Internati
o
nal
Journal of Ele
c
trical
and Computer
Engineering
(IJE
CE)
V
o
l.
5, N
o
. 1
,
Febr
u
a
r
y
201
5,
pp
. 71
~77
I
S
SN
: 208
8-8
7
0
8
71
Jo
urn
a
l
h
o
me
pa
ge
: h
ttp
://iaesjo
u
r
na
l.com/
o
n
lin
e/ind
e
x.ph
p
/
IJECE
Low bit Rate Video Qu
ality
Anal
ysis Usin
g NRDPF-VQA
Algorith
m
*
Subrahm
a
n
y
am Ch,
**
D
Venkat
a
Rao
,
**
*
N
Usha Ra
ni
*
Res
earch
S
c
hol
ar,
**
Prin
cip
a
l,
*
**
P
r
ofes
s
o
r & Head
*, ***
School of
Electronics, Vign
an’s Foundation
for Sc
ien
c
e,
Technolog
y
and
Res
earch
University
Vadlamudi, Guntur Dist, India
**
Narasar
a
opet
a
Institut
e
of
Eng
i
neeing
&
Te
chn
o
log
y
, Nar
a
sara
opeta
, Guntur
Di
st, Indi
a
Article Info
A
B
STRAC
T
Article histo
r
y:
Received Oct 1, 2014
Rev
i
sed
No
v
31
, 20
14
Accepted Dec 20, 2014
In this work, we propose NRDPF-V
QA (No
Referen
ce Distortion Patch
F
eatures
Video
Qualit
y As
s
e
s
s
m
ent) m
odel a
i
m
s
to us
e to m
eas
ur
e the v
i
deo
quality
assessment for H.264/AVC (Adva
nced Video Coding). Th
e proposed
me
thod ta
ke
s adva
nta
g
e
of the c
ontra
st c
h
a
n
ge
s in
the
vide
o qua
lity
by
luminance ch
an
ges. Th
e propos
ed quality
m
e
tric was tested
b
y
using LIVE
video datab
a
se. The
experim
e
ntal
results s
how that th
e
new index
perform
ance
co
m
p
ared with the
other NR-VQA m
odels
that req
u
ire tr
ainin
g
on LIVE video datab
a
ses, CSIQ vi
deo databas
e
, and VQEG
HDTV video
datab
a
s
e
.
The
v
a
lues
are
com
p
ared wi
th hum
a
n
s
c
ore
index
anal
ys
is
of
DMOS.
Keyword:
H.
26
4/
A
V
C
No-Refe
r
ence
NRDP
F-
V
Q
A
N
R
-V
QA
VQE
G H
D
T
V
Copyright ©
201
5 Institut
e
o
f
Ad
vanced
Engin
eer
ing and S
c
i
e
nce.
All rights re
se
rve
d
.
Co
rresp
ond
i
ng
Autho
r
:
Su
bra
h
m
a
ny
am
.C
h,
Research Sc
holar,
Scho
o
l
of
Electr
on
ics,
V
i
gn
an
’
s
Fo
und
atio
n fo
r
Scien
ce, Tech
no
log
y
an
d
Resear
ch Un
iver
sity
Vadl
am
udi
, G
unt
ur
Di
st
, I
ndi
a.
E-m
a
il
:
sub
r
a
h
m
a
ny
am
ch20
02
@
g
m
a
il
.com
1.
INTRODUCTION
We kno
w t
h
e
u
n
d
e
rlyin
g
`quality aware' n
a
t
u
ral
v
i
d
e
o
statistics
m
o
d
e
l in
th
e sp
ace-tim
e
d
o
m
ain
an
d
descri
be pe
rce
p
t
u
al
l
y
rel
e
va
n
t
t
e
m
poral
feat
ures t
h
at
are u
s
ed t
o
m
odel
i
n
t
e
r s
u
b
b
a
nd c
o
r
r
el
at
i
ons
o
v
e
r
b
o
t
h
l
o
cal
an
d gl
o
b
a
l
t
i
m
e
spans [
2
, 4,
6, 1
4
, 1
9
,
22]
.
