I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
p
ute
r
E
ng
in
ee
ring
(
I
J
E
CE
)
Vo
l.
10
,
No
.
2
,
A
p
r
il
2
0
2
0
,
p
p
.
1
3
1
7
~1
3
2
6
I
SS
N:
2
0
8
8
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v
1
0
i
2
.
p
p
1
3
1
7
-
1
3
2
6
1317
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m/in
d
ex
.
p
h
p
/I
JE
C
E
O
bje
cts d
etec
tion
a
nd t
ra
c
k
ing
usin
g
f
a
st princi
ple co
m
po
nen
t
purist and
k
a
l
m
a
n f
ilt
er
H
a
deel
N.
Abdu
lla
h
,
Nu
ha
H
.
Abdu
lg
ha
f
o
o
r
De
p
a
rtme
n
t
o
f
El
e
c
tri
c
a
l
En
g
in
e
e
rin
g
,
Un
iv
e
rsity
o
f
T
e
c
h
n
o
lo
g
y
,
Ir
a
q
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ma
y
2
4
,
2
0
1
9
R
ev
i
s
ed
Oct
1
5
,
2
0
1
9
A
cc
ep
ted
Oct
2
7
,
2
0
1
9
T
h
e
d
e
tec
ti
o
n
a
n
d
trac
k
in
g
o
f
m
o
v
in
g
o
b
jec
ts
a
tt
ra
c
ted
a
l
o
t
o
f
c
o
n
c
e
rn
b
e
c
a
u
se
o
f
th
e
v
a
st
c
o
m
p
u
ter
v
isio
n
a
p
p
li
c
a
ti
o
n
s.
T
h
is
p
a
p
e
r
p
r
o
p
o
se
s
a
n
e
w
a
lg
o
rit
h
m
b
a
se
d
o
n
se
v
e
ra
l
m
e
th
o
d
s
f
o
r
id
e
n
ti
fy
in
g
,
d
e
tec
ti
n
g
,
a
n
d
trac
k
in
g
a
n
o
b
jec
t
in
o
rd
e
r
t
o
d
e
v
e
lo
p
a
n
e
ff
e
c
ti
v
e
a
n
d
e
f
f
i
c
ien
t
sy
st
e
m
in
se
v
e
ra
l
a
p
p
li
c
a
ti
o
n
s.
T
h
is
a
lg
o
rit
h
m
h
a
s
th
re
e
m
a
in
p
a
rts:
th
e
f
irst
p
a
rt
f
o
r
b
a
c
k
g
ro
u
n
d
m
o
d
e
li
n
g
a
n
d
f
o
re
g
ro
u
n
d
e
x
trac
ti
o
n
,
th
e
se
c
o
n
d
p
a
rt
f
o
r
sm
o
o
th
i
n
g
,
f
il
t
e
rin
g
a
n
d
d
e
tec
ti
n
g
m
o
v
in
g
o
b
jec
ts
w
it
h
in
th
e
v
id
e
o
f
ra
m
e
a
n
d
th
e
las
t
p
a
rt
in
c
lu
d
e
s
trac
k
in
g
a
n
d
p
re
d
ictio
n
o
f
d
e
tec
ted
o
b
jec
ts.
In
th
is
p
r
o
p
o
se
d
w
o
rk
,
a
n
e
w
a
lg
o
rit
h
m
to
d
e
tec
t
m
o
v
in
g
o
b
jec
ts
f
ro
m
v
id
e
o
d
a
ta
is
d
e
sig
n
e
d
b
y
th
e
F
a
st
P
ri
n
c
ip
le
Co
m
p
o
n
e
n
t
P
u
rist
(F
P
CP
)
.
T
h
e
n
w
e
u
se
d
a
n
o
p
t
im
a
l
f
il
ter
th
a
t
p
e
rf
o
r
m
s
w
e
l
l
to
re
d
u
c
e
n
o
ise
t
h
ro
u
g
h
th
e
m
e
d
ian
f
il
ter.
T
h
e
F
a
s
t
Re
g
io
n
-
c
o
n
v
o
lu
ti
o
n
n
e
u
ra
l
n
e
tw
o
rk
s
(F
a
st
-
RCN
N)
is
u
se
d
to
a
d
d
sm
o
o
th
n
e
ss
to
th
e
sp
a
ti
a
l
id
e
n
ti
f
ica
ti
o
n
o
f
o
b
jec
t
s a
n
d
th
e
ir
a
re
a
s.
T
h
e
n
th
e
d
e
tec
ted
o
b
jec
t
is
trac
k
e
d
b
y
Ka
l
m
a
n
F
il
ter.
Ex
p
e
ri
m
e
n
tal
re
s
u
lt
s
s
h
o
w
th
a
t
o
u
r
a
lg
o
r
it
h
m
a
d
a
p
ts
to
d
if
fe
re
n
t
situ
a
ti
o
n
s a
n
d
o
u
tp
e
rf
o
rm
s
m
a
n
y
e
x
isti
n
g
a
lg
o
rit
h
m
s
.
K
ey
w
o
r
d
s
:
B
lo
b
an
aly
s
is
Fas
t
p
r
in
cip
le
co
m
p
o
n
e
n
t
p
u
r
i
s
t
(
FP
C
P
Kal
m
a
n
f
ilter
Ob
j
ec
t d
etec
tio
n
Ob
j
ec
t tr
ac
k
in
g
Co
p
y
rig
h
t
©
2
0
2
0
In
stit
u
te o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
rin
g
a
n
d
S
c
ien
c
e
.
Al
l
rig
h
ts re
se
rv
e
d
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Had
ee
l N
.
A
b
d
u
lla
h
,
Dep
ar
t
m
en
t o
f
E
lectr
ical
E
n
g
i
n
ee
r
in
g
,
Un
i
v
er
s
it
y
o
f
T
ec
h
n
o
lo
g
y
,
Al
-
S
in
aa
S
tr
ee
t,
B
ag
h
d
ad
,
I
r
aq
.
E
m
ail
:
3
0
0
0
2
@
u
o
ec
h
n
o
lo
g
y
.
e
d
u
.
iq
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
p
r
o
ce
s
s
o
f
tr
ac
k
in
g
i
s
o
n
e
o
f
th
e
m
ai
n
ta
s
k
s
in
co
m
p
u
t
er
v
is
io
n
[
1
]
.
I
t
h
as
an
i
m
p
o
r
t
an
t
t
u
r
n
i
n
m
an
y
ar
ea
s
o
f
r
esear
ch
,
s
u
c
h
as
m
o
v
e
m
e
n
t
o
b
j
ec
ts
,
r
ec
o
g
n
i
tio
n
an
d
a
n
al
y
s
is
h
u
m
an
a
n
d
n
o
n
h
u
m
a
n
Vital
it
y
,
3
D
r
ep
r
esen
tatio
n
,
m
o
b
ilit
y
i
n
v
eh
ic
les,
a
n
d
o
th
er
s
.
Ob
j
ec
t
tr
ac
k
i
n
g
is
th
e
m
o
s
t
co
m
m
o
n
in
a
u
to
m
ated
m
o
n
ito
r
i
n
g
ap
p
licatio
n
s
b
ec
a
u
s
e
th
e
in
d
i
v
id
u
a
l
h
u
m
a
n
e
m
p
lo
y
er
ca
n
n
o
t
m
an
a
g
e
t
h
e
co
n
tr
o
lled
ar
ea
,
esp
ec
iall
y
w
h
e
n
th
e
n
u
m
b
er
o
f
ca
m
er
as
r
aises
.
A
l
s
o
,
in
t
h
e
m
ed
ici
n
al
a
p
p
licatio
n
,
th
e
o
p
er
ato
r
ca
n
n
o
t
s
o
m
eti
m
es
a
n
al
y
s
e
th
e
v
id
eo
tak
e
n
b
y
th
e
d
e
v
ice
.
I
t
is
also
u
s
ed
in
s
ec
u
r
it
y
s
y
s
te
m
s
,
tr
a
f
f
ic
m
a
n
ag
e
m
e
n
t
s
y
s
te
m
s
,
a
n
d
o
th
er
s
.
T
h
e
tr
ac
k
in
g
s
y
s
te
m
ca
n
tr
ac
k
s
in
g
le
o
r
m
u
l
tip
le
an
i
m
atio
n
o
b
j
ec
ts
in
d
if
f
er
en
t
e
n
v
ir
o
n
m
e
n
ts
[
2
]
.
