
e
in
the
hos
t
plant
a
nd
c
a
n
be
obs
e
r
ve
d
di
r
e
c
tl
y
or
with
the
he
lp
of
the
c
a
mer
a
[
6]
.
T
he
textur
e
is
on
e
wa
y
that
c
a
n
identi
f
y
the
s
ympt
oms
or
o
bjec
ts
c
ontaine
d
in
a
n
im
a
ge
.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
F
e
atur
e
e
x
tr
ac
ti
on
of
j
abon
(
A
nthoce
phalus
s
p)
le
af
…
(
F
e
ll
iks
F
e
it
e
r
s
T
ampinongk
ol
)
741
T
e
xtur
e
c
a
n
de
s
c
r
ibe
the
s
moot
hne
s
s
,
r
oughne
s
s
,
a
nd
r
e
gular
it
y
of
a
n
im
a
ge
[
7]
.
One
tec
hnique
that
c
a
n
pr
ovide
inf
or
mation
ba
s
e
d
on
the
textur
e
c
ha
r
a
c
ter
is
ti
c
s
of
a
n
im
a
ge
is
the
dis
c
r
e
te
wa
ve
let
tr
a
ns
f
or
m
(
DW
T
)
.
De
tec
ti
on
o
f
medic
a
l
im
a
ge
s
o
f
br
a
in
r
e
gions
us
ing
pos
it
r
on
e
mi
s
s
ion
tom
ogr
a
p
hy
(
P
E
T
)
a
nd
magne
ti
c
r
e
s
ona
nc
e
im
a
ging
(
M
R
I
)
im
a
ge
s
us
ing
DW
T
tec
hnique
pr
ov
ides
good
a
c
c
ur
a
c
y
[
8]
.
De
tec
ti
on
of
blood
ve
nous
s
c
ler
a
in
the
human
e
ye
us
ing
40
e
ye
im
a
ge
s
a
mpl
e
s
with
DW
T
-
2D
f
e
a
tur
e
e
xtr
a
c
ti
on
indi
c
a
tes
a
s
igni
f
ica
nt
inc
r
e
a
s
e
in
a
c
c
ur
a
c
y
va
lue
[
9]
.
Automatic
tea
-
c
a
tegor
y
identi
f
ica
ti
on
by
[
10
]
u
s
ing
300
tea
im
a
ge
s
us
ing
dis
c
r
e
te
wa
ve
let
pa
c
ke
t
tr
a
ns
f
or
m
(
DW
P
T
)
a
nd
f
uz
z
y
S
VM
pr
ovides
a
n
a
c
c
ur
a
c
y
of
97.
7%
.
I
nf
or
mation
f
r
om
f
e
a
tu
r
e
e
xtr
a
c
ti
on
is
ve
r
y
im
po
r
tant
to
be
c
las
s
if
ied
ba
s
e
d
on
the
obs
e
r
ve
d
c
las
s
.
B
a
s
e
d
on
[
11
]
lea
f
dis
e
a
s
e
identif
ica
ti
on
ba
s
e
d
on
mor
phology
us
ing
S
VM
gives
a
n
a
c
c
ur
a
c
y
va
lue
o
f
87.
5
%
.
T
he
S
VM
c
las
s
if
ica
ti
on
method
is
c
a
pa
ble
of
p
r
ovidi
ng
be
tt
e
r
va
lue
than
the
mul
t
inom
ial
na
ïv
e
B
a
ye
s
c
las
s
if
ica
ti
on
method
a
nd
the
ne
ur
a
l
ne
twor
k
(
NN
)
to
e
mot
ions
in
the
text
[
12
]
.
I
n
th
is
s
tudy,
e
ne
r
gy
a
nd
e
ntr
opy
-
ba
s
e
d
wa
ve
let
method
us
e
d
f
or
r
e
c
ognize
J
a
bon
lea
f
dis
e
a
s
e
s
.
