Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
8
,
No.
6
,
D
ece
m
ber
201
8
, pp.
4343
~
43
51
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v8
i
6
.
pp
4343
-
43
51
4343
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
Accele
rating Co
mp
re
ssi
on
Time
of the
St
an
dard
JPEG b
y
Emp
loyin
g t
he
Qu
antize
d
YCbCr
Color Sp
ace Alg
orithm
Tr
ini
Sa
p
ta
ri
ani
, Sarifu
din
Madend
a, Er
na
s
tu
ti
,
Wid
ya Silfi
an
ti
Depa
rtment
o
f
C
om
pute
r
Scie
n
ce
,
Gunad
arma
Un
ive
rsit
y
,
Indon
esia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Sep
25
, 201
7
Re
vised
Feb
2
0
, 2
01
8
Accepte
d
Ma
r
6
, 2
01
8
In
thi
s
p
ape
r
,
we
propose
a
quant
i
ze
d
YCb
Cr
col
or
spac
e
(QY
CbCr
)
te
chn
ique
which
is
employ
ed
in
standa
rd
JP
EG.
The
objecti
v
e
of
thi
s
work
is
to
acce
l
erate
co
m
puta
ti
onal
ti
m
e
of
the
st
anda
r
d
JP
EG
image
compress
ion
al
gorit
hm
.
Thi
s
is
a
d
evelopm
ent
of
the
stan
dar
d
JP
EG
which
is
n
amed
QY
CBCr
al
gorit
hm
.
It
m
erg
es
two
proc
e
ss
es
i.
e
.
,
YCbCr
c
olor
spac
e
conve
rsion
and
Q
quantiz
at
ion
in
whi
ch
in
th
e
stand
ar
JP
EG
they
w
er
e
per
form
ed
sepa
r
at
e
l
y
.
The
m
erg
er
form
s
a
n
ew
single
in
te
gr
at
ed
proc
ess
o
f
col
or
conv
ersion
which
is
employ
ed
prior
to
DC
T
proc
ess
b
y
su
bseque
nt
l
y
el
iminating
th
e
q
uant
i
za
t
ion
proc
ess.
The
equ
atio
n
form
ula
of
QY
CbCr
col
or
cove
rsion
is
built
base
d
on
the
c
hrom
ina
nce
and
luminanc
e
prop
e
rti
es
of
th
e
hum
an
visual
s
ystem
which
de
ri
ved
from
qua
ti
z
at
ion
m
at
rices.
Expe
rimen
t
result
s
per
form
e
d
on
images
of
diffe
ren
t
siz
e
s
show
tha
t
the
co
m
puta
ti
ona
l
running
ti
m
e
of
QY
CbCr
al
gorithm
give
s
4
up
to
8
ti
m
es
faste
r
tha
n
JP
EG
standa
rd,
and
a
lso
provide
s
highe
r
compress
ion
rat
io
and
b
et
t
er
imag
e
qual
ity
.
Ke
yw
or
d:
Chrom
inace
Com
pr
ession
DCT
JPEG
Lum
inance
Qu
a
ntiza
ti
on
m
at
rix
QY
C
bCr
Copyright
©
201
8
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Trini Sa
ptaria
ni
,
Dep
a
rtm
ent Com
pu
te
r
Scie
nc
e,
Guna
dar
m
a U
ni
ver
sit
y,
Jakar
ta
,
In
done
sia
.
Em
a
il
: t
rini@staff.guna
dar
m
a.ac.id
1.
INTROD
U
CTION
The
de
velo
pme
nt
of
im
age
acqu
isi
ti
on
te
ch
no
l
og
y
t
od
ay
c
on
ti
nues
to
gro
w
ver
y
rap
i
dly
res
ulti
ng
i
n
excell
ent
im
age
qual
it
y.
This
giv
e
s
a
posit
ive
im
pact
on
us
ers
w
ho
nee
d
a
good
im
age
qual
it
y,
as
w
el
l
as
a
po
sit
ive
im
pact
on
the
de
velo
pm
ent
of
inf
or
m
at
ion
te
chnol
og
y
a
nd
ty
pe
s
of
te
c
hnologie
s
su
c
h
as
bi
ome
dical
,
ast
rono
m
ic
al
,
rem
ote
sensing
,
an
d
arc
hae
ologica
l
fiel
ds
that
m
os
tl
y
us
e
the
i
m
age
as
data
to
be
stud
ie
d.
Im
age
qu
al
it
y
can
be
im
pr
ov
ed
al
so
fo
ll
ow
ed
by
the
la
rge
vo
l
um
e
of
da
ta
gen
e
rated
so
that
im
pactin
g
t
he
need f
or increa
sed
m
e
m
or
y ca
pacit
y. It also
has
a
n
im
pact o
n t
he
tra
ns
m
is
sion
sp
ee
d
at
t
he
ti
m
e o
f
i
m
a
ge data
exch
a
nge
t
hro
ugh
a
c
omm
u
nicat
ion
net
work.
T
hese
im
pacts
can
be
an
ti
ci
pated
thr
ou
gh
the
ap
plica
ti
on
of
i
m
age
com
pr
ession
al
go
rith
m
s
in
i
m
age
a
cqu
isi
ti
on
te
ch
no
l
og
y
dev
ic
e
s
su
c
h
as
cam
eras
an
d
vid
e
os.
The
i
m
age
com
pr
es
sion
al
gorithm
com
m
on
ly
use
d
in
this
te
c
hnol
og
y
f
or
im
a
ge
file
stora
ge
is
JPEG.
J
PEG
is
an
al
gorithm
ty
pe
of
Lo
ssy
com
pr
essio
n
(c
om
pr
essio
n
accom
pan
ie
d
by
the
chang
e
in
data)
,
w
hich
is
a
com
pr
essio
n
t
echn
i
qu
e
by
m
ini
m
iz
ing
th
e
num
ber
of
bits
(
reducin
g
value
)
t
hroug
h
qu
a
ntiza
ti
on
proces
s
accom
pan
ie
d
by
rou
nd
i
ng
the
res
ult
val
ue
[1]
.
Q
ualit
y
of
c
om
pr
essio
n
ca
n
be
m
ai
ntain
with
t
he
qu
a
ntiza
ti
on
process
is
ca
rri
ed
out
in
a
f
re
qu
e
ncy
dom
ain
suc
h
as
DCT
(D
isc
reet
Cos
ine
Tra
ns
f
orm
)
wh
ic
h
is
ap
plied
i
n
JPEG al
gorith
m
[
2]
-
[4
]
.