I
n
g
en
er
al
,
th
e
o
b
j
ec
t
d
etec
tio
n
an
d
tr
ac
k
in
g
s
y
s
te
m
i
n
clu
d
e
s
th
e
d
i
f
f
er
en
t
s
ta
g
es
lik
e
b
ac
k
g
r
o
u
n
d
s
u
b
tr
ac
tio
n
,
o
b
j
ec
t
d
etec
tio
n
,
an
d
o
b
j
ec
t tr
ac
k
in
g
as sh
o
w
n
i
n
th
e
b
lo
ck
d
ia
g
r
a
m
in
Fi
g
u
r
e
1
.
I
n
m
an
y
f
ield
s
o
f
i
m
a
g
e
p
r
o
ce
s
s
i
n
g
an
d
co
m
p
u
ter
v
is
io
n
ap
p
licatio
n
s
,
th
e
b
ac
k
g
r
o
u
n
d
m
o
d
elin
g
is
an
i
m
p
o
r
tan
t
s
tep
,
also
k
n
o
w
n
a
s
f
o
r
eg
r
o
u
n
d
d
etec
tio
n
,
w
h
ich
e
x
tr
ac
ts
t
h
e
f
o
r
eg
r
o
u
n
d
o
f
th
e
im
ag
e
f
o
r
s
u
b
s
eq
u
e
n
t
p
r
o
ce
s
s
in
g
(
s
u
ch
as
o
b
j
ec
t
s
el
ec
tio
n
an
d
id
en
tif
ica
tio
n
)
f
r
o
m
b
ac
k
g
r
o
u
n
d
.
T
h
ese
ar
e
th
e
m
o
s
t
i
m
p
o
r
tan
t
ar
ea
s
o
f
th
e
p
ictu
r
e,
w
h
ic
h
ar
e
ca
lled
o
b
j
ec
ts
s
u
ch
as
h
u
m
a
n
s
,
ca
r
s
,
tex
t
s
,
etc.
T
h
is
s
tag
e
m
a
y
b
e
af
ter
th
e
p
r
e
-
p
r
o
ce
s
s
in
g
p
h
a
s
e
o
f
th
e
i
m
a
g
e,
w
h
ich
m
a
y
i
n
cl
u
d
e
n
o
is
e
r
ed
u
ctio
n
o
f
i
m
ag
e
s
,
an
d
b
ef
o
r
e
th
e
s
u
b
s
eq
u
e
n
t
tr
ea
t
m
e
n
t
p
h
ase
s
u
ch
as
m
o
r
p
h
o
lo
g
y
,
etc
[
3
]
.
A
m
o
r
e
co
m
m
o
n
w
a
y
to
d
etec
t
m
o
v
i
n
g
o
b
j
ec
ts
in
v
id
eo
s
is
b
ac
k
g
r
o
u
n
d
s
u
b
tr
ac
tio
n
.
T
h
e
b
asic
p
r
in
cip
le
is
to
d
etec
t
m
o
v
i
n
g
o
b
j
ec
ts
f
r
o
m
t
h
e
d
if
f
er
e
n
ce
b
et
w
ee
n
th
e
c
u
r
r
en
t
f
r
a
m
e
a
n
d
t
h
e
r
ef
er
e
n
ce
f
r
a
m
e,
o
f
ten
ca
lled
a
"
b
ac
k
g
r
o
u
n
d
i
m
a
g
e"
o
r
"
b
ac
k
g
r
o
u
n
d
m
o
d
el"
.
So
m
e
co
m
m
o
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
10
,
No
.
2
,
A
p
r
il 2
0
2
0
:
1
3
1
7
-
1326
1318
m
et
h
o
d
s
in
th
is
ar
ea
i
n
cl
u
d
e
t
h
e
u
s
e
o
f
f
r
a
m
e
d
i
f
f
er
en
tiat
io
n
,
o
p
tical
f
lo
w
,
a
n
al
y
s
is
o
f
p
r
in
cip
al
co
m
p
o
n
e
n
t
s
,
b
ac
k
g
r
o
u
n
d
m
i
x
t
u
r
e
m
o
d
els [
4
]
.
T
h
e
f
o
r
eg
r
o
u
n
d
r
ep
r
ese
n
ts
th
e
o
b
j
ec
t
o
f
in
ter
est
in
a
s
ce
n
e.
T
o
r
e
p
r
esen
t
an
o
b
j
ec
t,
f
ea
tu
r
es
ar
e
ex
tr
ac
ted
th
at
u
n
iq
u
el
y
d
e
f
in
e
s
an
o
b
j
ec
t.
T
h
e
s
e
f
ea
tu
r
e
s
o
r
th
e
d
escr
ip
to
r
s
o
f
a
n
o
b
j
ec
t
ar
e
th
e
n
u
s
ed
to
tr
ac
k
th
e
o
b
j
ec
t.
A
f
ea
tu
r
e
i
s
a
n
i
m
a
g
e
p
atter
n
t
h
at
d
i
f
f
er
en
tia
tes
a
n
o
b
j
ec
t
f
r
o
m
its
n
ei
g
h
b
o
r
h
o
o
d
.
T
h
e
f
ea
tu
r
es
o
f
an
o
b
j
ec
t
ar
e
co
n
v
er
ted
in
to
d
escr
ip
to
r
s
,
also
r
ef
er
r
ed
to
as
ap
p
ea
r
an
ce
f
ea
tu
re
s
,
u
s
i
n
g
s
o
m
e
o
p
er
atio
n
s
ar
o
u
n
d
th
e
f
ea
t
u
r
es
[
2
]
.
T
h
e
co
m
m
o
n
l
y
u
s
ed
o
b
j
ec
t
r
ep
r
esen
tatio
n
s
f
o
r
tr
ac
k
i
n
g
ar
e
ce
n
tr
o
id
,
m
u
l
tip
le
p
o
in
t
s
,
r
ec
tan
g
u
lar
p
atch
,
an
d
co
m
p
l
ete
o
b
j
ec
t
c
o
n
to
u
r
,
etc.
w
h
ile
th
e
d
escr
ip
to
r
s
o
f
an
o
b
j
ec
t
s
u
c
h
as
p
r
o
b
a
b
ilit
y
d
en
s
itie
s
o
f
o
b
j
ec
t a
p
p
ea
r
an
ce
(
His
to
g
r
a
m
)
,
te
m
p
late,
b
lo
b
an
al
y
s
i
s
,
etc.
T
h
e
o
b
j
ec
t
tr
ac
k
in
g
is
to
s
elec
t
an
d
g
iv
e
in
d
i
v
id
u
al
p
ath
s
to
e
ac
h
o
b
j
ec
t
in
th
e
v
id
eo
s
eq
u
en
ce
.
Ob
j
ec
ts
ca
n
b
e
h
u
m
an
s
o
n
th
e
s
tr
ee
t,
ca
r
s
o
n
th
e
r
o
ad
,
p
lay
er
s
o
n
t
h
e
p
itch
,
o
r
f
r
o
m
a
g
r
o
u
p
o
f
an
i
m
als.
T
h
e
o
b
j
ec
t
is
tr
ac
k
ed
to
ex
tr
ac
t
th
e
o
b
j
ec
t,
id
en
ti
f
y
t
h
e
o
b
j
ec
t
an
d
tr
ac
k
it,
an
d
t
h
e
d
ec
is
io
n
s
r
elate
d
to
th
eir
ac
t
iv
i
ties
.
T
r
ac
e
o
b
j
ec
ts
ca
n
b
e
class
if
i
ed
as
p
o
in
ts
tr
ac
k
in
g
,
k
er
n
el
tr
ac
k
in
g
,
an
d
tr
ac
e
s
h
ad
o
w
i
m
ag
e
s
.
T
h
e
g
en
er
al
tech
n
iq
u
es o
f
tr
ac
k
i
n
g
s
u
c
h
as
Kal
m
a
n
Fil
ter
,
P
ar
ticle
Fil
ter
,
Me
an
Sh
if
t M
eth
o
d
,
etc
[
5
]
.
T
h
e
m
ai
n
co
n
tr
ib
u
tio
n
o
f
th
i
s
w
o
r
k
i
s
:
a.
T
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
u
s
es t
h
e
FP
C
P
tech
n
iq
u
e
to
ex
tr
ac
t th
e
m
o
tio
n
ar
ea
s
f
r
o
m
d
if
f
er
en
t
b
ac
k
g
r
o
u
n
d
s
o
f
th
e
ca
p
tu
r
ed
v
id
eo
f
r
a
m
e
w
i
th
o
u
t
th
e
n
ee
d
f
o
r
f
u
r
t
h
er
in
p
u
t
.
As
a
r
es
u
lt,
s
o
th
e
o
u
tp
u
ts
h
a
v
e
g
o
o
d
s
p
ee
d
an
d
ac
cu
r
ac
y
.
b.