An
im
a
ge
c
a
n
be
de
c
ompos
e
d
int
o
a
s
e
t
of
s
ub
-
s
ignal
s
by
dis
c
r
e
te
wa
ve
let
tr
a
ns
f
or
m
(
DW
T
)
.
T
he
s
uppor
t
ve
c
tor
mac
hine
(
S
VM
)
models
us
e
d
to
s
e
pa
r
a
t
e
lea
f
s
pots
a
nd
lea
f
bli
ghts
dis
e
a
s
e
s
,
ba
s
e
d
on
leve
l
de
c
ompos
it
ion
dis
tr
ibut
ion
whic
h
e
ne
r
gy
a
nd
e
ntr
o
py
wa
ve
let
method
we
r
e
p
r
opos
e
d.
2.
RE
S
E
AR
CH
M
E
T
HO
D
2.
1.
Dat
a
S
e
t
T
he
da
ta
us
e
d
a
r
e
the
im
a
ge
o
f
r
e
d
J
a
bon
(
A
M
a
c
r
ophyll
us
)
a
nd
white
J
a
bon
(
A
C
a
da
mba
)
in
f
e
c
ted
with
the
dis
e
a
s
e
.
T
he
J
a
bon
lea
v
e
s
ob
s
e
r
ve
d
in
thi
s
s
tudy
a
r
e
J
a
bon
l
e
a
ve
s
that
a
r
e
s
ti
ll
in
the
gr
owth
pr
oc
e
s
s
of
3
-
6
mont
hs
of
a
ge
.
Da
ta
in
the
f
or
m
o
f
s
ympt
oms
of
s
pott
ing
a
nd
bl
ight
on
the
lea
ve
s
obtain
e
d
f
r
o
m
the
obs
e
r
va
ti
on
p
r
oc
e
s
s
a
t
f
our
nu
r
s
e
r
y
loca
ti
ons
n
ur
s
e
r
y
I
P
B
,
P
e
labuha
n
R
a
tu,
C
im
a
nggis
a
nd
B
P
K
M
a
na
do.
T
otal
s
a
mpl
e
s
obtaine
d
913
im
a
ge
s
of
r
e
d
a
nd
whit
e
J
a
bon
lea
ve
s
inf
e
c
ted
with
the
dis
e
a
s
e
F
igur
e
1
.
(
a
)
(
b
)
F
igur
e
1.
S
e
e
ds
of
r
e
d
J
a
bon
(
a
)
a
nd
S
e
e
ds
of
whit
e
J
a
bon
(
b)
2.
2.
M
e
t
h
od
ology
T
he
method
in
thi
s
r
e
s
e
a
r
c
h
c
ons
is
ts
of
s
ix
s
tage
s
including
da
ta
a
c
quis
it
ion
,
p
r
e
pr
oc
e
s
s
ing,
f
e
a
tur
e
e
xtr
a
c
ti
on,
c
las
s
if
ica
ti
on,
a
na
lys
is
,
a
nd
e
va
luatio
n.
T
he
f
low
o
f
the
r
e
s
e
a
r
c
h
p
r
oc
e
s
s
c
a
n
be
il
lus
tr
a
ted
in
F
igur
e
2.
F
igur
e
2.
F
lowc
ha
r
t
diagr
a
m
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
1693
-
6930
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
,
Vol.
18
,
No
.
2
,
Ap
r
il
2020:
7
40
-
75
1
742
2.
2.
1.
Dat
a
Ac
q
u
is
it
ion
P
hotos
of
r
e
d
a
nd
white
jabon
lea
ve
in
f
e
c
ted
by
the
dis
e
a
s
e
s
take
n
us
ing
a
digi
tal
c
a
mer
a
,
lea
f
pos
it
ion
is
in
the
mi
ddle
with
a
white
ba
c
kgr
ound
.
T
otal
913
i
mage
s
of
jabon
lea
ve
with
468
im
a
ge
s
of
white
jabon
lea
f
a
nd
445
r
e
d
jabon
lea
f
.