The
tw
o
m
ai
n
processes
i
n
JP
EG
im
age
com
pr
essi
on
are
t
he
DCT
a
nd
qua
ntiza
ti
on
p
r
oce
sses, w
he
re
these
tw
o
proc
esses
are
perform
ed
separ
at
el
y
and
gr
eat
ly
de
te
rm
ine
the
m
agn
it
ude
of
rat
io
an
d
qual
it
y
of
t
he
com
pr
essio
n
im
age.
Fig
ur
e
1
sho
ws
the
ge
ner
al
sc
hem
e
of
a
sta
nda
rd
JPEG
im
age
com
pr
essio
n
m
od
el
consi
sti
ng
of t
wo p
a
rts: i
m
ag
e com
pr
essio
n i
n
the t
op secti
on and
dec
om
pr
essio
n
i
n
the
bott
om
secti
on
[1].
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
20
88
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4343
-
4351
4344
Figure
1. Stan
da
rd JPE
G
im
age co
m
pr
essio
n and dec
om
pr
es
sion schem
e
2.
DC
T
A
ND Q
UANTIZ
ATION
PR
OCES
SES O
N
ST
A
NDA
RD J
PE
G
This
sect
ion
de
al
s
sp
eci
fical
ly
with
the
DCT
and
qua
ntiza
ti
on
processe
s
that
are
directl
y
relat
ed
to
the
resea
rc
h
pro
posal
s
i
n
this
pa
per.
E
quat
ion
1
de
note
s
t
he
DCT
proce
ss,
wh
ic
h
c
onve
rts
im
age
data
from
the
sp
at
ia
l
dom
ai
n
to
the
fr
e
qu
e
ncy
do
m
ain
,
w
her
e
[f
]
is
the
in
pu
t
of
an
i
m
age
blo
c
k
of
8x
8
pix
el
s,
[
DCT]
i
s
the D
CT
m
at
ri
x,
[D
C
T]
T
is t
he
tran
s
pose m
at
rix of
[
DCT
],
and [
F] i
s the
r
esult o
f
the
D
C
T pro
ce
ss
[5
]
.
[
]
=
(
[
]
.
[
]
)
.
[
]
(1)
with:
Fu
rt
her
m
or
e, t
he values i
n
th
e m
a
trix [
F] ar
e quantiz
ed a
s
sh
ow
n by eq
ua
ti
on
(2
).
Qu
a
nt
iz
at
ion
is the
process
of
divi
din
g
each
el
em
ent
of
t
he
m
at
rix
F(n,m
)
again
st
the
e
lem
ent
of
t
he
quantiz
at
ion
m
at
rix
Q(n,
m
)
[6
]
[7
]
.
Fig
ur
e
2
sho
w
s
an
e
xam
ple
of
DCT
pr
ocess
res
ult
an
d
f
oll
ow
e
d
by
qu
a
nt
iz
at
ion
pr
ocess
.
W
e
can
see
the
re
s
ult
of
the
quan
ti
zat
ion
proces
s,
the
val
ues
in
this
m
at
rix
beco
m
e
s
m
al
le
r
a
nd
a
lot
of
re
dunda
nt
values
,
especia
ll
y
the
value
of
0.
T
his
al
lows
the
m
at
rix
values
to
be
e
nc
od
e
d
wit
h
a
sm
al
le
r
nu
m
ber
of
bits
with the
h
el
p o
f
zi
g
-
zag
codi
ng a
nd RLE
-
H
uffm
an
co
ding.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
Accel
eratin
g
C
ompressi
on Ti
me
of the Sta
ndard
JPE
G b
y
Em
plo
yi
ng …
(
Trini
S
.
)
4345
(
,
)
=
(
(
,
)
(
,
)
)
(2)
Figure
2. Exa
m
ple o
f
DCT
a
nd qua
ntiza
ti
on
process
In
the
dec
om
pr
ession
pr
ocess,
to
return
t
he
m
at
rix
F
Q
(
n,m
)
to
it
s
init
ia
l
da
ta
f
(
n,m
)
,
t
hen
the
in
ve
rs
e
process
m
us
t
be
done
.
This
process
is
carried
ou
t
by
equ
at
io
ns
(
3)
and
(4),
w
hic
h
is
pr
ecede
d
by
the
dequa
ntiza
ti
on
process a
nd
fo
l
lowe
d by the i
nv
e
rse DCT
process
.
He
re
[iDCT]
=
[D
CT]
T
and [iDCT]
T
=
[
DCT]
.
Exam
ple inv
e
r
s pro
ce
ss: i
DC
T and
d
e
qu
a
ntiza
ti
on
as
sho
w
n
in
Fig
ure
3.
(
,
)
=
(
,
)
.
(
,
)
(3)
[
]
=
(
[
]
.
[
]
)
.
[
]
(4)
1
8
0
6
.7
-
7
.1
0
0
8
.
1
5
6
4
.
2
3
8
-
1
.7
5
0
-
3
.5
9
7
-
1
.2
1
3
2.5
0
9
-
1
.5
0
9
0.4
2
4
0
.
0
0
4
1
.
8
9
6
0
.
7
8
4
1
.2
4
7
-
3
.3
5
8
-
0
.8
6
7
-
0
.7
0
9
-
0
.7
9
4
-
0
.
7
3
7
-
0
.9
2
5
-
0
.
1
3
5
-
0
.2
0
6
0.9
2
6
1
.
0
1
9
2
.6
8
2
1
.035
1
.
3
6
7
0
.2
0
7
-
0
.4
8
4
-
0
.4
6
3
0.9
5
4
-
1
.2
7
5
-
3
.2
5
0
-
0.
9
2
5
-
1
.
4
9
7
-
0
.2
4
8
0
.7
5
0.1
8
6
-
1
.0
0
3
1.5
5
6
2
.3
9
3
0.5
2
5
1.
1
2
4
0
.846
-
0
.
6
2
4
0.
7
4
9
-
0
.
4
2
5
-
1
.
6
8
9
-
1
.0
5
9
-
0
.1
1
7
-
0
.
5
7
3
-
1
.2
8
4
0
.3
2
6
-
1
.4
3
3
1.7
3
7
1.