Usi
n
g
t
h
e
m
et
h
o
d
o
f
a
n
al
y
s
i
n
g
t
h
e
s
p
atial
an
d
te
m
p
o
r
al
s
i
m
u
ltan
eo
u
s
l
y
to
s
elec
t
t
h
e
e
f
f
ec
tiv
e
p
i
x
els
in
th
e
m
o
tio
n
zo
n
e
s
an
d
to
d
eter
m
i
n
e
th
e
ar
ea
o
f
th
e
o
b
j
ec
t
at
t
h
e
s
a
m
e
ti
m
e
u
s
in
g
Fas
t
R
C
N
N
.
I
n
ad
d
itio
n
to
u
s
i
n
g
e
f
f
icie
n
t
tr
ac
k
in
g
te
ch
n
iq
u
e,
Kal
m
an
Fil
ter
i
s
th
u
s
a
n
ef
f
ic
ien
t
a
n
d
in
teg
r
ate
d
w
a
y
to
tr
ac
k
m
u
ltip
le
o
b
j
ec
ts
in
th
e
s
a
m
e
c
ap
tu
r
ed
v
id
eo
f
r
a
m
e.
T
h
is
p
ap
er
is
o
r
d
er
e
d
as
f
o
llo
w
s
.
Sec
tio
n
2
e
x
p
lain
s
th
e
r
elate
d
w
o
r
k
s
.
Sectio
n
3
ex
p
lain
s
Me
th
o
d
o
lo
g
ies
(
Ma
th
e
m
at
ical
B
ac
k
g
r
o
u
n
d
)
,
Sectio
n
4
ex
p
lain
s
t
h
e
p
r
o
p
o
s
ed
A
lg
o
r
ith
m
.
Sectio
n
5
ex
p
lai
n
s
th
e
r
esu
lts
a
n
d
p
er
f
o
r
m
an
ce
a
n
al
y
s
i
s
.
Fi
n
all
y
th
e
co
n
c
lu
s
io
n
s
in
Sectio
n
6
.
Fig
u
r
e
1
.
T
h
e
b
lo
ck
d
iag
r
am
o
f
o
b
j
ec
t a
n
d
tr
ac
k
in
g
s
y
s
te
m
2.
RE
L
AT
E
D
WO
RK
S
Sin
ce
th
e
la
s
t
f
e
w
d
ec
ad
es,
m
an
y
r
esear
ch
er
s
h
av
e
p
r
o
v
en
alg
o
r
ith
m
s
f
o
r
d
etec
tin
g
an
d
tr
ac
k
in
g
o
b
j
ec
ts
.
I
n
th
is
s
ec
tio
n
,
w
e
d
e
m
o
n
s
tr
ated
s
o
m
e
o
f
th
ese
alg
o
r
ith
m
s
r
elate
d
to
th
e
p
r
o
p
o
s
ed
s
y
s
te
m
.
A
cc
o
r
d
in
g
to
[
6
]
,
m
o
tio
n
b
r
i
m
is
ex
tr
ac
t
ed
in
p
o
lar
-
lo
g
co
o
r
d
in
ate,
t
h
en
t
h
e
g
r
ad
ie
n
t
o
p
er
ato
r
is
em
p
lo
y
ed
to
co
m
p
u
te
th
e
o
p
tical
f
lo
w
d
ir
ec
tl
y
i
n
e
v
er
y
m
o
tio
n
r
eg
io
n
s
,
t
h
en
t
h
e
o
b
j
ec
t
is
tr
ac
k
e
d
.
I
n
th
e
p
r
o
p
o
s
ed
w
o
r
k
in
[
7
]
,
th
e
ac
tiv
e
b
ac
k
g
r
o
u
n
d
r
ec
o
n
s
tr
u
cted
an
d
th
e
o
b
j
ec
t
s
ize
d
ete
r
m
in
ed
as
a
p
r
eli
m
in
ar
y
ta
s
k
,
t
o
ex
tr
ac
t
an
d
tr
ac
k
th
e
o
b
j
ec
t
in
th
e
f
o
r
eg
r
o
u
n
d
.
T
h
e
m
et
h
o
d
in
[
8
]
is
th
e
o
b
j
e
ct
d
etec
tio
n
is
d
o
n
e
b
y
Ga
u
s
s
ian
Mix
tu
r
e
Mo
d
el
(
GM
M)
,
an
d
th
e
tr
ac
k
i
n
g
is
d
o
n
e
b
y
Kal
m
an
Fil
ter
.
I
n
t
h
is
m
et
h
o
d
,
Ob
j
ec
t
d
etec
tio
n
is
d
eter
m
i
n
ed
b
ased
o
n
th
e
s
ize
o
f
t
h
e
f
o
r
eg
r
o
u
n
d
.
T
h
er
ef
o
r
e
E
r
r
o
r
s
w
ill
o
cc
u
r
in
d
eter
m
in
i
n
g
th
e
o
b
j
ec
t
s
u
ch
a
s
th
e
o
b
j
ec
t
an
d
its
s
h
ad
o
w
ar
e
m
er
g
ed
as
a
n
o
b
ject
o
r
r
ep
r
esen
tin
g
t
w
o
ad
j
ac
en
t
co
m
p
o
u
n
d
s
a
s
a
s
i
n
g
le
o
b
j
ec
t.
T
h
e
p
ap
er
[
9
]
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
Ob
jects d
etec
tio
n
a
n
d
tr
a
ck
in
g
u
s
in
g
fa
s
t p
r
in
cip
le
co
m
p
o
n
en
t p
u
r
is
t a
n
d
ka
lma
n
filt
er
(
H
a
d
ee
l N.
A
b
d
u
lla
h
)
1319
d
ev
elo
p
ed
;
th
e
alg
o
r
ith
m
i
n
cl
u
d
es
o
p
tical
f
lo
w
an
d
t
h
e
m
o
tio
n
v
ec
to
r
esti
m
atio
n
f
o
r
o
b
j
ec
t
d
etec
tio
n
an
d
tr
ac
k
in
g
.
T
h
e
d
etec
tio
n
an
d
t
r
ac
k
in
g
s
y
s
te
m
i
n
[
1
0
]
is
s
o
p
h
is
t
icate
d
d
ep
en
d
o
n
o
p
tical
f
lo
w
f
o
r
d
etec
tio
n
;
th
e
o
b
j
ec
t tr
ac
k
in
g
i
s
d
o
n
e
b
y
b
lo
b
an
aly
s
is
.
I
n
P
r
ab
h
ak
ar
et
al.
[
1
1
]
,
a
m
o
v
in
g
o
b
j
ec
t
tr
ac
k
in
g
s
y
s
te
m
u
s
in
g
m
o
r
p
h
o
lo
g
ical
p
r
o
ce
s
s
i
n
g
an
d
b
lo
b
an
al
y
s
is
,
w
h
ic
h
ab
le
to
d
is
tin
g
u
i
s
h
b
et
w
ee
n
ca
r
an
d
p
ed
estrian
i
n
th
e
s
a
m
e
v
id
eo
.
I
n
th
e
p
ap
er
[
1
2
]
,
th
e
f
o
r
eg
r
o
u
n
d
is
ex
tr
ac
ted
f
r
o
m
t
h
e
b
ac
k
g
r
o
u
n
d
u
s
i
n
g
m
u
ltip
l
e
-
v
ie
w
ar
ch
itect
u
r
e.
Af
ter
th
at,
th
e
f
o
r
w
ar
d
m
o
v
e
m
e
n
t
d
ate
an
d
ed
itin
g
s
c
h
e
m
e
s
ar
e
u
s
ed
to
d
etec
t
th
e
an
i
m
ated
o
b
j
ec
ts
.
Fin
all
y
,
b
y
d
etec
tin
g
th
e
ce
n
ter
o
f
g
r
av
it
y
o
f
t
h
e
m
o
v
in
g
o
b
j
ec
t,
it
i
s
u
s
ed
to
tr
ac
e
t
h
e
o
b
j
ec
t
b
ased
o
n
t
h
e
Kal
m
a
n
f
ilter
.
I
n
t
h
e
m
et
h
o
d
[
1
3
]
,
an
i
m
ated
o
b
j
ec
ts
ar
e
r
ep
r
esen
ted
as
g
r
o
u
p
s
o
f
s
p
atial
a
n
d
t
e
m
p
o
r
al
p
o
in
ts
u
s
in
g
t
h
e
Gab
o
r
3
D
f
ilter
,
w
h
ic
h
w
o
r
k
s
o
n
t
h
e
s
p
atia
l
an
d
te
m
p
o
r
al
an
al
y
s
i
s
o
f
t
h
e
s
eq
u
e
n
ti
al
v
id
eo
an
d
is
th
e
n
j
o
in
t
b
y
u
s
i
n
g
th
e
Min
i
m
u
m
Sp
an
n
in
g
T
r
ee
.