P
hotog
r
a
ph
of
ja
bon
lea
f
is
s
hown
in
F
igur
e
3
.
L
e
a
f
s
pots
a
r
e
us
ua
l
ly
r
a
ther
de
f
ini
te
s
pots
of
va
r
ying
s
ize
s
,
s
ha
pe
s
a
nd
c
ol
or
s
.
T
he
r
e
is
ne
a
r
ly
a
lwa
ys
a
dis
ti
nc
ti
ve
mar
gin.
[
1
3
]
.
T
his
dis
e
a
s
e
is
c
a
us
e
d
by
the
f
ungi
R
hizoc
toni
a
s
p
[
1
4
]
a
nd
is
c
a
us
e
d
a
ls
o
by
the
C
oll
e
tot
r
i
c
hum
s
p
f
ungi
[
1
5
].
L
e
a
f
bli
ghts
a
r
e
ge
ne
r
a
ll
y
lar
ge
r
dis
e
a
s
e
d
a
r
e
a
s
than
lea
f
s
pots
a
nd
mo
r
e
ir
r
e
gular
ly
s
ha
p
e
d
[
16]
.
S
ometim
e
s
the
bl
ight
ing
a
ppe
a
r
a
nc
e
of
lea
ve
s
is
the
r
e
s
ult
o
f
the
c
oa
les
c
e
nc
e
of
numer
ous
s
ma
ll
s
pots
.
T
he
c
a
us
e
of
thi
s
dis
e
a
s
e
is
by
F
us
a
r
ium
s
p
[
17]
.
(
a
)
(
b)
F
igur
e
3.
L
e
a
f
dis
e
a
s
e
on
J
a
bon
,
(
a
)
lea
f
s
pot
,
(
b
)
l
e
a
f
bli
ght
2.
2.
2.
P
r
e
p
r
oc
e
s
s
in
g
T
he
f
ir
s
t
s
tep
im
a
ge
of
jabon
lea
ve
s
labe
led
f
or
e
a
s
y
r
e
c
ognit
ion
a
nd
s
e
pa
r
a
ted
then
the
s
ize
of
the
im
a
ge
is
c
ompr
e
s
s
e
d
int
o
±
300
KB
f
or
e
a
c
h
i
mage
to
f
a
c
il
it
a
te
the
p
r
oc
e
s
s
of
c
omput
ing
.
T
he
i
mage
of
the
jabon
lea
f
that
ha
s
be
e
n
labe
led
a
nd
c
omp
r
e
s
s
e
d
is
then
s
e
gmente
d
u
s
ing
the
R
GB
c
olor
int
e
r
c
e
pt
r
e
duc
ti
on
tec
hnique
[
18
]
.
T
he
R
GB
c
olor
c
ha
n
ne
l
s
e
pa
r
a
ti
on
c
ons
is
ts
of
R
-
G,
R
-
B
,
G
-
R
,
G
-
B
,
B
-
G,
a
nd
B
-
R
[
1
9
].
T
he
s
e
gmente
d
lea
f
a
r
e
a
is
s
e
gmente
d
in
the
r
e
duc
ti
on
of
g
r
e
e
n
-
r
e
d
(
G
-
R
)
c
ha
nne
ls
a
nd
the
he
a
lt
hy
lea
f
a
r
e
a
is
s
e
gmente
d
on
the
r
e
d
-
gr
e
e
n
c
ha
nne
l
(
R
-
G)
.
C
ha
nne
l
s
e
lec
ti
on
de
pe
nds
on
the
inf
or
mation
r
e
quir
e
d.
I
n
the
G
-
R
c
ha
nne
l,
the
s
e
gmente
d
dis
e
a
s
e
s
e
gmen
t
a
s
white
a
nd
blac
k
is
a
he
a
l
thy
lea
f
a
r
e
a
F
igur
e
4
.
Or
igi
na
l
I
mage
R
e
d
J
a
bon
Ar
e
a
of
Dis
e
a
s
e
s
G
-
R
He
a
lt
hy
Ar
e
a
s
R
-
G
W
hit
e
J
a
bon
G
-
R
R
-
G
F
igur
e
4.