4
8
7
0
.2
0
9
-
0
.0
5
6
0
.1
7
9
0
.9
7
7
-
0
.0
9
8
1.1
4
6
-
1
.5
6
9
-
0
.8
8
1
F
(n
,m
)
=
(
[
]
⬚
.
[
]
)
⬚
.
[
]
⬚
(
,
)
(
,
)
2
2
6
2
2
6
2
2
3
2
2
3
2
2
6
2
2
6
2
2
8
2
2
7
2
2
6
2
2
6
2
2
3
2
2
3
2
2
6
2
2
6
2
2
8
2
2
7
2
2
6
2
2
6
2
2
3
2
2
3
2
2
6
2
2
6
2
2
8
2
2
7
2
2
6
2
2
6
2
2
3
2
2
3
2
2
6
2
2
6
2
2
8
2
2
7
2
2
6
2
2
6
2
2
3
2
2
3
2
2
6
2
2
6
2
2
8
2
2
7
2
2
7
2
2
7
2
2
7
2
2
2
2
2
6
2
2
8
2
2
6
2
3
0
2
2
8
2
2
8
2
2
5
2
2
4
2
2
5
2
2
9
2
2
9
2
2
9
2
2
3
2
2
3
2
2
6
2
2
1
2
2
7
2
2
5
2
2
6
2
2
8
f
(x,y)
=
9
0
3
-
4
3
1
0
-
1
0
0
-
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
-
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
F
Q
(n
,m
)
=
2
2
3
4
5
6
8
1
1
2
2
2
4
5
7
9
1
1
3
2
3
5
7
9
1
1
1
2
4
4
5
7
9
1
1
12
1
2
5
5
7
9
1
1
1
2
12
1
2
6
7
9
1
1
1
2
1
2
12
1
2
8
9
1
1
1
2
1
2
1
2
12
1
2
1
1
1
1
1
2
1
2
1
2
1
2
12
1
2
Q
(n
,m
)
=
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
20
88
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4343
-
4351
4346
F
igur
e
3. Exa
m
ple I
nv
er
s P
r
ocess
: i
DCT
&
D
e
qu
a
ntiza
ti
on
3.
PROP
OSE
D MET
HO
D
In
E
qu
at
io
n
2
,
the
qu
a
ntiza
ti
on
process
of
a
m
at
rix
hav
in
g
8×
8
el
e
m
ents
requires
64
di
vision
s
a
nd
the
de
quantiz
a
ti
on
pr
ocess
in
Eq
uation
3
ne
eds
64
m
ulti
plica
ti
on
s.
T
his
m
eans
for
the
i
m
a
ge
siz
e
of
Nx
M
pix
el
s,
Nx
M
di
vision
a
nd
NxM
m
ulti
plica
t
i
on
a
re
re
qu
ir
e
d
[
6].
This
pr
oc
ess
is
qu
it
e
ti
m
e
con
su
m
ing,
so
we
need
to
fin
d
s
ol
ution
s
on
how
to
acce
le
rate
t
he
c
om
pr
essio
n
a
nd
dec
om
pr
ession
proce
ss
es.
T
he
al
gorit
hm
of
com
bin
ing
DC
T
an
d
qua
ntiza
ti
o
n
processe
s
to
ac
cel
erate
the
c
om
pr
essi
on
a
nd
dec
ompressi
on
proce
sses
of
JPEG
im
age
ha
s
bee
n
stu
die
d
pre
viously
in
[
8]
-
[
11]
.
I
n
this
sect
ion,
w
e
pro
po
se
so
l
utions
by
us
in
g
the
YCb
Cr
-
qu
a
ntize
d
c
olor
s
pac
e
co
nv
e
rsion
m
od
el
.
The
ba
sic
idea
of
t
his
m
e
tho
d
is
to
us
e
Q
m
at
rix
that
ha
s
a
ho
m
og
e
ny
val
ue,
s
o
that
one
const
ant
value
can
ex
pr
e
ss
Q.
Th
ere
f
or
e,
t
he
qu
a
ntiza
ti
on
p
r
ocess
can b
e
c
arr
ie
d
ou
t at
t
he be
gin
ni
ng alo
ng w
i
th the YC
bCr
c
olo
r
c
onversi
on
process
.
[
12]
-
[
14
]
In
sta
ndar
d
JP
EG,
the
RGB
to
YCb
C
r
colo
r
conver
si
on
is
giv
e
n
by
Eq
ua
ti
on
5
an
d
fro
m
YCb
Cr
to
RGB
gi
ven
by
Eq
uatio
n
6
.
O
ur
pro
posed
m
et
hod
is
to
int
egr
at
e
t
he
Q
quantiz
at
io
n
pr
oc
ess
into
the
YCb
Cr
equ
at
io
n,
s
o
it
is
called
YC
bCr
-
qua
ntize
d
or
Q
YCb
Cr
.
Equ
at
io
ns
7
and
8
show
t
he
RGB
to
Q
YCb
Cr
conve
rsion
pr
oc
ess
that
we
propose
.
The
ne
xt
quest
io
n
is
wh
at
a
re
the
quantiz
at
io
n
val
ues
f
or
the
lu
m
inance
com
po
ne
nt Y (
Q
Y
)
a
nd fo
r
the
chrom
inance c
om
po
ne
nts Cb
and Cr
(
Q
C
)?
[
]
=
[
0
.
299
0
.
587
0
.
114
0
.
500
−
0
.
419
0
.
081
−
1
.
169
−
0
.
331
0
.
500
]
[
]
(5)
[
]
=
[
1
0
.
0
1
.
402
1
1
.
772
0
.
0
1
−
0
.
344
−
0
.
741
]
[
]
(6)
[
]
=
[
(
0
.
299
0
.
587
0
.
114
)
/
(
0
.
500
−
0
.
419
0
.
081
)
/
(
−
1
.
169
−
0
.
331
0
.
500
)
/
]
[
]
(7)
(
[
]
⬚
.
[
]
)
⬚
.