T
h
e
p
r
o
p
o
s
ed
tech
n
i
q
u
e
d
escr
ib
ed
in
[
1
4
]
,
s
p
lit
in
to
th
r
ee
s
ta
g
es;
Fo
r
eg
r
o
u
n
d
s
eg
m
e
n
tatio
n
s
tag
e
b
y
u
s
in
g
Mi
x
tu
r
e
o
f
A
d
a
p
tiv
e
Gau
s
s
ia
n
m
o
d
el,
tr
ac
k
in
g
s
ta
g
e
b
y
u
s
i
n
g
t
h
e
b
lo
b
d
etec
tio
n
an
d
ev
al
u
atio
n
s
tag
e
w
h
ich
in
cl
u
d
es
th
e
clas
s
i
f
icatio
n
ac
co
r
d
in
g
to
t
h
e
f
ea
tu
r
e
ex
tr
ac
tio
n
.
T
h
e
p
r
o
p
o
s
ed
w
o
r
k
i
n
[
1
5
]
,
it
m
er
g
ed
B
ac
k
g
r
o
u
n
d
Su
b
tr
ac
tio
n
w
it
h
L
o
w
R
an
k
tech
n
iq
u
es
f
o
r
e
f
f
ec
tiv
e
o
b
j
ec
t
d
etec
tio
n
.
Fin
all
y
th
e
s
u
g
g
ested
alg
o
r
ith
m
i
n
[
1
6
]
,
it
em
p
lo
y
ed
B
ac
k
g
r
o
u
n
d
Su
b
tr
ac
tio
n
an
d
K
-
Me
a
n
s
C
l
u
s
ter
in
g
te
ch
n
iq
u
es
to
d
etec
t
th
e
m
o
v
in
g
o
b
j
ec
t
an
d
tr
ac
k
in
g
it.
Af
ter
ex
p
lo
r
in
g
s
o
m
e
o
f
th
e
p
u
b
lis
h
ed
r
esear
ch
o
n
th
e
d
etec
tio
n
an
d
tr
ac
k
in
g
o
f
th
e
o
b
j
ec
t,
it
w
a
s
f
o
u
n
d
th
at
th
e
d
i
s
co
v
er
y
an
d
tr
ac
k
i
n
g
o
f
th
e
o
b
j
ec
t
is
a
co
m
p
le
x
tas
k
b
ec
au
s
e
o
f
m
a
n
y
ele
m
e
n
ts
o
f
d
y
n
a
m
ic
tr
ac
k
i
n
g
s
u
c
h
as
d
eter
m
in
i
n
g
t
h
e
t
y
p
e
o
f
ca
m
er
a
m
o
v
in
g
o
r
s
tatic,
t
h
e
r
an
d
o
m
ch
a
n
g
e
o
f
th
e
s
p
ee
d
o
f
th
e
o
b
j
ec
t,
th
e
in
t
en
s
it
y
o
f
li
g
h
t a
n
d
d
ar
k
n
es
s
,
etc.
3.
M
E
T
H
O
DO
L
O
G
I
E
S (
M
AT
H
E
M
AT
I
CAL B
ACK
G
RO
UND)
3
.
1
.
F
a
s
t
P
rincipa
l C
o
m
po
n
ent
P
urs
uit
FP
C
P
w
a
s
r
ec
en
tl
y
s
u
g
g
e
s
ted
as
a
p
o
w
er
f
u
l
alter
n
a
tiv
e
to
P
r
in
c
ip
al
C
o
m
p
o
n
e
n
t
s
An
al
y
s
is
(
P
C
A
)
.
T
h
is
m
e
th
o
d
w
il
l
b
e
u
s
ed
in
v
ar
io
u
s
ap
p
licatio
n
s
,
in
cl
u
d
in
g
f
o
r
eg
r
o
u
n
d
/b
ac
k
g
r
o
u
n
d
m
o
d
el
lin
g
,
d
ata
an
al
y
s
i
s
,
w
h
et
h
er
in
te
x
t o
r
v
id
eo
f
o
r
m
a
t a
n
d
i
m
a
g
e
p
r
o
ce
s
s
in
g
.
T
h
e
PC
A
w
as
f
o
r
m
u
lated
in
i
tiall
y
[
1
7
]
.
,
|
|
|
|
+
|
|
|
|
1
s
.
t.
D
= L
+ S
(
3
)
W
h
er
e
D
∈
R
m×
n
is
th
e
o
b
s
er
v
ed
m
atr
i
x
,
||
L||
o
is
th
e
n
u
clea
r
n
o
r
m
o
f
m
atr
ix
L
(
i.e
.
∑
|
(
)
|
)
an
d
|
|
S
|
|
1
is
th
e
l
1
n
o
r
m
o
f
m
atr
ix
S
.
Nu
m
er
o
u
s
ch
a
n
g
es
h
av
e
b
ee
n
m
ad
e
t
o
eq
.
(
1
)
b
y
ch
an
g
i
n
g
t
h
e
r
estri
ctio
n
s
o
n
s
an
ctio
n
s
an
d
v
ice
v
er
s
a.
So
th
at
t
h
e
eq
.
(
1
)
b
e
ca
m
e:
,
1
2
|
|
+
−
|
|
+
|
|
|
|
1
s
.
t.
||
L |
|
o
<
t
(
2
)
T
h
e
co
n
s
tr
ain
t
|
|
L
|
|
o
<
t
is
ac
tiv
e,
r
ep
r
esen
ts
a
co
n
s
tr
ain
t
o
f
eq
u
alit
y
,
s
o
it
is
s
u
g
g
ested
th
at
th
e
alg
o
r
it
h
m
r
an
k
s
t
h
e
s
a
m
e
r
ath
er
th
a
n
r
elax
t
h
e
n
u
clea
r
b
ase,
s
o
th
e
f
u
n
ctio
n
i
s
as
f
o
llo
ws
,
1
2
|
|
+
−
|
|
+
|
|
|
|
1
s
.
t.
r
a
n
k(
L)
≈
t
(
3
)
T
h
is
ad
j
u
s
t
m
en
t
i
g
n
o
r
ed
th
e
in
itial
s
elec
tio
n
o
f
t
h
e
p
ar
a
m
eter
λ
,
th
e
b
ac
k
g
r
o
u
n
d
m
o
d
ellin
g
c
o
m
p
o
u
n
d
L
is
o
f
te
n
lo
w
,
a
n
d
in
p
r
ac
tice,
t
h
er
e
is
n
o
d
if
f
ic
u
lt
y
i
n
s
e
lecti
n
g
th
e
ap
p
r
o
p
r
iate
v
alu
e
f
o
r
t
.
A
n
o
r
m
al
p
r
o
ce
s
s
to
s
o
lv
e
(
3
)
is
th
e
alter
n
ati
v
e
m
in
i
m
izat
io
n
:
+
1
=
|
|
+
−
|
|
s
.
t.
r
a
n
k
(
L)
≈
t
(
4
)
+
1
=
|
|
+
1
+
−
|
|
+
|
|
|
|
1
(
5
)
T
h
e
(
3
)
ca
n
b
e
s
o
lv
ed
b
y
tak
i
n
g
a
p
ar
tial
Si
n
g
u
lar
Val
u
e
D
ec
o
m
p
o
s
i
tio
n
SVD
o
f
(
D
-
S
k
)
w
i
th
r
e
s
p
ec
t
to
t.
w
h
ile
t
h
e
eq
.
4
ca
n
b
e
s
o
lv
ed
b
y
e
le
m
e
n
t
-
w
i
s
e
s
h
r
i
n
k
a
g
e.
T
h
e
b
ac
k
g
r
o
u
n
d
o
f
t
h
e
v
id
eo
s
i
s
s
u
p
p
o
s
e
d
to
lie
in
a
lo
w
-
lev
el
s
u
b
-
s
p
ac
e,
an
d
th
e
m
o
v
i
n
g
o
b
j
ec
ts
s
h
o
u
ld
b
e
in
th
e
f
o
r
eg
r
o
u
n
d
as
i
f
th
e
y
wer
e
g
r
ad
u
all
y
s
o
f
t
in
th
e
s
p
atial
a
n
d
te
m
p
o
r
al
d
ir
ec
tio
n
[
1
8
]
.