S
e
gmenta
ti
on
C
ha
nne
l
R
e
d,
Gr
e
e
n,
B
lue
(
R
GB
)
2.
2.
3.
F
e
a
t
u
r
e
E
xt
r
ac
t
ion
T
he
f
e
a
tur
e
e
xt
r
a
c
ti
on
tec
hnique
us
e
d
dis
c
r
e
te
wa
ve
let
tr
a
ns
f
or
m
(
DW
T
)
a
c
c
or
ding
to
[
20
-
21
]
wa
ve
let
is
a
s
mall
wa
ve
that
ha
s
the
a
bil
it
y
to
c
las
s
if
y
im
a
ge
e
ne
r
gy
a
nd
c
onc
e
ntr
a
ted
on
a
g
r
oup
of
s
mall
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
F
e
atur
e
e
x
tr
ac
ti
on
of
j
abon
(
A
nthoce
phalus
s
p)
le
af
…
(
F
e
ll
iks
F
e
it
e
r
s
T
ampinongk
ol
)
743
c
oe
f
f
icie
nts
.
W
a
ve
let
is
a
f
unc
ti
on
t
ha
t
c
a
n
s
hr
e
d
da
ta
int
o
dif
f
e
r
e
nt
s
e
ts
o
f
f
r
e
que
nc
ies
,
making
it
e
a
s
y
to
lea
r
n
[
22
]
.
T
he
pr
oc
e
s
s
of
wa
ve
let
de
c
ompos
it
ion
c
a
n
be
s
e
e
n
in
F
igu
r
e
5
.
ℎ
ℎ
a
nd
a
r
e
a
high
-
pa
s
s
f
il
ter
(
HP
F
)
a
nd
low
-
pa
s
s
f
il
ter
(
L
P
F
)
,
ℎ
ℎ
is
r
e
f
e
r
r
e
d
to
a
s
the
DW
T
c
oe
f
f
icie
nt.
ℎ
ℎ
is
the
de
tail
of
the
s
ignal
inf
or
mation
,
whe
r
e
a
s
is
a
c
r
ude
a
pp
r
oxim
a
t
ion
of
the
s
c
a
li
ng
f
unc
ti
on
[
23]
.
T
he
in
f
or
mation
f
r
om
e
a
c
h
wa
v
e
let
de
c
ompos
it
ion
s
ubgr
oup
will
be
c
a
lcula
ted
us
ing
wa
ve
let
e
ne
r
gy
a
nd
S
ha
nnon
e
ntr
opy
to
s
e
r
ve
a
s
a
f
e
a
tur
e
s
pa
c
e
f
or
pr
e
diction
[
24
-
25]
.
F
or
mul
a
f
e
a
tur
e
e
xtr
a
c
ti
on
a
s
s
hown
in
T
a
ble
1
.
F
igur
e
5.
W
a
ve
let
dis
c
r
e
te
de
c
ompos
it
ion
T
a
ble
1.
F
or
mul
a
f
e
a
tur
e
e
xt
r
a
c
ti
on
F
e
a
tu
r
e
E
xt
r
a
c
ti
on
F
or
mul
a
D
e
s
c
r
ip
ti
on
W
a
ve
le
t
E
ne
r
gy
=
∑
|
(
)
|
=
1
E
ne
r
gy f
or
e
a
c
h s
ub
-
ba
nd of
w
a
ve
le
t
de
c
ompos
it
io
n
=
∑
2
=
1
T
he
t
ot
a
l
e
ne
r
gy va
lu
e
obt
a
in
e
d f
r
om e
a
c
h s
ub
-
ba
nd
=
N
or
ma
li
z
a
ti
on of
w
a
ve
le
t
e
ne
r
gy va
lu
e
S
ha
nnon E
nt
r
opy
=
−
∑
2
(
)
=
1
P
r
ovi
de
s
i
nf
or
ma
ti
on i
n t
he
f
or
m of
a
r
a
ndom va
lu
e
of
t
he
w
a
ve
le
t
de
c
ompos
it
io
n s
p
e
c
tr
um
3.