[
]
⬚
[
(
,
)
]
∗
[
(
,
)
]
2
2
3
4
5
6
8
1
1
2
2
2
4
5
7
9
1
1
3
2
3
5
7
9
1
1
1
2
4
4
5
7
9
1
1
12
1
2
5
5
7
9
1
1
1
2
12
1
2
6
7
9
1
1
1
2
1
2
12
1
2
8
9
1
1
1
2
1
2
1
2
12
1
2
1
1
1
1
1
2
1
2
1
2
1
2
12
1
2
Q
(n
,m
)
=
9
0
3
-
4
3
1
0
-
1
0
0
-
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
-
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
F
Q
(n
,m
)
=
1
8
0
6
-
8
9
4
0
-
6
0
0
-
2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
4
0
0
0
0
0
0
0
-
5
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
F
(n
,m
)
=
2
2
5
2
2
6
2
2
3
2
2
2
2
2
5
2
2
6
2
2
6
2
2
8
2
2
6
2
2
6
2
2
4
2
2
3
2
2
6
2
2
7
2
2
7
2
2
9
2
2
6
2
2
6
2
2
3
2
2
2
2
2
6
2
2
7
2
2
6
2
2
8
2
2
5
2
2
5
2
2
2
2
2
2
2
2
5
2
2
6
2
2
6
2
2
8
2
2
6
2
2
6
2
2
3
2
2
3
2
2
6
2
2
7
2
2
6
2
2
8
2
2
7
2
2
8
2
2
5
2
2
4
2
2
7
2
2
8
2
2
8
2
3
0
2
2
7
2
2
7
2
2
5
2
2
4
2
2
7
2
2
8
2
2
8
2
3
0
2
2
5
2
2
5
2
2
3
2
2
2
2
2
5
2
2
6
2
2
6
2
2
8
f
(x,y)
=
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
Accel
eratin
g
C
ompressi
on Ti
me
of the Sta
ndard
JPE
G b
y
Em
plo
yi
ng …
(
Trini
S
.
)
4347
[
]
=
[
(
0
.
299
0
.
587
0
.
114
)
/
(
0
.
500
−
0
.
419
0
.
081
)
/
(
−
1
.
169
−
0
.
331
0
.
500
)
/
]
−
1
[
]
(8)
The
J
PE
G
sta
nd
a
r
d
c
om
pr
ession
al
ways
us
es
a
ch
r
om
i
nan
ce
qua
ntiza
ti
on
value
gr
eat
er
tha
n
a
lum
inance
qua
ntiza
ti
on
val
ue
.
This
co
rr
es
ponds
to
the
hu
m
an
visu
al
syst
e
m
that
is
m
or
e
sensiti
ve
to
lum
inance
c
ha
ng
e
tha
n
sm
al
l
cha
ng
e
of
c
olo
r
or
c
hrom
ina
nce.
Exam
ple,
Photos
hop
software
us
es
a
dif
fer
e
nt
qu
a
ntiza
ti
on
m
at
rix
Q
f
or
phot
ography
an
d
f
or
we
b
ap
plica
ti
on
s.
For
phot
ogra
ph
y
a
pp
li
c
at
ion
,
P
hotos
hop
C
S
us
es
12
qua
ntiza
ti
on
m
at
rice
s
of
Q
12
,
Q
11
,
...
,
Q
1
resu
lt
in
g
di
ff
e
ren
t
c
om
pr
e
ssion
rati
o
an
d
im
age
qual
it
y.
Be
low
are
tw
o
exam
ples
of
quantiz
at
io
n
m
a
tric
es
Q
12
and
Q
9
,
w
her
e
Q
iL
and
Q
iC
(
i
=
1,
2,
.
.
.,
12)
are
use
d
to
qu
a
ntize
the
lu
m
inance a
nd c
hrom
inance com
po
nen
ts,
r
es
pe
ct
ively
.
Our
ap
proac
h,
in
ob
ta
ini
ng
a
value
of
scal
ar q
ua
ntiza
ti
on,
is
to
cal
culat
e
the
aver
a
ge
valu
e
of
m
at
rix
Q
i
by
us
i
ng
Eq
uations
9
a
nd
10
.
E
xam
ple
for
m
at
rix
Q
12
,
we
ob
ta
i
ne
tw
o
val
ues
of
Q
L
=
1.7
031
a
nd
Q
C
=
2.6
719,
w
hile
fo
r
m
at
rix
Q
9
we
get
Q
L
=
10.43
75
an
d
Q
C
=
12.96
88.
Fu
rt
her
m
or
e
f
or
eac
h
Q
i
use
d
in
Photos
hop
is
c
al
culat
ed
the
va
lues
of
Q
L
an
d
Q
C
,
a
nd
t
hen
em
bed
ed
t
he
m
to
the
Eq
uat
ion
s
7
an
d
8
to
obta
in
the f
i
nal form
ula o
f
the
Q
YCbC
r
co
l
or co
nve
rsion a
nd it
s in
ver
se
.
=
1
2
∑
∑
(
,
)
=
1
=
1
(9)
=
1
2
∑
∑
(
,
)
=
1
=
1
(10)
Our
pro
posed
m
et
ho
d
can
re
place
sta
nda
rd
J
PEG
com
pr
es
sion
schem
e
(in
the
F
i
gure
1)
with
a
ne
w
on
e
a
s
sho
wn
in
the
Fi
gure
4
.
T
his
sc
hem
e
is
si
m
pler
an
d
faste
r
beca
use
it
has
integ
r
at
ed
the
qu
a
ntiza
ti
on
process
i
nto
t
he
RGB
to Q
Y
CbCr
c
olor
sp
a
ce
co
nv
e
rsion p
r
ocess
a
nd
al
s
o
the d
e
qu
a
ntiza
ti
on
p
r
ocess
i
nto
t
he
QY
C
bCr
t
o
R
GB c
olo
r
s
pac
e co
nv
e
rsion
proces
s.
1
1
1
2
3
3
3
3
1
1
1
2
3
3
3
3
1
1
2
3
3
Q
12C
=
1
1
1
1
1
1
1
2
1
1
1
1
1
1
1
2
1
1
1
1
1
Q
12L
=
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
20
88
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4343
-
4351
4348
Figure
4. O
ur
pro
po
se
d
m
et
ho
d f
or n
e
w JPE
G
c
om
pr
essio
n an
d decom
pr
e
ssion sch
em
e
4.
E
X
PERI
MEN
T
RES
ULTS
Our
im
age
co
m
pr
ession
a
nd
dec
om
pr
essio
n
al
go
rithm
s
hav
e
bee
n
im
pl
e
m
ented
a
nd
te
ste
d
in
th
e
Ma
tl
ab
pr
og
ra
m
m
ing
la
ng
ua
ge.