Th
e ove
ra
l
l
m
odel
i
s
t
h
e
basi
s on
a
n
al
g
o
ri
t
h
m
fo
r pre
d
i
c
t
i
n
g
v
i
d
e
o
q
u
a
lity th
at is sh
own
t
o
co
rrelate wel
l
with
h
u
m
an
ju
dg
m
e
n
t
s o
f
visu
al qu
ality. We also
co
m
p
are it's
perform
a
nce to state-of-t
he-a
rt FR and NR
VQA approac
h
e
s
[
18]
.
B
e
f
o
re
we
desc
ri
be t
h
e m
odel
i
n
det
a
i
l
,
we
revi
e
w
rel
e
va
nt
pri
o
r
w
o
r
k
i
n
t
h
e a
r
ea
of
V
Q
A
.
We
creat
e a
`zer
o
sh
ot
'
NR
V
Q
A
m
odel
by
m
a
ki
ng
m
easurem
ent
s
o
f
perce
p
t
u
a
l
l
y
rel
e
vant
t
e
m
poral
vi
de
o
st
at
i
s
t
i
c
s [1
4
]
. Nat
u
ral
vi
d
e
os c
o
nt
ai
n
r
e
gul
a
r
structures a
nd ha
ve
general
l
y piece-wise
sm
ooth lum
i
nances i
n
s
p
ac
etim
e separated
by s
p
arse
s
p
atio-
te
m
p
o
r
al edg
e
d
i
scon
tinu
ities [16
]
. Th
is st
ro
ng
p
r
o
p
erty of n
a
t
u
ral v
i
d
e
os h
a
s
b
een
explo
ited
in
a v
a
ri
ety o
f
applications.
It induces self sim
ilarity over s
p
ace a
nd t
i
m
e
which, for exam
ple, ha
s bee
n
e
xpl
oited for
reso
l
u
tio
n enhace
m
e
n
t
, action
recogn
itio
n
,
RR an
d NR
VQA [1
2,
1
9
,
21
]. Self sim
i
larity statist
i
cs co
m
p
u
t
ed
usi
n
g di
f
f
ere
n
ces bet
w
ee
n cons
ecut
i
v
e f
r
a
m
es have bee
n
use
d
t
o
cap
t
u
re di
st
o
r
t
i
o
n
-
i
n
duce
d
ana
m
ol
ous
b
e
h
a
v
i
or an
d
to
con
d
u
c
t v
i
su
al qu
ality in
feren
c
e. Deriv
i
n
g
i
n
sp
iratio
n fro
m
th
ese ex
am
p
l
es, we
m
o
d
e
l
te
m
p
o
r
al self similarities
u
s
ing
fram
e
d
i
fferen
ces
b
e
tween
con
s
ecu
tiv
e
fram
e
s. On
ce t
h
e
no
rm
alized
coefficients for each subba
nd
are co
m
puted,
each
coefficient m
a
p is
par
titi
one
d into P X
P patches [14, 19].
Features
are e
x
tracted
using t
h
e c
o
effi
cients
of each patc
h. The c
o
e
fficien
ts of each sub-band a
r
e m
ode
led as
obey
i
n
g
a
ge
n
e
ral
i
zed
Gaus
s
i
an di
st
ri
but
i
o
n,
w
h
i
c
h e
ffec
t
i
v
el
y
capt
u
res
t
h
e
beha
vi
o
r
of t
h
e c
o
ef
fi
ci
ent
s
o
f
nat
u
ral
an
d
di
s
t
ort
e
d
o
f
vi
de
o
s
[
17]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E V
o
l
.