RE
S
UL
T
S
A
ND
AN
AL
YSI
S
T
he
r
e
duc
ti
on
o
f
the
G
-
R
c
olo
r
c
ha
nne
l
is
a
ble
to
s
e
pa
r
a
te
the
objec
t
of
the
dis
e
a
s
e
with
the
ba
c
kgr
ound
s
o
that
in
the
textur
e
f
e
a
tur
e
e
xtr
a
c
ti
on
pr
oc
e
s
s
us
ing
DW
T
c
olor
c
ha
nne
l
G
-
R
is
s
e
lec
ted
f
or
the
de
c
ompos
it
ion
pr
oc
e
s
s
.
De
c
ompos
it
ion
is
done
a
s
much
a
s
3
-
leve
l
b
e
c
a
us
e
a
t
de
c
ompos
it
io
n
leve
l
-
3
nois
e
de
c
r
e
a
s
e
s
a
nd
c
ha
r
a
c
ter
is
ti
c
of
dis
e
a
s
e
be
c
omes
c
lea
r
e
r
.
T
he
c
las
s
if
ica
ti
on
pr
oc
e
s
s
us
e
s
the
f
e
a
tur
e
va
lues
obtaine
d
f
r
om
wa
ve
let
e
ne
r
gy
a
nd
e
ntr
opy
.
T
he
e
va
luation
tec
hnique
us
e
s
a
c
onf
us
ion
mat
r
ix.
3.
1.
Wave
let
d
e
c
o
m
p
os
it
ion
W
a
ve
let
de
c
ompos
it
ion
pr
oc
e
s
s
is
done
a
s
much
a
s
3
-
leve
l
de
c
ompos
it
ion
with
the
r
e
s
ult
of
G
-
R
s
e
gmenta
ti
on
whic
h
be
c
omes
the
im
a
ge
of
i
nput
in
doing
the
de
c
ompos
it
ion
pr
oc
e
s
s
.
T
h
e
higher
the
de
c
ompos
it
ion
p
r
oc
e
s
s
the
nois
e
will
de
c
r
e
a
s
e
,
s
o
the
va
lue
o
f
dis
e
a
s
e
f
e
a
tur
e
s
obtaine
d
f
r
om
the
c
a
lcula
ti
on
of
wa
ve
let
e
ne
r
gy
a
n
d
S
ha
nnon
e
ntr
opy
mo
r
e
de
tail.
W
a
ve
let
de
c
ompos
it
ion
pr
ovides
inf
or
mation
o
f
a
ppr
ox
im
a
ti
on
(
L
L
)
,
hor
izonta
l
(
H
L
)
,
(
L
H)
ve
r
ti
c
a
l,
a
nd
(
HH
)
diagona
l
.
T
he
s
ubba
nd
ha
s
dif
f
e
r
e
nt
inf
o
r
mation
on
the
L
L
s
ub
-
ba
nd
of
inf
or
mation
given
in
the
f
o
r
m
of
c
r
ude
a
ppr
oxim
a
te
va
lues
of
the
input
im
a
ge
.
S
ubba
nd
L
L
is
the
r
e
s
ult
of
DW
T
whos
e
x
-
a
xis
is
c
onve
r
ge
d
with
low
pa
s
s
f
il
ter
ing
(
L
P
F
)
a
nd
the
y
-
a
xis
is
c
onve
r
ted
by
high
pa
s
s
f
il
ter
ing
(
HPF
)
.
HL
a
xis
x
is
c
onvol
ve
d
with
HPF
a
nd
the
y
-
a
xis
is
c
onvolved
with
L
P
F
s
o
a
s
t
o
m
a
ke
a
hor
izonta
l
dir
e
c
ti
on
de
tec
ted.