The
c
om
pu
te
r
sp
eci
ficat
io
ns
us
e
d
duri
ng
the
exp
erim
ent
has
process
or
Du
al
Core
i5,
1.8
G
Hz,
4
GB
D
D
R3L
RAM
m
e
m
or
y
and
128
GB
SS
D
.
The
perform
ance
of
the
de
velo
pe
d
m
et
ho
d
is
evaluated
by
m
easur
ing
it
s
com
pu
ta
ti
on
al
tim
e,
co
m
pr
ession
rati
o
a
nd
i
m
age
qu
al
it
y
and
the
n
we
c
om
par
e
them
with
tho
s
e
ob
ta
ine
d
by
s
ta
nd
a
rd
JP
EG
.
Table
1
sho
ws
the
perform
ance
of
sta
ndar
d
JPEG
al
go
rith
m
and
our
al
gorithm
i
n
te
rm
o
f
c
om
pu
ta
ti
on
al
ti
m
e.
The
first
col
um
n
sh
ow
s
10
t
est
i
m
ages
with
di
ff
e
ren
t
siz
e
an
d
co
ntaini
ng
di
ff
e
ren
t
c
ha
racteri
sti
cs
of
colo
r,
s
hap
e
a
nd
te
xture.
T
he
seco
nd
a
nd
thir
d
col
um
ns
sh
ow
the
c
om
pu
ta
ti
on
al
ti
m
e
of
t
he
sta
ndar
d
JPEG
com
pr
ess
io
n
al
gorithm
and
t
he
pro
posed
co
m
pr
ession
al
gorithm
,
resp
ect
ively
.
F
ro
m
this
ta
ble,
it
ca
n
be
see
n
that
com
pu
ta
ti
on
al
ti
m
e
us
ing
QY
C
bCr
rea
ches
4
to
8
ti
m
es
faster
tha
n
sta
nda
rd
JPE
G.
Her
e
,
we
c
an
no
te
that
the
com
pu
ta
ti
on
al
tim
e
of
i
m
age
com
pr
ession
dep
e
nds
no
t
only
on
it
s
siz
e
bu
t
al
so
on
the
div
e
rsity
of
it
s
colo
r,
sh
a
pe
an
d
te
xtu
re
.
The
m
or
e
var
ie
d
co
lors,
s
hap
e
s
an
d
/
or
te
xtu
re
s
con
ta
ine
d
in
th
e
i
m
age,
the
lo
ng
e
r
of the c
om
pu
ta
ti
on
ti
m
e is require
d.
Table
1.
C
om
pu
ta
ti
on
al
Tim
e
of JPE
G
Sta
nd
ard
an
d
J
PE
G usin
g QY
C
bCr
I
m
ag
e si
ze
(M
eg
a Pixel
)
JPEG stan
d
ard
(secon
d
)
JPEG us
in
g
QYCb
Cr
(secon
d
)
5
1
2
x
5
1
2
=
0
.2
4
2
4
.41
1
3
6
7
1
.08
0
9
7
2
1
0
0
0
x 4
8
3
=
0
.4
8
3
4
.80
3
1
6
6
1
.32
4
3
2
3
7
0
0
x 8
0
0
=
0
.560
6
.73
5
5
5
5
1
.33
2
6
7
5
1
6
3
0
x 1
4
8
0
=
2
.412
8
.68
4
1
2
7
1
.34
7
5
6
1
2
1
0
2
x 1
5
8
7
=
3
.336
7
.72
0
8
2
5
1
.53
0
8
8
7
2
1
9
2
x 2
0
2
0
=
4
.428
8
.83
0
2
3
1
1
.06
1
6
6
7
4
4
8
2
x 1
2
1
0
=
5
.423
7
.46
8
4
0
0
1
.71
1
7
6
3
3
7
5
8
x 1
9
0
7
=
7
.167
6
.29
8
6
2
8
1
.92
7
5
9
1
6
5
0
0
x 3
6
3
7
=
2
3
.6
4
1
7
.60
4
4
6
8
1
.11
0
3
6
1
6
4
4
6
x 4
0
9
6
=
2
6
.4
0
3
1
4
.43
9
8
5
3
1
.61
1
0
7
4
Be
sides
of
c
om
pu
ta
ti
on
al
ti
m
e,
we
al
s
o
m
easur
e
the
c
om
pr
ession
rati
o
a
nd
im
age
qu
al
it
y
resu
lt
in
g
by
the
sta
nd
ar
d
JPEG
a
nd
our
m
et
ho
d.
F
or
t
his
m
easur
em
e
nt,
seve
n
of
quantiz
at
ion
m
a
tric
es
fr
om
Ph
oto
s
ho
p
CS:
Q
12, Q1
1, Q10,
Q9, Q8,
Q7, and
Q
6
ar
e
u
sed
, and ju
st
two of
ten test
i
m
ages ar
e p
re
sented
. Th
e
fir
st on
e
is
Lena
im
age,
has
m
any
are
as
with
hom
og
eneous
col
or
s
,
a
li
tt
le
te
xtu
re
,
but
it
has
m
a
ny
sh
a
pe
va
riat
ion
s
.
The
se
co
nd
one
is
Ba
bo
on
im
age
has
a
he
te
rogen
e
ous
c
olor,
s
ha
pe
a
nd
te
xt
ur
e
.
T
he
Com
pr
essio
n
r
at
io
i
s
cal
culat
ed
us
i
ng
eq
uatio
n
(
11)
and
the
im
age
qu
al
it
y
is
per
f
or
m
ed
by
m
eas
ur
i
ng
the
pea
k
sign
al
to
no
ise
rati
o
(P
S
NR)
as
sho
wn
by
equ
at
io
n
(
12).
f
is
un
c
om
pr
essed
im
a
ge,
f
i
is
com
pr
essed
im
age,
ij
f
and
ij
i
f
are
pix
e
l
value o
f un
c
om
pr
essed
a
nd
com
pr
essed
im
age at
posit
ion
(
i,j
).
=
(
)
(
)
(11)
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
Accel
eratin
g
C
ompressi
on Ti
me
of the Sta
ndard
JPE
G b
y
Em
plo
yi
ng …
(
Trini
S
.
)
43
49
M
S
E
L
o
g
P
S
N
R
2
5
5
.