5, No
. 1, Feb
r
uar
y
20
1
5
:
7
1
– 77
72
We
have
p
r
o
p
o
se
d t
h
e
use
o
f
exem
pl
ar nat
u
ral
pi
ct
ure c
o
nt
ent
as
gr
o
u
n
d
t
r
ut
h
rel
a
t
i
v
e t
o
w
h
i
c
h
statistical
reg
u
larity
may b
e
d
e
term
in
ed
. Such
a m
o
d
e
l, h
o
wev
e
r, m
a
y b
e
li
mited
in
th
at
it can
o
n
l
y cap
ture
comm
on basel
i
ne cha
r
acteris
tics of a
speci
fic collection
of
non
-d
istorted con
t
en
t, and
i
s
th
ereb
y
n
o
t
ab
le to
uni
versal
l
y
re
prese
n
t
vi
de
o
speci
fi
c i
n
t
r
i
n
si
c charact
eri
s
t
i
c
s [1, 9, 1
2
]
.
Al
so, t
h
e c
o
nst
r
uct
i
on
of
suc
h
a
d
a
tab
a
se
requ
i
r
es th
e unb
iased
selectio
n
an
d
m
a
in
tain
ence o
f
h
und
reds o
f
n
a
tural und
isto
rted
v
i
d
e
os. Th
is
also
raises
the que
stion of how
m
a
ny
exemplar
vide
os a
r
e
nee
d
ed to
desi
gn an accurate
natural vi
deo
m
odel,
and
how
distinctive these nee
d
to
be relative
to each
othe
r
and t
o
the world of
vide
os
[4, 9,
12, 17]. Fi
nally,
gi
ve
n t
h
e l
i
m
it
at
i
ons o
f
i
m
age/
vi
de
o cam
era capt
u
re,
d
i
sto
r
tion
s
are inev
itab
l
y in
trodu
ced
i
n
th
e cap
t
ure
pr
ocess
an
d
he
nce t
h
e
p
r
ocu
r
em
ent
of
pe
rfe
ct
l
y
nat
u
ra
l
`p
r
i
st
i
n
e'
vi
deos i
s
p
r
act
i
cal
l
y
impos
si
bl
e
[1
6]
.
2.
R
E
SEARC
H M
ETHOD
Th
e ap
pro
ach for th
e NR
DPF-
VQA
(No
Refere
n
c
e
Distortio
n
Patch
Feature Vid
e
o
Qu
ality
Anal
y
s
i
s
) t
h
at
we ha
ve de
vel
ope
d can
be su
m
m
a
ri
zed as f
o
l
l
o
w
s
. Gi
ve
n
a (po
ssi
bl
y
di
st
ort
e
d
)
vi
de
o h
a
vi
n
g
l
o
w
bi
t
rat
e
, fi
r
s
t
com
put
e e
n
c
odi
ng
a
n
d
dec
o
di
n
g
of
t
h
e
f
r
a
m
es sel
ect
ed b
y
t
h
e
vi
de
o
du
r
a
t
i
on.
The
f
o
l
l
o
wi
ng
are t
h
e
eq
uat
i
o
ns t
o
t
o
ap
pl
i
e
d
t
o
a
gi
ve
n
di
st
ort
i
o
n
vi
de
o
[1
5]
.
The
e
q
u
a
t
i
ons
rep
r
ese
n
t
t
h
e f
eat
ures of
t
h
e di
st
ort
i
o
n pa
tch
e
s
o
f
v
i
deo. It also
ob
ser
v
ed
th
at t
h
e normalized
l
u
m
i
nance
val
u
es st
r
o
ngl
y
t
e
nd t
o
wa
r
d
s a
uni
t
no
rm
al
Gaus
si
an c
h
ar
act
eri
s
t
i
c
for
vi
de
o. C
o
m
put
e t
h
e
M
A
TLAB
p
r
o
g
ram
fo
r the
eq
uation
s
.
.
.
Whe
r
e
∑
,
1
,
,
,
and
,
are
h
o
ri
z
ont
al
a
n
d
ve
rt
i
cal
m
o
t
i
on
vec
t
ors at
pi
xel
(i
,
j
) re
spect
i
v
el
y
,
w i
s
t
h
e
wi
n
d
o
w
of
di
m
e
nsi
on m
X m
ove
r
whi
c
h t
h
e
l
o
cal
i
zed
com
put
at
i
on
of
t
h
e t
e
n
s
o
r
i
s
p
e
rf
orm
e
d.
|
,
,
|
|
2Γ
1/
/
/
Γ
z>
0
The
NR
D
PF-
V
Q
A
al
g
o
ri
t
h
m
i
s
desi
gne
d
f
o
r t
h
i
s
vi
de
o
qual
i
t
y
assessm
ent
pu
r
pose
.