T
h
e
p
r
o
p
o
s
ed
m
et
h
o
d
in
teg
r
ates
t
h
e
Fro
b
en
i
u
s
a
n
d
l
1
-
n
o
r
m
b
a
s
e
in
to
a
u
n
i
f
ied
f
r
a
m
e
w
o
r
k
f
o
r
s
i
m
u
ltan
eo
u
s
n
o
i
s
e
r
ed
u
ctio
n
a
n
d
d
etec
tio
n
.
T
h
e
Fro
b
en
iu
s
b
ase
u
s
e
s
th
e
lo
w
-
le
v
el
p
r
o
p
er
ty
in
t
h
e
b
ac
k
g
r
o
u
n
d
; t
h
e
co
n
tr
ast is
i
m
p
r
o
v
ed
b
y
t
h
e
l
1
n
o
r
m
s
ta
n
d
ar
d
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
10
,
No
.
2
,
A
p
r
il 2
0
2
0
:
1
3
1
7
-
1326
1320
3
.
2
.
No
is
e
f
ilte
ring
An
i
m
ated
d
ig
ital
p
ictu
r
es
o
f
t
en
o
v
er
lap
w
it
h
a
s
et
o
f
n
o
is
e
b
ased
o
n
p
r
ev
ailin
g
co
n
d
itio
n
s
.
So
m
e
o
f
th
is
n
o
is
e
is
v
er
y
d
is
t
u
r
b
in
g
wh
en
i
m
p
licated
in
alter
i
n
g
t
h
e
i
n
ten
s
it
y
o
f
v
id
eo
f
r
a
m
es.
I
t
s
p
o
ils
p
ix
els
r
an
d
o
m
l
y
an
d
d
iv
id
es
i
n
to
t
w
o
e
x
tr
e
m
e
lev
els:
r
elati
v
el
y
lo
w
o
r
r
elativ
el
y
h
ig
h
,
co
m
p
ar
ed
to
ad
j
a
ce
n
t
p
ix
els
[
1
1
]
.
Su
b
s
eq
u
e
n
tl
y
,
it
is
n
ec
es
s
ar
y
t
o
ap
p
ly
r
ef
i
n
e
m
e
n
t
tech
n
icalit
ies
th
at
ar
e
ab
le
to
h
an
d
le
v
ar
i
o
u
s
t
y
p
es
o
f
n
o
is
e.
Mo
r
p
h
o
lo
g
ical
p
r
o
ce
s
s
es a
r
e
p
er
f
o
r
m
ed
to
ex
tr
ac
t
i
m
p
o
r
tan
t
f
ea
t
u
r
es o
f
u
s
ef
u
l
i
m
a
g
e
s
i
n
t
h
e
i
m
p
er
s
o
n
a
tio
n
o
f
s
h
ap
es
in
th
e
r
e
g
io
n
a
n
d
t
h
eir
d
escr
ip
tio
n
.
W
e
h
a
v
e
u
s
ed
b
o
th
th
e
m
o
r
p
h
o
lo
g
y
o
f
t
h
e
clo
s
u
r
e
a
n
d
co
r
r
o
s
io
n
,
r
esp
ec
tiv
el
y
,
to
r
e
m
o
v
e
p
ar
ts
o
f
th
e
r
o
ad
an
d
u
n
w
a
n
ted
th
in
g
s
.
Af
ter
th
e
m
o
r
p
h
o
lo
g
ic
al
clo
s
u
r
e
p
r
o
ce
s
s
,
p
r
o
v
id
ed
th
at
th
e
ap
p
ea
r
an
ce
o
f
th
e
o
b
j
ec
t is n
o
t d
estro
y
ed
,
an
d
th
at
m
a
n
y
s
m
a
ll p
u
n
ctu
r
e
s
an
d
s
ep
ar
ate
p
ix
el
s
ar
e
f
illed
in
th
e
f
o
r
m
o
f
o
n
e
la
r
g
e
r
ea
l o
b
j
ec
t.
T
h
e
f
o
llo
w
i
n
g
is
th
e
d
ef
i
n
itio
n
o
f
m
o
r
p
h
o
lo
g
i
ca
l c
lo
s
u
r
e
p
r
o
ce
s
s
an
d
th
e
ap
p
licab
le
s
tr
u
ct
u
r
al
el
e
m
en
t
B
.
P
*
B
=
(
P
⊕
B)
⊕
B
(
6
)
W
h
er
e:
=
[
0
0
1
0
1
0
1
0
0
]
(
7
)
T
h
e
m
atr
ix
P
,
w
h
ich
in
cl
u
d
es
m
o
v
i
n
g
o
b
j
ec
t
in
f
o
r
m
a
tio
n
,
i
s
o
b
tain
ed
th
r
o
u
g
h
th
e
d
etec
tio
n
p
r
o
ce
s
s
.
An
i
n
te
g
r
al
p
ar
t
o
f
t
h
e
m
o
r
p
h
o
lo
g
ical
ex
p
a
n
s
io
n
a
n
d
er
o
s
io
n
p
r
o
ce
s
s
es
is
a
s
tr
u
ct
u
r
al
ele
m
en
t
o
f
a
f
lat
s
h
ap
e.
T
h
er
e
is
a
b
in
ar
y
f
la
t
s
tr
u
ctu
r
e
ele
m
e
n
t
w
i
th
a
li
v
in
g
v
al
u
e,
e
ith
er
2
-
D
o
r
m
u
lti
-
d
i
m
en
s
io
n
a
l,
in
w
h
ic
h
t
h
e
r
ea
l
p
ix
els
ar
e
in
c
lu
d
ed
in
t
h
e
m
o
r
p
h
o
lo
g
ical
ca
lc
u
latio
n
,
a
n
d
f
al
s
e
p
ix
els
ar
e
n
o
t.
T
h
e
m
id
d
le
p
ix
el
o
f
t
h
e
s
tr
u
ct
u
r
e
ele
m
e
n
t,
ca
lled
th
e
p
ar
en
t,
d
et
er
m
in
e
s
th
e
p
i
x
el
i
n
th
e
i
m
a
g
e
b
ein
g
p
r
o
ce
s
s
ed
[
1
1
]
3
.
3
.
F
a
s
t
-
re
g
io
n
co
nv
o
lutio
n neu
ra
l net
w
o
rk
T
h
e
r
ef
er
en
ce
to
th
e
r
ec
en
t
ad
v
an
ta
g
es
o
f
co
n
v
o
l
u
tio
n
al
n
eu
r
al
n
et
w
o
r
k
s
(
C
NN)
h
as
b
ee
n
v
er
y
s
u
cc
e
s
s
f
u
lin
a
v
ar
iet
y
o
f
co
m
p
u
ter
v
is
io
n
tas
k
s
,
esp
ec
iall
y
th
o
s
e
ass
o
ciate
d
w
it
h
d
etec
ti
n
g
o
b
j
ec
ts
.
I
n
th
is
r
esear
ch
,
w
e
u
s
e
C
NN
n
et
wo
r
k
s
to
id
en
tify
t
h
e
co
m
m
er
cial
o
b
j
e
ct
as
a
s
u
p
er
v
is
ed
lear
n
in
g
tas
k
[
1
9
]
.
T
h
e
Fas
t
R
-
C
N
N
Ob
j
ec
t
Det
ec
tio
n
to
o
l
(
R
eg
io
n
s
w
it
h
C
o
n
v
o
l
u
tio
n
a
l
Neu
r
al
Net
w
o
r
k
s
)
is
u
s
ed
to
l
o
ca
te
d
etec
ted
o
b
j
ec
ts
th
at
ar
e
r
etu
r
n
ed
as
a
s
e
t
o
f
p
er
ip
h
er
al
b
o
x
es.
T
h
e
d
etec
to
r
h
as
h
ig
h
co
n
f
id
en
ce
i
n
th
e
d
is
co
v
er
ies.
An
n
o
tate
t
h
e
i
m
ag
e
w
it
h
th
e
s
u
r
r
o
u
n
d
in
g
s
q
u
ar
es
o
f
th
e
d
etec
to
r
an
d
th
e
co
r
r
esp
o
n
d
in
g
d
etec
tio
n
g
r
ad
es.
T
h
ese
p
er
ip
h
er
al
b
o
x
es w
er
e
ca
lled
ar
ea
s
u
g
g
e
s
tio
n
s
o
r
o
b
j
ec
t
s
u
g
g
e
s
tio
n
s
.
Z
o
n
e
s
u
g
g
e
s
tio
n
s
w
er
e
j
u
s
t
lis
ts
o
f
s
q
u
ar
e
b
o
x
es
w
it
h
a
s
m
all
p
r
o
b
ab
ilit
y
o
f
co
n
tain
i
n
g
an
o
b
j
ec
t
.