20
with
N
i
M
j
ij
i
ij
f
f
M
xN
M
S
E
1
1
2
)
(
1
(12)
Tables
2
a
nd
3
show
the
cal
c
ulati
on
res
ults
of
c
om
pr
essio
n
rati
o
an
d
im
age
qu
al
it
y
(PSNR)
f
or
t
w
o
exam
ples
Len
a
an
d
Ba
bo
on
i
m
ages
res
pe
ct
ively
.
Figure
5
s
hows
t
he
gr
a
ph
of
PS
NR
(
dB)
a
gai
ns
t
the
com
pr
essio
n
r
at
io
of
Ta
ble
2,
w
hile
Fig
ur
e
6
is
the
grap
h
for
Ta
ble
3.
T
he
c
urve
in
bl
ue
li
ne
represe
nts
th
e
perform
ance
of
JPE
G
wit
h
QY
C
bCr
al
gor
it
h
m
and
the
r
ed
li
ne
cu
r
ve
corres
ponds
to
the
pe
rfor
m
ance
of
sta
nd
a
rd
J
PEG.
On
t
hese
tw
o
gr
a
ph
s
,
it
appears
that
both
the
rati
o
an
d
qual
it
y
of
JPEG
com
pr
ession
us
i
ng
QY
C
bCr
are
al
so
bette
r
th
an
sta
nd
a
r
d
J
PEG
.
Im
ages
with
heter
og
e
neous
co
ntent
ha
ve
c
ha
racteri
sti
cs
increase
d
in
ra
ti
o
and
qu
al
it
y
of
com
pr
essi
on.
T
hat’s
m
at
t
er
are
not
to
o
sign
ific
a
nt,
but
an
increase
i
n
rati
o
and quali
ty
o
f
com
pr
e
ssio
n b
ecom
e
m
or
e sign
i
ficant f
or a
m
or
e h
om
og
e
ne
ous im
age.
Table
2.
T
he
Com
pr
essio
n
Ra
ti
o
an
d
Im
age Q
ualit
y
of Le
na
Ph
o
to
shop
CS Q
-
Matr
ix
JEPG stan
d
ard
JEPG with p
rop
o
s
ed
QYCb
Cr
PSNR
(dB
)
Co
m
p
r
ess
io
n
Ratio
PSNR
(dB
)
Co
m
p
r
ess
io
n
Ratio
12
4
6
.14
2
5
2
.40
1
4
6
.49
4
4
2
.53
9
8
11
3
9
.11
0
8
4
.51
1
7
3
8
.97
5
.07
5
10
3
6
.93
4
1
7
.45
6
6
3
6
.37
8
1
9
.00
2
2
9
3
6
.11
7
2
9
.48
9
5
3
5
.72
9
1
1
1
.01
6
5
8
3
5
.39
5
6
1
1
.16
9
6
3
5
.22
8
2
1
2
.98
5
7
3
5
.36
5
3
1
1
.50
3
8
3
5
.14
6
1
3
.10
4
7
6
3
4
.54
4
5
1
3
.02
7
3
4
.70
5
2
1
5
.28
1
5
Figure
5. Cu
rves of P
SN
R
vs.
Com
pr
essio
n R
at
io o
f
Le
na’s Im
age
Table
3.
T
he
Com
pr
essio
n
Ra
ti
o
an
d
Im
age Q
ualit
y
of
Ba
boon
’s
Im
age
Ph
o
to
shop
CS
Q
-
Mat
rix
JEPG stan
d
ard
JEPG us
in
g
QYCb
Cr
PSNR
(dB
)
Co
m
p
r
ess
io
n
Ratio
PSNR
(dB
)
Co
m
p
r
ess
io
n
Ratio
12
4
6
.10
2
1
1
.53
4
4
4
6
.48
1
2
1
.60
6
6
11
3
7
.69
5
5
2
.32
8
2
3
8
.02
1
1
2
.53
5
7
10
3
3
.81
4
4
3
.14
3
7
3
4
.01
6
6
3
.41
1
1
9
3
2
.80
4
9
3
.55
4
3
3
2
.99
7
5
3
.75
4
4
8
3
2
.05
0
1
3
.85
7
0
3
2
.26
1
7
4
.01
4
7
7
3
1
.91
2
2
3
.87
2
0
3
2
.22
1
5
4
.05
9
6
3
0
.91
4
2
4
.15
2
9
3
1
.40
0
6
4
.41
1
4
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
20
88
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4343
-
4351
4350
Figure
6. Cu
rves of P
SN
R
vs.
Com
pr
essio
n R
at
io o
f
Ba
bo
on’s
I
m
age
5.
CONCL
US
I
O
N
On
e
of
the
pro
blem
of
the
JPEG
al
gorithm
l
ie
s
in
the
schem
e
(ar
chite
c
ture
)
de
sign
of
it
s
dev
el
op
m
ent
in
order
to
get
op
ti
m
u
m
of
c
om
pu
ta
ti
on
tim
e,
rati
o
an
d
qu
al
it
y
of
the
co
m
pr
ession
im
a
ges.
I
n
this
our
stu
dy,
the
QY
CbC
r
al
gorithm
is
pr
op
os
ed
as
on
e
of
de
velo
pme
nt
of
Sta
nd
a
r
d
JPE
G
in
w
hi
ch
it
m
erg
es two
processes i.e.
, YCbCr and qua
nt
iz
at
ion
in
JPE
G
Stan
dard form
ing
o
ne
inte
grat
ed
proce
ss of co
lo
r
conve
rsion
deri
ved
from
the
chrom
inance
a
nd
l
um
inance
qu
a
ntiza
ti
on
m
at
rices
.
E
xp
e
rim
ental
resu
lt
s
that
perform
ed
on
t
he
pr
ocess
or
D
ual
Core
i
5,
1.8
G
Hz,
4
GB
DD
R
3L
RAM
m
e
m
or
y
and
128
GB
SS
D
usi
ng
10
te
st
i
m
ages
wit
h
dif
fere
nt
siz
e
wh
ic
h
are
res
pecti
vely
con
ta
ining
dif
fer
e
nt
char
act
e
risti
cs
of
c
olor,
s
hap
e
and
te
xtu
re
,
show
that
t
he
QYbCr
al
go
rithm
has
com
pu
ta
ti
on
al
tim
e
4
up
to
8
tim
es
faster
than
sta
ndar
d
JPE
G
al
gorithm
.