T
h
e
M
A
TLAB
co
d
e
i
s
devel
ope
d f
o
r ent
i
r
e
e
quat
i
o
n
s
.
Fi
gu
re
1.
Ex
pe
ri
m
e
nt
al
set
up
usi
n
g L
I
VE
da
t
a
base
L
I
VE
VI
D
E
O
DATAB
ASE R
E
L
E
ASE
NRDPF-V
QA
ALGOR
ITH
M
DIST
ORTI
ON VI
DEOS
QUAL
ITY
METR
IC
Pear
so
n and
Spea
rm
an correlation c
o
efficient
Out
put im
ages
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Lo
w b
it
Ra
te Vid
e
o
Qu
a
lity
An
a
l
ysis Usi
n
g NRDPF-VQA Alg
o
r
ithm
(Subra
h
ma
n
y
am
.C
h
)
73
Figu
re
2.
NR
D
PF-
VQ
A m
ode
l fram
e
wo
rk
3.
R
E
SU
LTS AN
D ANA
LY
SIS
Wh
ile
we are u
s
ing
a pro
b
ab
ilistic fram
e
work fo
r d
i
st
o
r
tion
classifi
catio
n
wh
ere
we
u
s
e the
p
r
ob
ab
ility o
f
v
i
d
e
o
b
e
ing
d
i
sto
r
ted
with
a p
a
rticu
l
ar d
i
stortio
n, bu
t j
u
st
as a p
r
oof o
f
ho
w
g
ood
th
e featu
e
s
use
d
i
n
t
h
e
f
r
a
m
ewor
k act
as
di
st
o
r
tion
id
en
tifiers and
also
wh
ich
d
i
stortio
n
s
are m
i
scl
a
ssified
with
wh
ich
one
s,
we are
re
porting the c
onfusi
o
n m
a
trix for
first st
a
g
e cl
assification.
We woul
d
like to point out that
each
ent
r
y
i
n
t
h
e c
o
n
f
usi
o
n m
a
t
r
ix i
s
t
h
e
m
ean of
co
nf
usi
ons
acro
ss L
I
VE
vi
de
o
dat
a
base
.
W
e
can
see
fr
om
dat
a
base t
h
at
H.
26
4/
A
V
C
f
o
rm
at i
s
conf
us
ed wi
t
h
ot
her
fo
rm
at
s. Al
so, M
P
EG-
2
an
d
IP are al
so co
nf
use
d
som
e
t
i
m
e
s. Whi
t
e
noi
se a
nd
B
l
ur are c
o
m
p
arat
i
v
el
y
m
o
re ro
bust
i
n
det
ect
i
on an
d n
o
t
c
o
n
f
used
us
ual
l
y
wi
t
h
ot
he
r di
st
ort
i
o
n
s
.
(a)
(b
)
(c)
NRDPF-
VQA
MOD
E
L
TEST VIDE
O
TRAI
NE
D
M
ODEL
SPAT
I
AL AN
D
TEM
P
O
R
A
L
ACTI
VI
TY
FE
ATUR
ES
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E V
o
l
.
5, No
. 1, Feb
r
uar
y
20
1
5
:
7
1
– 77
74
(d
)
(e)
(
f)
(g
)
Fi
gu
re
3.
Im
ages
fr
om
(a) t
o
(
g
) a
r
e c
o
nsi
d
er f
o
r
H.
26
4
V
i
deo
wi
t
h
l
o
w
bi
t
rat
e
As we com
put
ed ha
ve correl
ations for each algorithm
ove
r train test trials, we find m
e
a
n
SROCC value and
the standa
rd error associate
d
with
these correlation val
u
es
.