T
h
e
f
r
a
m
e
wo
r
k
o
f
Fa
s
t
R
-
C
NN
as
s
h
o
w
n
i
n
F
ig
u
r
e
2
[
2
0
]
.
I
n
m
an
y
ap
p
licatio
n
s
w
h
er
e
t
h
e
ca
lcu
latio
n
is
t
i
m
e
-
co
n
s
u
m
in
g
,
o
n
e
ca
n
u
s
e
p
o
in
t
an
al
y
s
is
to
eli
m
in
a
te
p
o
in
ts
th
at
d
o
n
o
t
m
atter
b
ased
o
n
s
p
ec
i
f
ic
s
p
atia
l
p
r
o
p
er
ties
an
d
r
etain
r
elev
a
n
t
p
o
in
ts
f
o
r
f
u
r
t
h
er
an
al
y
s
is
T
h
e
o
b
j
ec
t
co
r
r
esp
o
n
d
in
g
to
t
h
e
p
o
in
t
ar
ea
is
d
etec
ted
as
a
c
o
m
p
o
s
ite
o
b
ject
an
d
f
ea
t
u
r
es
as
a
b
o
u
n
d
ed
b
o
x
.
Fig
u
r
e
2
.
T
h
e
f
r
am
e
w
o
r
k
o
f
F
ast R
-
C
NN
3
.
4
.
K
a
l
m
a
n
f
ilte
r
Ob
j
ec
t
tr
ac
k
in
g
is
a
w
a
y
to
f
i
n
d
an
d
cr
ea
te
a
p
ath
to
th
e
o
b
j
ec
t
th
at
w
as
d
is
co
v
er
ed
.
I
n
t
h
is
s
ea
r
c
h
,
th
e
Kal
m
a
n
f
ilter
m
et
h
o
d
w
as
u
s
ed
to
tr
ac
k
a
n
o
b
j
ec
t in
s
eq
u
en
ce
w
it
h
ca
p
t
u
r
ed
v
id
eo
[
2
1
]
.
T
h
e
Kal
m
a
n
f
ilter
is
a
lin
ea
r
ap
p
r
o
ac
h
th
at
o
p
er
ates
in
t
w
o
b
as
ic
p
h
ases
o
f
p
r
ed
ictio
n
an
d
co
r
r
ec
tio
n
(
u
p
d
a
te)
.
T
h
e
p
r
ed
ictio
n
p
h
ase
is
ac
co
u
n
tab
le
f
o
r
th
e
s
co
o
p
o
f
th
e
n
ex
t
s
ta
te
an
d
p
o
s
itio
n
o
f
th
e
p
r
esen
t
o
b
j
ec
t.
Ho
w
ev
er
th
e
co
r
r
ec
tio
n
p
h
ase
p
r
o
v
id
es
th
e
p
ar
a
m
eter
s
w
it
h
t
h
eir
i
n
s
tan
ce
,
th
e
y
co
m
b
in
e
th
e
ac
t
u
al
m
ea
s
u
r
e
m
en
t
w
it
h
t
h
e
p
r
ev
io
u
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
Ob
jects d
etec
tio
n
a
n
d
tr
a
ck
in
g
u
s
in
g
fa
s
t p
r
in
cip
le
co
m
p
o
n
en
t p
u
r
is
t a
n
d
ka
lma
n
filt
er
(
H
a
d
ee
l N.
A
b
d
u
lla
h
)
1321
esti
m
ate
to
i
m
p
r
o
v
e
t
h
e
tr
aj
ec
to
r
y
w
h
er
e
t
h
e
o
b
j
ec
t
in
f
o
r
m
a
tio
n
d
etec
ted
in
t
h
e
p
r
ev
io
u
s
f
r
a
m
e
i
s
u
s
ed
an
d
p
r
o
v
id
es
an
esti
m
ate
o
f
t
h
e
o
b
j
ec
t'
s
n
e
w
p
o
s
it
io
n
.
T
h
e
Kalm
an
f
ilter
h
a
s
th
e
ab
ilit
y
to
r
atin
g
th
e
tr
ac
k
i
n
g
lo
ca
tio
n
s
w
it
h
m
in
i
m
al
d
at
u
m
o
n
th
e
lo
ca
tio
n
o
f
t
h
e
o
b
j
ec
t.
I
n
itiall
y
,
t
h
e
s
tat
u
s
St
an
d
m
ea
s
u
r
e
m
en
t
Xt
p
ar
ad
ig
m
ar
e
d
eter
m
i
n
ed
to
p
r
ed
ict
th
e
n
ex
t sit
e
a
s
s
h
o
w
n
i
n
Fi
g
u
r
e
3
.
T
h
e
Kal
m
an
f
il
ter
is
a
r
ep
etiti
v
e
alg
o
r
it
h
m
to
esti
m
ate
t
h
e
e
v
o
lv
i
n
g
s
tate
o
f
a
p
r
o
ce
s
s
w
h
en
m
a
k
i
n
g
m
ea
s
u
r
e
m
e
n
t
s
i
n
th
e
p
r
o
ce
s
s
.
T
h
e
Kal
m
an
f
i
lter
is
li
n
ea
r
w
h
e
n
th
e
e
v
o
l
u
tio
n
o
f
t
h
e
ca
s
e
f
o
llo
w
s
t
h
e
li
n
ea
r
m
o
tio
n
m
o
d
el
an
d
t
h
e
m
ea
s
u
r
e
m
en
ts
ar
e
li
n
ea
r
f
u
n
ctio
n
s
o
f
th
e
ca
s
e.
T
h
e
Kal
m
a
n
f
ilter
c
an
ac
cu
r
atel
y
tr
ac
k
m
o
tio
n
b
ased
o
n
ad
ap
tiv
e
f
ilt
er
in
g
u
s
i
n
g
th
e
s
tate
ar
ea
m
o
d
el.
I
t
is
n
ec
ess
a
r
y
to
d
esig
n
a
d
y
n
a
m
ic
m
o
d
el
o
f
th
e
tar
g
et
m
o
v
e
m
e
n
t
w
h
ic
h
i
s
th
e
f
i
x
ed
s
p
ee
d
m
o
d
el
(
C
V)
,
w
h
ic
h
as
s
u
m
e
s
t
h
at
t
h
e
s
p
ee
d
is
co
n
s
ta
n
t
d
u
r
in
g
th
e
s
a
m
p
li
n
g
p
er
io
d
.
T
h
is
m
o
d
el
h
as b
ee
n
u
s
ed
in
v
ar
io
u
s
ap
p
licatio
n
s
f
o
r
its
ea
s
e
an
d
e
f
f
e
ctiv
e
n
ess
[
2
2
]
.
Fig
u
r
e
3
.
T
h
e
b
lo
ck
d
iag
r
am
o
f
Kal
m
an
Fil
ter
A
l
g
o
r
ith
m
W
h
er
e;
A
-
s
tate
tr
an
s
itio
n
m
atr
ix
,
B
-
co
v
er
ts
co
n
tr
o
l
i
n
p
u
t
,
Q
-
p
r
o
ce
s
s
n
o
is
e
co
v
ar
ia
n
c
e
K
-
Kal
m
a
n
g
a
in
,
X
-
m
ea
s
u
r
e
m
e
n
t
m
atr
ix
,
S
-
m
ea
s
u
r
e
m
en
t e
r
r
o
r
co
v
ar
ian
ce
a
n
d
H
-
m
o
d
el
m
a
tr
ices.
T
h
e
p
r
ed
ictio
n
o
f
th
e
n
ex
t
s
tate
S
t
+
1
is
d
o
n
e
b
y
in
teg
r
ati
n
g
th
e
ac
t
u
al
m
ea
s
u
r
e
m
e
n
t
w
i
th
th
e
p
r
e
-
esti
m
ate
o
f
th
e
s
itu
a
ti
o
n
S
t
-
1
.
4.
T
H
E
P
RO
P
O
SE
D
SYS
T
E
M
I
n
th
is
p
ag
er
,
o
b
j
ec
t d
etec
tio
n
an
d
tr
ac
k
in
g
alg
o
r
it
h
m
,
a
co
ll
ec
tio
n
o
f
t
w
o
f
a
m
o
u
s
co
m
p
u
te
r
v
is
ib
ilit
y
tech
n
o
lo
g
ies,
Fas
t
P
r
in
cip
le
C
o
m
p
o
n
en
t
P
u
r
is
t
(
FP
C
P
)
an
d
t
h
e
Kal
m
a
n
f
ilter
,
w
a
s
i
n
tr
o
d
u
ce
d
.