T
he
L
H
s
ub
-
ba
nd
is
a
n
x
-
a
xis
DW
T
pr
oc
e
s
s
c
onvolved
wi
th
L
P
F
a
nd
the
po
r
ti
on
on
the
y
-
a
xis
is
c
onvolved
with
HP
F
s
o
that
ve
r
ti
c
a
l
li
ne
is
de
tec
ted.
HH
va
lues
a
r
e
obtaine
d
f
r
om
the
x
-
a
xis
a
nd
the
y
-
a
xis
is
c
onvolved
us
ing
HPF
s
o
that
li
ne
s
with
diagona
l
dir
e
c
ti
on
a
r
e
de
tec
ted.
T
he
pr
oc
e
s
s
of
wa
ve
let
leve
l
-
1
de
c
ompos
it
ion
c
a
n
be
s
e
e
n
in
F
igur
e
6
(
a
)
.
T
he
DW
T
de
c
ompos
it
ion
pr
oc
e
s
s
a
t
leve
l
2
pr
oduc
e
s
4
s
ub
-
ba
nds
L
L
,
L
H,
HL
,
a
nd
HH
jus
t
a
s
in
the
pr
e
vious
de
c
ompos
it
ion
pr
oc
e
s
s
.
How
e
ve
r
,
in
the
p
r
oc
e
s
s
of
de
c
ompos
it
ion
leve
l
-
2
a
nd
s
o
on
no
longer
us
e
the
im
a
ge
o
f
s
e
gmenta
ti
o
n
r
e
s
ult
s
a
s
input
but
the
va
lue
o
f
L
L
s
u
b
-
ba
nd
on
d
e
c
ompos
it
ion
leve
l
-
1
whic
h
be
c
omes
input
f
o
r
f
ur
the
r
de
c
ompos
it
ion
pr
oc
e
s
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
1693
-
6930
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
,
Vol.
18
,
No
.
2
,
Ap
r
il
2020:
7
40
-
75
1
744
T
he
y
-
a
xis
of
the
s
pe
c
tr
um
s
hows
the
f
r
e
que
nc
y
va
lue
a
nd
the
x
-
a
xis
s
hows
the
pixel
pos
it
ion
of
the
input
im
a
ge
.
T
he
r
e
f
or
e
,
by
us
ing
DW
T
the
dis
e
a
s
e
of
J
a
bon
lea
f
c
a
n
be
de
tec
ted
by
looki
ng
a
t
the
f
luctua
ti
ng
s
pe
c
tr
um
va
lue
in
e
a
c
h
de
c
ompos
it
ion
pr
oc
e
s
s
(
a
)
,
(
b
)
a
nd
(
c
)
.
As
in
F
ig
ur
e
6
(
c
)
t
he
nois
e
is
de
c
r
e
a
s
ing
s
o
that
the
pixel
pos
it
ion
of
the
dis
e
a
s
e
s
of
s
pott
ing
a
nd
bli
ght
be
c
omes
mor
e
a
ppa
r
e
nt.
LL
–
s
pe
c
t
r
um
L
H
–
s
pe
c
tr
um
HL
–
s
pe
c
tr
um
F
igur
e
6.
(
a
)
De
c
ompos
it
ion
pr
oc
e
s
s
leve
l
-
1
(
De
c
ompos
it
ion
wa
ve
let
leve
l
-
1
)
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
F
e
atur
e
e
x
tr
ac
ti
on
of
j
abon
(
A
nthoce
phalus
s
p)
le
af
…
(
F
e
ll
iks
F
e
it
e
r
s
T
ampinongk
ol
)
745
HH
-
s
pe
c
tr
um
F
igur
e
6.
(
a
)
De
c
ompos
it
ion
pr
oc
e
s
s
leve
l
-
1
De
c
ompos
it
ion
wa
ve
let
leve
l
-
1
(
c
onti
nue
)
LL
–
s
pe
c
tr
um
L
H
–
s
pe
c
tr
um
F
igur
e
6.