W
e
pl
ot th
e sam
e
across the dataset along with
err
o
r
ba
rs
o
n
e
st
anda
rd
de
vi
at
i
on
wi
de
f
o
r
ea
ch
of
t
h
e e
v
al
u
a
t
e
d al
g
o
ri
t
h
m
s
.
Tabl
e
I
.
Gr
o
u
n
d
t
r
ut
h
an
d NR
DPF
-
V
Q
A
fo
r vi
de
os (a
)
,
(c
),
(d
),
(f
) a
n
d (
g
)
fr
om
live VQ
A
databa
se
Video a
c
d
f
g
Actual
0.
9500
1.
4600
1.
1800
1.
1800
0.
5000
VL
bit-r
a
te
1.
0923
1.
2651
1.
1981
1.
0189
0.
9365
NRDPF-
VQA
(
P
roposed)
1.
0021
1.
3642
1.
1842
1.
1792
0.
6245
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Lo
w b
it
Ra
te Vid
e
o
Qu
a
lity
An
a
l
ysis Usi
n
g NRDPF-VQA Alg
o
r
ithm
(Subra
h
ma
n
y
am
.C
h
)
75
Tab
l
e II. Med
i
an
SR
OCC co
rrelatio
n
s
on
every po
ssi
b
l
e com
b
in
atio
n
of train
/test set sp
lits (subj
ectiv
e
DM
OS
V
s
NRDPF
-
V
Q
A
D
M
OS)
.
80%
o
f
co
ntent
use
d
f
o
r
trainin
g
Distortion PSNR
SSIM
VQM
STMA
D
MOV
I
E
RRED
VIDE
O-
BLIIN
DS
NRDPF-
VQA
M
P
E
G
-
2
0.
667
0.
786
0.
828
0.
9484
0.
9286
0.
809
0.
882
0.
9514
H.
264
0.
714
0.
762
0.
828
0.
9286
0.
9048
0.
885
0.
851
0.
9312
W
i
r
e
less
0.
680
0.
714
0.
714
0.
7976
0.
800
0.
771
0.
802
0.
8124
I
P
0.
660
0.
600
0.
770
0.
7143
0.
788
0.
771
0.
826
0.
8436
AL
L 0.
671
0.
650
0.
7451
0.
8250
0.
807
0.
826
0.
821
0.
8514
Fig
u
re
4
.
Probab
ility Plo
t
for
DMOS ind
e
x fo
r H.26
4 format
Figure 5.
Plot for
Fram
e
features for H.264
4.
CO
NCL
USI
O
N
We propo
sed
a No
referen
ce
Vid
e
o
b
a
sed
qu
ality
assess
men
t
m
o
d
e
l NRDPF-VQA wh
i
c
h
p
e
rfo
r
m
s
q
u
a
lity assessmen
t
o
f
Vi
d
e
o with
ou
t an
y in
fo
rm
atio
n
from
d
i
sto
r
tio
n
i
m
ag
e. No
d
i
st
o
r
tion
sp
ecific
featu
r
es
suc
h
as n
o
i
s
e;
bl
u
r
has been
m
odel
e
d
i
n
t
h
e
al
go
ri
t
h
m
in specific. T
h
e
algorithm
onl
y
q
u
ant
i
f
i
e
s t
h
e
bl
i
nd i
n
t
h
e vi
de
o d
u
e t
o
p
r
esen
ce of
d
i
st
ort
i
o
n
s
. T
h
e desi
g
n
e
d
fram
e
wo
r
k
i
s
spat
i
a
l
dom
ai
n, hum
an pe
rcept
i
o
n
base
d,
sim
p
l
e
r and
fa
st
er w
h
i
c
h m
a
kes i
t
supe
ri
o
r
t
o
ot
he
r n
o
r
e
fere
nce al
g
o
ri
t
h
m
s
. The i
n
d
e
x i
s
bee
n
sh
o
w
n t
o
per
f
o
r
m
wel
l
across
di
f
f
ere
n
t
di
st
ort
i
o
ns
ve
ri
fy
i
ng i
t
s
di
st
ort
i
o
n a
g
n
o
st
i
c
nat
u
re.