In
add
it
io
n,
this
m
et
ho
d
al
so
pro
vid
es
bette
r
com
pr
essio
n
r
at
io
and
im
ag
e
qu
al
it
y.
The
fu
tu
re
work,
t
o
ob
ta
i
n
m
or
e
e
ff
ic
ie
nt
r
unni
ng
ti
m
e,
the
st
ud
y
can
be
de
vel
op
by
in
vestigat
ing
the
lo
w
le
ve
l
arch
it
ect
ure
of
how Q
YCb
Cr
al
gorithm
s ar
e com
pu
te
d
t
hro
ugh ha
rdwar
e
desig
n.
REFERE
NCE
S
[1]
Gregor
y
K.
W
allace,
“
The
JP
EG
stil
l
pi
ct
ur
e
co
m
pre
ss
ion
standa
rd”
,
IEEE
Tr
an
sac
ti
ons
on
Con
sum
er
El
ec
tron
i
cs,
Vol.
38
,
No.
1,
FEBRUA
RY 1992.
[2]
Ahm
ed,
N., Nat
a
raj
am T
.
,
Rao
K,
“
Discre
te Cosin
e
Tr
ansform
“
,
IE
EE
,
1974
.
[3]
S
y
ed
Al
i
Kha
y
a
m
,
“
The
Discre
t
e
Cosine
Tra
nsf
orm
(DCT):
Theor
y
and
Appli
cation”,
Mich
iga
n
Stat
e
Univ
ersi
t
y
,
Marc
h
10th
200
3.
[4]
Sari,
W
el
l
ia
Shi
nta
,
R
ac
hm
awa
nto,
Eko
Har
i,
Seti
adi
,
De
Ros
al
Igna
ti
us
Mos
es,
Sari
,
Chr
ist
y
Atika
,
“
A
Good
Perform
anc
e
OTP
Enc
r
y
pti
on
Im
age
base
d
o
n
DCT
-
DW
T
Stega
nogra
ph
y
”
,
TE
LKOM
NIK
A
,
Vol.
15,
No.
4
,
Dec
ember
2017
,
pp.
1987~1995
ISS
N:
1693
-
6930,
a
cc
r
edi
t
ed
A
b
y
DIK
TI
,
Dec
r
ee
No:
58
/DIKT
I/Ke
p/2013
DOI
:
10.
12928/TELK
OM
NIK
A.v15i
4.
5883
[5]
Made
nda
Sar
ifu
ddin,
“
Pengol
ah
an
Ci
tra da
n
Vid
eo
Digi
ta
l
”, Boo
k,
2016
.
[6]
Set
y
ani
ngsih
,
E
m
y
,
Harj
oko,
A
gus,
“
Survey
of
H
y
brid
Im
age
C
om
pre
ss
ion
Te
chni
ques
”
,
Int
ern
at
ion
al
Journal
o
f
El
e
ct
ri
ca
l
and
C
om
pute
r
Engi
ne
eri
ng
(IJECE)
Vol.
7,
No.
4,
Augus
t
2017,
pp.
2
206~2214
ISS
N:
2088
-
8708,
DO
I
:
10.
11591/ijece.
v
7i4.
pp2206
-
221
4
[7]
El
awa
d
y
,
Im
an,
La
khdar
,
Abdelmounaim
Moula
y
,
Mus
ta
pha
,
Kheli
fi,
“
The
Noise
Reduc
ti
o
n
over
W
ire
le
s
s
Channe
l
Us
ing
Vec
tor
Quantiza
ti
on
Com
pre
ss
ion
and
Filt
eri
ng
”
,
Inte
rna
t
iona
l
Jo
urna
l
of
El
e
ct
r
icala
nd
Com
pute
r
Engi
ne
eri
ng
(
IJE
CE)
,
Vol
.
6
,
No
.
1,
Februa
r
y
201
6,
pp
.
130~138 I
SS
N:
2088
-
8708,
DO
I:
10
.
11591
/i
jece
.
v6i1
.
8483
[8]
Alen
Doce
f
,
Fao
uzi
Kos
senti
ni,
Khanh
Nguuy
en
-
Phiand
Ism
ae
il
Raga
b
,
“
The
Quantize
d
DCT
an
d
Its
A
ppli
ca
ti
o
n
to
DCT
-
Based
V
ide
o
Cod
ing
”,
IE
EE
Tra
nsa
ct
ion
on
Im
age
Proc
essing,
Vol.
11,
N
o.
3
,
Mar
ch, 200
2.
[9]
Sukirm
an
Edi
,
“
Peningka
t
an
Kiner
ja
Algor
it
m
a
Kom
pre
si
dan
Dekom
pre
si
JPE
G
Mela
lui
Pen
ggabunga
n
pros
es
DCT
dan
Ku
ant
i
sasi”,
Dise
rta
si
,
Octobe
r
2009
.
[10]
Su
kirman
Edi
,
E
rna
stuti
,
Made
n
da,
Sarifuddi
n
,
“
Model
DCT
-
te
rk
uant
isasi
dan
IDCT
-
te
rku
ant
isasi
untuk
kom
pre
si
ci
tr
a”
,
Proce
edi
n
gs Sem
ina
r
Ilmi
ah
Nasiona
l
KO
MM
IT,
2010.
[11]
Li
nli
n
Deng
and
Ping
Fu,
Aijun
Sang
and
Sili
n
Mang
,
“
Quanti
z
at
ion
Algor
it
hm
for
Color
Im
age
Coding
Based
o
n
Multi
-
dimension
al
Ve
ct
or
Matri
x
DCT
Orthogon
a
l
Tr
ansform
at
ion
”,
I
EEE,
2009
[12]
M
ahe
ndra
M.Di
xit
,
Pri
y
a
ta
m
kum
ar,
“
Com
par
at
ive
Anal
y
sis
of
Vari
abl
e
Quan
t
iz
a
ti
on
DCT
an
d
Vari
abl
e
R
an
k
Matri
x
SV
D Alg
orit
hm
s for
Im
ag
e
Com
pre
ss
ion A
ppli
cation”
,
I
E
EE
,
2010
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
Accel
eratin
g
C
ompressi
on Ti
me
of the Sta
ndard
JPE
G b
y
Em
plo
yi
ng …
(
Trini
S
.