(
b
)
De
c
ompos
it
ion
pr
oc
e
s
s
leve
l
-
2
(
De
c
ompos
it
ion
wa
ve
let
leve
l
-
2
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
1693
-
6930
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
,
Vol.
18
,
No
.
2
,
Ap
r
il
2020:
7
40
-
75
1
746
HL
–
s
pe
c
tr
um
HH
-
s
pe
c
tr
u
m
F
igur
e
6.
(
b
)
De
c
ompos
it
ion
pr
oc
e
s
s
leve
l
-
2
(
De
c
ompos
it
ion
wa
ve
let
l
e
ve
l
-
2
)
(
c
onti
nue
)
LL
–
s
pe
c
tr
um
F
igur
e
6.
(
c
)
De
c
ompos
it
ion
pr
oc
e
s
s
leve
l
-
3
(
W
a
ve
let
leve
l
-
3
de
c
ompos
it
ion
)
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
F
e
atur
e
e
x
tr
ac
ti
on
of
j
abon
(
A
nthoce
phalus
s
p)
le
af
…
(
F
e
ll
iks
F
e
it
e
r
s
T
ampinongk
ol
)
747
L
H
–
s
pe
c
tr
um
HL
–
s
pe
c
tr
um
HH
–
s
pe
c
tr
um
F
igur
e
6.
(
c
)
De
c
ompos
it
ion
pr
oc
e
s
s
leve
l
-
3
(
W
a
ve
let
leve
l
-
3
de
c
ompos
it
ion
)
(
c
onti
nue
)
3.
2.
Wave
let
e
n
e
r
gy
an
d
e
n
t
r
op
y
W
he
n
wa
ve
lets
a
r
e
a
ppli
e
d
to
a
dis
c
r
e
te
s
ignal,
low
-
pa
s
s
a
nd
high
-
pa
s
s
f
il
ter
s
a
r
e
us
e
d,
s
pli
tt
ing
the
da
ta
int
o
a
low
f
r
e
que
nc
y
(
a
pp
r
oxim
a
ti
on)
pa
r
t
a
nd
a
high
f
r
e
que
nc
y
(
de
tail)
pa
r
t.
T
he
da
ta
dis
tr
ibut
ion
f
r
om
the
L
L
s
ub
-
ba
nd
gives
a
high
a
c
c
ur
a
c
y
va
lu
e
s
o
that
the
e
ne
r
gy
a
nd
e
ntr
opy
va
lues
of
the
L
L
will
be
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
1693
-
6930
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
,
Vol.
18
,
No
.
2
,
Ap
r
il
2020:
7
40
-
75
1
748
pr
oc
e
s
s
e
d
to
buil
d
the
S
VM
model.
T
he
e
ne
r
gy
wa
ve
let
s
hows
the
r
e
gular
it
y
va
lue
of
the
s
pe
c
t
r
um
a
nd
e
ntr
opy
mea
s
ur
e
s
a
r
a
ndom
va
lue
ba
s
e
d
on
the
oc
c
ur
r
e
nc
e
of
the
s
pe
c
tr
um.
B
oth
o
f
thes
e
va
lues
wi
ll
be
the
f
ounde
r
s
to
pe
r
f
o
r
m
the
c
las
s
if
ica
ti
on
pr
oc
e
s
s
.
T
he
a
ppr
oxim
a
ti
on
e
ne
r
gy
a
nd
e
ntr
opy
leve
l
1,
2
a
nd
3
va
lues
c
a
n
be
s
e
e
n
in
F
igur
e
7.
LL
-
1
S
pe
c
tr
um
Nilai
e
ne
r
gy:
8948533
.
49
Nilai
e
ntr
opy:
15
.
978967518
LL
-
2
S
pe
c
tr
um
Nilai
e
ne
r
gy:
4465997
.
20
Nilai
e
ntr
opy:
14
.
022177246
LL
-
3
S
pe
c
tr
um
Nilai
e
ne
r
gy:
2232297
.
42
Nilai
e
ntr
opy:
12
.
091749071
F
igur
e
7.