An e
x
hau
s
t
i
v
e anal
y
s
i
s
of
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E V
o
l
.
5, No
. 1, Feb
r
uar
y
20
1
5
:
7
1
– 77
76
per
f
o
r
m
a
nce i
s
d
one
u
s
i
n
g L
I
VE
V
Q
A
dat
a
base,
C
S
I
Q
Vi
deo
dat
a
base a
n
d
V
Q
E
G
H
D
T
V Vi
de
o dat
a
base o
n
fo
ur
ki
n
d
s o
f
di
st
ort
i
o
ns t
h
r
o
u
g
h
spea
rm
an ra
nk
or
dere
d
correl
a
t
i
o
n co
effi
cent
.
T
h
e f
r
am
e work i
s
f
o
u
n
d
t
o
per
f
o
r
m
st
at
i
s
tical
l
y
bet
t
e
r t
h
a
n
ot
he
r
pr
op
os
ed
no
re
fere
nc
e al
go
ri
t
h
m
s
.
REFERE
NC
ES
[1]
Kalpana Sesh
adr
i
nath
an and A
.
C. Bovik, “Motion Tuned
Spa
tio-
t
em
poral Qua
lit
y Assessm
ent of Natural
Videos
”
,
vol. 19
, no
. 2
,
pp
. 335-350
,
I
EEE Transactions
on Image
Processing
, Feb
.
2010
.
[2]
M.A. Saad, A.C. Bovik and C.
Charrier
, “
A
DCT Statistics-Based Blind Im
age Qualit
y
I
ndex”,
IEEE Signa
l
Processing Letters
, pp. 583-586
,
vol. 17
, no
. 6
,
Ju
ne 2010.
[3]
H.R. Sheikh
.and
A.C. Bo
v
i
k, “
I
m
a
ge inform
atio
n and visual qu
ality
”
,
IEEE Tra
n
sactions on Im
age Processing
,
vol.15, no.2
,
pp
.
430- 444, Feb. 2
006.
[4]
H.R. Sheikh
.and
A.C. Bo
v
i
k, “
I
m
a
ge inform
atio
n and visual qu
ality
”
,
IEEE Tra
n
sactions on Im
age Processing
,
vol.15, no.2
,
pp
.
430- 444, Feb. 2
006.
[5]
H.R. Sheikh, A.C. Bovik and
G.
de Vecian
a, “
A
n inform
ation fi
delit
y
cr
iterion f
o
r image quality assessment using
natura
l sc
ene st
a
tistics”
,
I
E
EE Transactions on I
m
age Processin
g
, vol.14, no.12p
p. 2117-
2128, Dec.2005
.
[6]
X. Zhu and
P.
Milanfar
, “
A
utom
atic par
a
m
e
ter
selec
ti
on for
denoising algor
ith
ms using a no-referen
ce measure o
f
im
age con
t
en
t”,
IEEE
T
r
ans.Im
age Process
, vol.
19, no
. 12
, pp
. 3
116–3132, Dec.
2010.
[7]
E
h
sa
n
Ol
l
a
h Sh
ey
ba
n
i
,
“
An Algorithm
for Real-Time Blin
d
Im
age Quality Comparison and Assessment
”,
International Jo
urnal of
Electrical
and Computer Engin
eering
(
I
JECE)
,
pp. 120-1
29, vol. 2
,
no
. 1
,
Februar
y
2012
.
[8]
A.
Bua
d
e
s
,
B.
Coll,
a
nd J.
M.
More
l,
“
A non-local algorithm
for image denois
i
ng
”,
in P
r
oc. I
EEE Com
put. S
o
c.
Conf. Comput.
Vis. Pattern R
e
cogn
it., vol. 2. Jun. 2005
, pp
. 60–
65.
[9]
Akshay Gore, H.K. Kansal, Savita Gupta
, “
L
ocal Standa
rd Deviat
ion Base
d Im
age Quality Metri
c
s for JPE
G
Compressed Images”,
TELKOMNIKA, vol.