FP
C
P
is
u
s
ed
i
n
th
e
o
b
j
ec
t
d
is
co
v
er
y
p
h
a
s
e.
I
t
p
r
o
v
id
es
q
u
ic
k
an
d
d
elica
te
o
b
j
ec
t
d
etec
tio
n
o
n
o
t
h
er
m
e
th
o
d
s
s
u
ch
a
s
b
ac
k
g
r
o
u
n
d
s
u
b
tr
ac
tio
n
.
FP
C
P
d
o
es
n
o
t
p
r
o
v
id
e
th
e
p
ath
o
f
m
o
tio
n
,
in
s
t
ea
d
it
s
u
p
p
lies
ac
q
u
ain
ta
n
ce
ab
o
u
t
th
e
o
r
ien
tatio
n
o
f
th
e
o
b
j
ec
t a
n
d
its
m
o
tio
n
i
n
v
ec
to
r
f
o
r
m
.
T
h
e
P
r
o
p
o
s
ed
alg
o
r
ith
m
o
f
FP
C
P
as sh
o
w
n
in
Alg
o
r
it
m
1
.
A
l
g
o
r
ith
m
1
: FP
C
P
A
lg
o
r
it
h
m
-
I
n
p
u
t :
C
ap
t
u
r
ed
Vid
eo
V
∈
,
R
eg
u
lar
iza
tio
n
P
ar
a
m
eter
s
λ
=
2
0
/(
Ma
x
(
s
ize(
V)
,
F),
μ
o
=0
.
0
1
,
T
o
le
r
an
ce
Valu
e
τ
.
I
n
itializatio
n
: S
o
=
0
,
L
o
=
0
,
R
an
k
r
=1
,
k
=0
.
W
h
ile
n
o
co
n
v
er
g
e
n
ce
d
o
[
U
Ʃ
V
to
l]
=
lm
S
VD(
D
-
S
(
k
)
,
r
)
L
(
k+
1)
=
U
*
s
h
r
i
n
k
(
Ʃ
,
1
/μ
k
)
*
V
T
S
(k+
1)
=
s
h
r
in
k
(
D
-
L
(
k+
)
,
λ
)
I
f
∑
=
>
th
e
n
r
=r
+1
%
co
n
v
er
g
en
ce
co
n
d
it
io
n
.
(
ξ =
‖
+
−
‖
)
<
to
l
k
=k
+1
μ
k
+
1
=
m
in
(
τ
*
μ
k
, μ
m
ax
)
E
n
d
o
f
W
h
ile
Ou
tp
u
t: th
e
o
p
ti
m
al
v
al
u
e
s
o
f
L
an
d
S
T
h
is
alg
o
r
ith
m
h
as
a
s
i
m
p
ler
an
d
clea
r
er
s
t
y
le.
S
in
ce
t
h
e
v
id
eo
b
ac
k
g
r
o
u
n
d
m
o
d
eli
n
g
ap
p
licatio
n
co
m
p
o
n
e
n
t
L
is
u
s
u
a
ll
y
v
er
y
l
o
w
,
w
e
s
u
g
g
est a
s
i
m
p
le
p
r
o
ce
d
u
r
e
f
o
r
esti
m
ati
n
g
th
e
u
p
p
er
li
m
it o
f
r
,
s
o
w
e
ca
n
esti
m
ate
th
e
co
n
tr
ib
u
tio
n
o
f
th
e
n
e
w
s
i
n
g
le
v
ec
to
r
;
if
t
h
i
s
in
p
u
t
is
s
m
all
e
n
o
u
g
h
,
w
e
s
to
p
in
cr
ea
s
i
n
g
t
h
e
v
a
lu
e
o
f
r
.
I
n
o
u
r
ex
p
er
i
m
en
tal
s
i
m
u
latio
n
s
,
th
i
s
co
r
r
esp
o
n
d
s
to
a
n
esti
m
ate
o
f
t
h
e
o
r
d
er
m
ad
e
b
y
t
h
e
i
n
e
x
ac
t
AL
M.
T
h
is
alg
o
r
ith
m
is
u
s
ed
L
MSV
D
(
L
i
m
ited
Me
m
o
r
y
Si
n
g
u
lar
Valu
e
Dec
o
m
p
o
s
itio
n
s
)
.
T
h
is
f
u
n
ctio
n
is
b
ased
o
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
10
,
No
.
2
,
A
p
r
il 2
0
2
0
:
1
3
1
7
-
1326
1322
a
b
lo
ck
Kr
y
lo
v
s
u
b
s
p
ac
e
o
p
ti
m
izatio
n
tec
h
n
iq
u
e
w
h
ic
h
s
i
g
n
i
f
ica
n
tl
y
ac
ce
ler
ate
s
th
e
cl
ass
ic
s
i
m
u
lta
n
eo
u
s
iter
atio
n
m
et
h
o
d
[
2
3
]
.
Ou
r
co
m
p
u
tat
io
n
al
r
e
s
u
l
ts
al
s
o
s
h
o
w
t
h
at
o
u
r
p
r
o
p
o
s
ed
alg
o
r
ith
m
h
a
s
lo
w
m
e
m
o
r
y
s
p
ac
e
th
at
ca
n
b
e
ap
p
lied
in
r
e
al
-
tim
e
v
id
eo
ap
p
licatio
n
s
,
f
aster
th
an
m
o
d
er
n
ap
p
licatio
n
s
an
d
d
eliv
er
co
m
p
ar
ab
le
q
u
alit
y
r
es
u
lt
s
.
T
h
e
p
r
o
p
o
s
ed
s
y
s
te
m
a
s
s
h
o
w
in
in
Fi
g
u
r
e
4
.
I
t
s
h
o
w
s
th
e
d
e
tail
s
tag
e
s
o
f
th
e
s
y
s
te
m
ar
e
f
o
llo
w
ed
to
th
at
d
ea
ls
w
it
h
b
ac
k
g
r
o
u
n
d
s
ep
ar
atio
n
,
o
b
j
ec
t
an
d
f
ea
tu
r
e
e
x
tr
ac
tio
n
.
A
t
f
ir
s
t,
th
e
v
id
eo
is
ta
k
en
b
y
th
e
s
tatio
n
ar
y
ca
m
er
a.
T
h
e
v
id
eo
is
o
n
l
y
a
s
er
ies
o
f
ca
s
ca
d
in
g
f
r
a
m
e
s
,
s
o
th
e
o
b
j
ec
t
d
et
ec
tio
n
m
an
n
e
r
m
u
s
t
f
ir
s
t
d
etec
t
th
e
m
o
v
i
n
g
o
b
j
ec
t
in
th
ese
c
ascad
in
g
f
r
a
m
es.
T
h
en
th
e
al
g
o
r
ith
m
co
n
v
er
ts
t
h
e
v
id
eo
i
n
to
t
w
o
d
i
m
en
s
io
n
m
atr
ices
to
f
ac
ilit
ate
h
a
n
d
li
n
g
in
m
at
h
e
m
atica
l
ca
lcu
la
tio
n
s
,
to
r
ed
u
ce
th
e
tim
e
ca
lcu
l
atio
n
an
d
m
e
m
o
r
y
r
eq
u
ir
e
m
en
ts
T
h
e
p
r
o
ce
s
s
i
n
g
i
n
clu
d
e
s
r
e
m
o
v
i
n
g
th
e
n
o
i
s
e.
T
h
en
th
e
p
r
o
ce
s
s
s
tatu
s
o
f
all
p
i
x
el
i
s
te
s
ted
b
y
Fas
t
R
C
N
N
an
d
cl
u
s
ter
ed
it
t
o
d
etec
t
th
e
o
b
j
ec
t.
I
n
itialize
th
e
tr
ac
k
in
g
s
tag
e
a
n
d
u
p
d
ate
th
e
tr
ac
k
er
in
e
v
er
y
n
e
w
f
r
a
m
e.
T
h
is
s
y
s
te
m
h
as m
a
n
y
f
ea
tu
r
es,
in
cl
u
d
in
g
th
e
p
o
s
s
ib
ili
t
y
o
f
tr
ac
k
in
g
m
o
r
e
th
a
n
o
n
e
o
b
j
ec
t
an
d
th
e
s
p
ee
d
o
f
r
esp
o
n
s
e
to
a
c
h
an
g
e
i
n
s
p
e
ed
an
d
ch
an
g
e
i
n
t
h
e
s
ca
le
.
Fi
g
u
r
e
4
s
h
o
w
s
th
e
o
r
ig
in
al
v
id
e
o
w
as
co
n
v
er
ted
to
in
d
iv
id
u
al
f
r
a
m
es to
b
e
p
r
o
ce
s
s
ed
in
n
ex
t sta
g
e
s
.