E
ne
r
gy
a
nd
S
ha
nnon
E
nt
r
opy
Va
lue
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
F
e
atur
e
e
x
tr
ac
ti
on
of
j
abon
(
A
nthoce
phalus
s
p)
le
af
…
(
F
e
ll
iks
F
e
it
e
r
s
T
ampinongk
ol
)
749
3.
3
.
Clas
s
if
icat
ion
T
a
ble
2
pr
ovides
the
c
ompar
is
on
of
S
VM
model
a
c
c
ur
a
c
y
leve
l
a
c
c
or
ding
to
ke
r
ne
l
R
a
dial
B
a
s
is
F
unc
ti
on
(
R
B
F
)
.
T
he
identi
f
ier
us
e
d
f
or
c
las
s
if
ica
ti
on
is
the
e
ne
r
gy
a
nd
e
ntr
opy
va
lue
of
e
a
c
h
im
a
ge
of
J
a
bon
lea
f
.
Dis
tr
ibut
ion
of
da
ta
us
e
d
in
the
c
las
s
if
ica
ti
on
is
the
dis
tr
ibut
ion
of
e
ne
r
gy
da
ta
L
L
x
a
xis
a
nd
e
nt
r
opy
L
L
a
xis
y.
T
he
s
uppor
t
ve
c
tor
model
is
obtaine
d
f
r
om
c
ompar
ing
the
10
-
f
old
c
r
os
s
va
li
da
ti
on
va
l
ue
with
the
r
e
s
ult
ing
va
lue
o
f
e
a
c
h
f
old
.
T
he
va
lue
of
the
f
old
is
c
los
e
to
the
a
ve
r
a
ge
v
a
lue
to
be
s
e
lec
ted
a
s
the
s
uppor
t
ve
c
tor
model
.
Ave
r
a
ge
leve
l
of
a
c
c
ur
a
c
y
o
f
leve
l
-
1
83.
723
%
,
l
e
ve
l
-
2
84.
270%
a
nd
leve
l
-
3
84.
416
%
.
B
a
s
e
d
on
T
a
ble
2
the
a
c
c
ur
a
c
y
va
lue
a
t
leve
l
-
1
i
s
c
los
e
to
the
a
ve
r
a
ge
va
lue
is
model
3,
leve
l
-
2
model
3
a
nd
leve
l
-
3
model
1.
T
he
highes
t
a
c
c
ur
a
c
y
is
s
hown
on
the
leve
l
-
3
a
c
c
ur
a
c
y
va
lue
84.
672%
model
-
1.
Da
ta
dis
tr
ibut
ion
a
nd
S
VM
model
f
or
e
a
c
h
leve
l
c
a
n
be
s
e
e
n
in
F
igur
e
8.
T
a
ble
2.
S
uppor
t
ve
c
tor
10
-
f
old
c
r
os
s
va
li
da
ti
on
m
ode
ls
K
-
fo
ld
A
c
c
ur
a
c
y V
a
lu
e
L
e
ve
l
-
1
(%)
A
c
c
ur
a
c
y V
a
lu
e
L
e
ve
l
-
2
(%)
A
c
c
ur
a
c
y V
a
lu
e
L
e
ve
l
-
3
(%)
1
84.672%
84.672%
84.672%
2
85.766%
86.496%
85.036%
3
83.212%
84.307%
84.672%
4
82.482%
82.117%
83.212%
5
85.401%
86.861%
85.401%
6
84.307%
85.401%
85.401%
7
81.387%
81.752%
82.117%
8
83.212%
83.942%
85.036%
9
84.307%
84.307%
85.401%
10
82.482%
82.847%
83.212%
A
ve
r
age
83.723%
84.270%
84.416%
M
ode
l
3
L
L
leve
l
-
1
M
ode
l
3
L
L
leve
l
-
2
F
igur
e
8.
S
uppor
t
ve
c
tor
mac
hine
models
Evaluation Warning : The document was created with Spire.PDF for Python.