12
, n
o
. 10
, pp
. 7280-
7286, Octob
e
r 2
014.
[10]
F. Luisier,
T. Blu,
and M. Uns
e
r,
“A new SURE appro
ach
to
image d
e
noisin
g
: Inters
cale orthonormal wavelet
thresholding
”,
I
EEE
T
r
ans.Im
age Process
, vol. 1
6
, no
. 3
,
pp
. 593
–606, Mar
.
2007
.
[11]
K. Dabov, A. Foi, V. Ka
tkovnik, and K. Egiazarian
, “Image denoising by
sparse 3-D transform-domain
collaborative filtering”,
I
EEE Trans. Image
Process
, vol. 16
, no
.
8, pp
. 2080–209
5, Aug. 2007.
[12]
Wang Yu Ling, Yang Hu, “Machine Learning to De
s
i
gn F
u
ll-refe
re
nce Im
age Q
u
alit
y As
s
e
s
s
m
ent
Algorithm”,
TELKOMNIKA
,
pp. 3414-3421, vol. 11,
no. 6, June
2013.
[13]
A. Mittal
,
A. K. Moorth
y
and A. C. Bovik
,
“
N
o-referen
ce im
ag
e qualit
y assessm
ent in the spatial dom
ain”
,
IEE
E
Transactions on
Image Processing
, Vol: 21 No: 1
2
, Page(s
): 4695
-4708, December 2012.
[14]
Yang X.K.,
Ling
W.S., Lu
Z.K.,
Ong E.P.
and Y
a
o S.S. (2005) J
u
st noticeable
distortion model and
its applicatio
ns
in vid
e
o
coding.
Signal
Process, I
m
age commun
, 2
0
, 662–680
.
[15]
Chin Y.J. and B
e
rger T
.
(1999)
Asoftware-onl
y
videocod
ec usin
g pixelwise
con
d
itional
diff
eren
tial repl
enishm
ent
and per
cep
tual
e
nhancem
ents
.
IEEE Trans. Circu
its
Syst. Video
Technol
, 9
,
438–4
50.
[16]
Chou C.H. and
Chen C.W. (19
96) A percep
tually
op
timized 3-
D subband codec for vid
e
o com
m
unication ov
er
wire
le
ss c
h
a
nnels.
IEEE Trans. C
i
rcuits S
y
st. Vid
e
o Technol
, 6
,
14
3–156.
[17]
Ahumada A.J. and Peterson
H.A.
(1992) Luminance-M
odel-Based DC
T Quantization
for Color Image
Compression. SPIE Conf. on H
u
man Vision,
V
i
sual Processing, and
Digital
Display
III, Vol. 1
666, pp. 365–37
4,
SPIE
,
Sa
n Die
g
o.
[18]
Watson, A.B. (1
993)
DCT Quantization Ma
trices Visually
Optimized for Individual Images
. Pr
oc. SPIE, pp. 20
2–
216.
[19]
Seshadrinath
an K., Soundararajan R.,
Bovik A.C. and Cormack L.K. (2010)
Stud
y
of subjective and objective
quality
assessment of v
i
deo
.
I
E
EE Trans. Image
Process
, 19
, 142
7–1441.
[20]
Seshadrinath
an,
K. and Bovik, A.C.
(2010) Motion tuned spatiotemporal qua
lity
assessment of na
tural vid
e
os.
IEEE
Trans. Image Pr
ocess
, 19
, 335–3
50.
[21]
Wang, Z. and Li, Q. (2007)
Vid
e
o quali
ty
assessment using a sta
tistical model o
f
human visual
speed per
ceptio
n
.
JOSA, 24, B61–
B69.
[22]
Ninassi, A.,
Le
Meur, O., Le Callet, P
.
and B
a
rba, D.
(2009)
Considering tem
poral var
i
ations of spatial visu
al
distortions in
vid
e
o quality
assessment.