)
4351
[13]
Ramin
Sam
ada
ni,
“
Chara
cteri
z
ing
and
Esti
m
ating
Bloc
k
DCT
Im
age
Com
pre
ss
ion
Quanti
za
tion
Para
m
et
ers”
,
As
il
om
ar
Confer
enc
e
on
Signa
l
s,
S
y
stems
,
and
Com
pute
rs,
30
Octobe
r
–
2
Novem
ber
2005,
Paci
f
ic
Grove
,
CA,
USA.
[14]
Zi
m
bic
o,
A
.
Sch
nei
der
,
F.,
Maia,
J,
“
Com
par
at
ive
stud
y
of
the
p
er
form
anc
e
of
th
e
JP
EG
al
gorit
hm
using
opti
m
iz
e
d
quant
i
za
t
ion
m
atrice
s for
ul
tra
sou
nd
image
compress
ion”
,
I
EEE,
2
014
.
BIOGR
AP
H
I
ES
OF
A
UTH
ORS
Tri
ni
Sapt
ariani
,
SK
o
m
,
MM
SI
-
has
rec
ei
v
ed
B.
S.
degr
ee
in
Com
pute
r
Scie
nc
e
of
Gunada
rm
a
Univer
sit
y
in
19
88
and
the
Mag
iste
r
in
Magist
er
Mana
gement
fr
om
Gunada
rm
a
Univer
sit
y
in
1994.
She
is
cu
rre
ntly
an
senio
r
lectur
er
in
th
e
fac
u
lty
of
com
pute
r
sc
ie
n
ce
an
d
Inform
at
ion
Engi
ne
eri
ng,
Gunada
rm
a
Univer
sit
y
.
Her
cur
ren
t
rese
ar
ch
in
te
r
e
sts
inc
lud
e
infor
m
at
ion
s
y
st
em,
image
proc
essin
g.
Emai
l: tr
in
i@
staff
.
gunad
arma.ac
.
id
Prof.
Sarifuddi
n
Made
nda
-
Curr
ent
l
y
Dir
ec
tor
of
Doctor
al
Progra
m
of
Inform
at
ion
Te
chno
lo
g
y
and
Com
pute
r
S
ci
en
ce,
Gunada
r
m
a
Univer
sit
y
–
Indone
sia.
He
rec
e
ive
d
the
B.
S
.
degr
ee
from
Univer
sit
y
of
I
ndonesia
,
in
19
89,
the
M.S.
degr
ee
from
Instit
ut
e
Nati
ona
l
des
Scie
nce
s
Appliqué
es
d
e
L
y
on
(INS
A
de
L
y
on)
–
Frenc
h
,
i
n
1992,
and
th
e
Ph.D.
degr
ee
fro
m
Univer
sit
y
of
Burgund
y
-
Frenc
h,
in
1995.
From
1995
to
19
96,
he
was
a
Resea
rch
As
sociate
at
LI
ESIB
La
bora
tor
y
-
Univer
sit
y
of
Bur
gund
y
.
In
1997,
he
becam
e
a
Le
c
ture
r
in
Co
m
pute
r
Scie
nc
e
Depa
rtment,
Gu
nada
rm
a
Univer
sit
y
–
Indone
si
a
.
From
2002
to
2007,
he
was
a
rese
ar
che
r
a
t
Aca
demic
Rese
a
rch
Consortium
on
digi
ta
l
imagi
ng,
vide
o,
aud
io
and
m
ult
imedia
(CoRIMedia
),
Cana
da
.
His
res
ea
rch
intere
sts
a
re
image
pro
ce
s
sing:
image
compress
ion,
co
lor
i
m
agi
ng,
image
dat
ab
ase
and
se
arc
hing
,
m
edica
l
image
ana
l
y
sis,
implementi
ng
o
f
image
pro
ce
ss
i
ng
al
gori
thms
on
FP
GA
for
real
ti
m
e
image
an
a
l
y
sis.
Emai
l: sari
f@staff.
gunad
ar
m
a.
ac.i
d
Dr.
Ern
astut
i
-
has
re
ceive
d
B.
S.
degr
ee
in
Mathe
m
at
i
cs
fr
om
Univer
sit
y
of
Indone
sia
in
Dec
ember
1985
,
and
the
M.S
.
in
Com
pute
r
Sci
e
nce
from
Univ
er
sit
y
of
Indon
esia,
in
Jul
y
1994,
and
PhD
degr
ee
s
in
Com
pute
r
Scie
nc
e
from
G
unada
rm
a
Unive
rsit
y
,
Indone
sia
,
in
April
2008.
She
is
cur
ren
tly
an
associa
t
e
profe
ss
or
in
the
fac
ul
t
y
of
computer
scie
n
ce
an
d
Inform
at
ion
Engi
ne
eri
ng,
Gunada
rm
a
Univer
sit
y
.
Her
cur
ren
t
rese
ar
ch
in
te
r
e
sts
in
cl
ude
gr
ap
h
the
or
y
and
combinat
ori
al
o
pti
m
iz
ation,
gra
ph
-
the
ore
ti
c
interc
onne
ct
ion
ne
tworks,
par
a
llel
and
distr
ibute
computing,
and design
and
an
alys
is of
al
gor
it
hm
s.
Emai
l: e
rn
as_tut
i@
y
ahoo.co.id
Dr.
W
id
y
a
Si
lfi
a
nti
–
now
is
a
se
nior
l
ec
tur
er
in
t
he
fa
cul
t
y
of
co
m
pute
r
scie
n
ce
a
nd
informati
on
engi
ne
eri
ng,
Gu
nada
rm
a
Unvier
sit
y
.
She
has
rec
e
ive
d
Com
pute
r
Degre
e
fro
m
Gunada
rma
Univer
sit
y
in
A
pril
1995
,
and
t
he
Magist
er
fro
m
Gunada
rm
a
Univer
sit
y
in
1
998,
and
PhD
degr
ee
s
in
Com
pute
r
Sci
enc
e
fr
om
Gunada
rm
a
Univer
sit
y
,
Ind
onesia
,
in
2010.
Her
cur
ren
t
rese
arc
h
in
te
rest
s inc
lud
e
in
form
at
ion
s
y
st
em,
da
t
aba
se.
Emai
l: ws
il
fi@sta
ff
.
gunad
arma.
a
c.
id
.
Evaluation Warning : The document was created with Spire.PDF for Python.