I
EEE J. S
e
l. Topi
cs Signal
Process
, 3
,
253–
265.
[23]
Lidong Wang, “The
Effect of
Force on
Fingerpri
nt Image Quality and Fing
erprint
Distortion”,
International Journ
a
l
of Electrical and
Comput
er
Eng
i
neer
ing (
I
JEC
E
)
, pp
. 294-300
, v
o
l. 3
,
no
. 3
,
June 2013.
[24]
Ba
rkowsky
M.,
Bia
l
kowski J.,
Eskofie
r
B.,
Bitto R.
a
n
d Ka
up A
.
(2009)
Te
m
pora
l
tra
j
e
c
t
ory
a
w
a
r
e
vide
o qua
lity
m
eas
ure.
I
EEE
J
.
Se
l.
T
op. S
i
gna
l Process
, 3, 266
–279.
[25]
H
e
ks
tra A
.
P
., B
eerends
J
.
G
.
, Le
derm
ann D
., D
e
Caluw
e
F
.
E.
,
K
ohler S
., K
o
en
en R.H
., Rihs
S
., Ehrs
am
M
.
an
d
Schlauss D. (200
2) PVQM—A p
e
r
cep
tual vid
e
o
quality
m
easure.
Signal Process.
Image commun
, 17,
781–798
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Lo
w b
it
Ra
te Vid
e
o
Qu
a
lity
An
a
l
ysis Usi
n
g NRDPF-VQA Alg
o
r
ithm
(Subra
h
ma
n
y
am
.C
h
)
77
BIOGRAP
HI
ES OF
AUTH
ORS
S
ubrahm
a
n
y
am
Ch rece
ived B.
T
ech degr
ee in
El
ectron
i
cs
and Co
m
m
unication En
ginee
i
ng from
Jawaharlal Nehr
u Technol
ogical University
, H
yderabad
,
Ind
i
a,
and M.Tech
degree
in Digital
Electronics
and
Communication Engine
eing fr
om JNTU, H
y
derab
d
, India.
He joined as a research schol
ar
(part time), ECE department
in Januar
y
2010
. His research
inter
e
sts includ
e signal
and im
age processing,
Computer visi
o
n
. He published
more than 20
res
earch
pap
e
rs
i
n
journa
ls
and
c
onferenc
e
s
.
Dr. Venkata Rao
published more
than 40
techn
i
cal
papers
in Inter
n
ation
a
l
and National Journals
and confer
enc
e
s
of interna
tion
a
l repu
te.
His
res
earch
int
e
res
t
s
include s
i
gn
al and
im
age
processing, wireless networks. He
carried out AICTE projects wo
rth Rs. 23 lakhs as Principal
Investigator and
Project coordinator. Dr.D
. Venkata Rao is rec
ognized guide under JNTU
Kakinada. Curr
ently
h
e
is
guidin
g
6 res
earch
sch
o
lars in
diff
eren
t univ
e
rsiti
es for
the
i
r Ph.D
in
the
areas of sign
al pro
cessing. I
m
age processing,
wireless networks. He gu
ided m
o
re th
an 50
UG
and PG projects.
He is Fellow Mem
b
er of Institu
t
i
on of Engine
ers (India) and m
e
m
b
er of Board
of Studies, Dep
a
rtment of
Electronics and
Inst
rumentation
En
gineer
ing, B
a
patla Eng
i
neering
coll
ege (Auto
n
o
m
ous
).
Dr. N Usha Ran
i
published
more than 25
technica
l p
a
pers in
Inte
rnation
a
l
and Na
tional
Journals
and conferen
ces of internation
a
l repute. Her
research in
ter
e
sts include signal and im
age
processing, VLSI. He carr
i
ed out AICTE projec
ts like MODROB
S and Project co
ordinator
.
She
is recogn
ized
gu
ide und
er Vignan University
. Cu
rre
ntly
sh
e is gu
iding 5 res
earch
scholars in
the
university
for
th
eir Ph.D
in
the areas of
signal pr
ocessing.
Evaluation Warning : The document was created with Spire.PDF for Python.