T
h
e
f
o
r
eg
r
o
u
ial
n
d
w
a
s
ex
tr
ac
ted
b
y
th
e
F
P
C
P
d
etec
tio
n
an
d
s
h
o
w
ed
h
o
les
a
n
d
n
o
i
s
e
o
n
t
h
e
f
r
a
m
e.
I
n
o
r
d
er
to
clar
if
y
an
d
s
o
f
ten
th
e
f
r
a
m
e,
t
h
e
m
o
r
p
h
o
lo
g
y
w
a
s
d
o
n
e.
T
h
e
f
o
r
eg
r
o
u
n
d
f
r
a
m
e
is
s
h
o
w
n
a
f
ter
th
e
m
o
r
p
h
o
lo
g
ical
p
r
o
ce
s
s
an
d
th
e
f
in
al
o
u
tp
u
t
f
r
a
m
e
w
h
ic
h
in
cl
u
d
es
tr
ac
k
ed
o
b
j
ec
ts
w
it
h
b
o
u
n
d
in
g
b
o
x
es.
Fig
u
r
e
4
.
T
h
e
f
o
w
ch
ar
t o
f
t
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
5.
RE
SU
L
T
S AN
D
P
E
RF
O
RM
ANCE AN
AL
Y
SI
S
T
h
e
alg
o
r
ith
m
p
r
o
p
o
s
ed
in
MA
T
L
A
B
(
2
0
1
8
b
)
h
as
b
ee
n
ap
p
lied
,
an
d
th
eir
ex
p
er
im
en
ts
w
er
e
p
er
f
o
r
m
ed
o
n
a
C
o
m
p
u
ter
t
y
p
e
MSI
GV6
3
w
it
h
I
n
tel
C
o
r
e
i7
8
7
5
0
H,
NVI
DI
A
R
T
X
2
0
6
0
6
G,
2
5
6
GB
SS
D+
1
T
B
an
d
1
6
GB
R
A
M.
I
t
h
as
th
r
ee
s
ta
g
es
ar
e
f
o
r
eg
r
o
u
n
d
d
etec
tio
n
,
f
ilter
i
n
g
an
d
tr
ac
k
in
g
.
T
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
d
etec
t
s
t
h
e
m
o
v
ab
l
e
o
b
j
ec
ts
ac
cu
r
atel
y
a
n
d
k
ee
p
s
tr
ac
k
o
f
t
h
eir
ap
p
ea
r
an
ce
i
n
th
e
s
eq
u
e
n
ce
v
id
eo
f
r
a
m
e
s
.
Vid
eo
d
ata
h
as
b
ee
n
u
s
ed
in
a
n
y
f
o
r
m
at
a
s
a
n
in
p
u
t
t
o
th
e
p
r
o
p
o
s
ed
w
o
r
k
,
a
n
d
g
o
o
d
r
esu
lts
h
a
v
e
b
ee
n
o
b
tain
ed
in
v
ar
io
u
s
ar
ticle
co
n
d
itio
n
s
o
n
t
h
is
i
n
d
o
o
r
,
o
u
td
o
o
r
,
lig
h
t
tr
a
f
f
ic
an
d
d
en
s
e
tr
a
f
f
ic
.
T
h
e
ef
f
icien
c
y
o
f
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
w
as e
v
alu
ated
,
th
e
e
x
p
er
ien
tial o
u
tco
m
es
w
er
e
as
f
o
llo
w
s
Sev
er
al
s
a
m
p
led
v
id
eo
is
u
s
ed
in
th
e
v
ar
io
u
s
e
n
v
ir
o
n
m
e
n
t
in
o
r
d
er
to
test
th
e
p
er
f
o
r
m
an
c
e
o
f
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
.
T
h
e
ex
p
er
ien
ti
al
o
u
tco
m
es
ar
e
g
i
v
en
i
n
Fi
g
u
r
e
5
.
T
h
e
f
ir
s
t
co
lu
m
n
as
s
h
o
w
n
i
n
Fig
u
r
e
5(
a)
s
h
o
w
s
t
h
e
s
a
m
p
le
d
f
r
a
m
es
o
f
th
e
v
id
eo
th
en
th
e
s
ec
o
n
d
co
lu
m
n
as
s
h
o
w
n
i
n
Fig
u
r
e
5
(
b
)
s
h
o
w
s
a
clea
n
f
o
r
eg
r
o
u
n
d
e
x
tr
ac
ted
f
r
a
m
es
b
y
FP
C
P
d
etec
tio
n
a
r
e
g
iv
e
n
.
T
h
e
th
ir
d
co
lu
m
n
a
s
s
h
o
w
n
in
Fi
g
u
r
e
5
c)
in
cl
u
d
es
th
e
d
etec
ted
an
d
tr
ac
k
ed
o
b
j
ec
ts
b
y
m
ar
k
i
n
g
it
w
it
h
a
cir
cu
m
f
er
en
tial
b
o
x
.
T
ab
le
1
ex
p
lain
s
th
e
m
ea
n
ex
ec
u
t
io
n
t
i
m
e
s
(
s
ec
)
o
f
t
h
e
p
r
o
p
o
s
ed
s
y
s
te
m
w
h
e
n
i
m
p
le
m
en
ted
o
n
3
6
5
s
a
m
p
led
v
id
e
o
f
r
a
m
e
s
b
y
Ma
tlab
v
er
s
io
n
2
0
1
8
b
.
A
cc
u
r
ac
y
is
a
m
ea
s
u
r
e
o
f
t
h
e
p
er
f
o
r
m
an
ce
ef
f
icie
n
c
y
o
f
t
h
e
o
b
j
ec
t
tr
ac
k
i
n
g
s
y
s
te
m
.
T
h
e
d
etec
tio
n
an
d
tr
ac
k
in
g
s
y
s
te
m
p
r
ec
is
io
n
ca
n
b
e
ca
lcu
late
d
u
s
in
g
t
h
e
n
e
x
t
f
o
r
m
u
la:
A
c
c
ura
c
y
=
The
tota
l
n
umb
e
r
of
de
te
c
te
d
ob
je
c
ts
by
s
ys
te
m
The
tota
l
n
umb
e
r
of
a
c
tua
l
ob
je
c
ts
in
vide
o
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
Ob
jects d
etec
tio
n
a
n
d
tr
a
ck
in
g
u
s
in
g
fa
s
t p
r
in
cip
le
co
m
p
o
n
en
t p
u
r
is
t a
n
d
ka
lma
n
filt
er
(
H
a
d
ee
l N.
A
b
d
u
lla
h
)
1323
(
a)
(
b
)
(
c)
Fi
g
u
r
e
5
.
T
h
e
p
r
o
p
o
s
ed
alg
o
r
it
h
m
r
e
s
u
lt
s
o
n
s
e
v
er
al
ca
p
tu
r
ed
v
id
eo
,
(
a)
Or
ig
in
al
f
r
a
m
q
,
(
b
)
T
h
e
f
ilt
er
ed
f
o
r
eg
r
o
u
n
d
an
d
(
c)
T
h
e
r
esu
lta
n
t
f
r
a
m
e
T
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
f
o
r
t
h
e
d
if
f
er
e
n
t
v
id
eo
in
p
u
t
h
a
s
b
ee
n
test
ed
w
ith
d
i
f
f
er
en
t
m
et
h
o
d
s
to
ev
alu
ate
its
ac
cu
r
ac
y
.
T
h
e
ac
cu
r
ac
y
o
f
t
h
e
p
r
o
p
o
s
ed
tr
ac
k
er
in
d
if
f
er
e
n
t i
n
p
u
t
s
ce
n
e
s
w
as c
o
m
p
ar
ed
an
d
co
m
p
ar
ed
w
it
h
o
th
er
tr
ac
k
i
n
g
s
y
s
te
m
s
,
a
s
s
h
o
w
n
in
T
ab
le
2
.
I
t
s
h
o
w
s
t
h
at
th
e
d
etec
tio
n
a
n
d
tr
ac
k
i
n
g
ac
cu
r
ac
y
r
ate
u
s
i
n
g
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
s
is
1
0
0
%.
T
h
e
r
esu
lts
ar
e
o
p
ticall
y
ac
ce
p
tab
le
ex
ce
p
t
f
o
r
an
alg
o
r
ith
m
t
h
at
p
r
o
v
es
th
a
t
th
is
m
u
lti
-
o
b
j
ec
t
tr
ac
k
in
g
m
eth
o
d
is
v
alid
ated
.
I
t
w
a
s
co
n
clu
d
ed
th
at
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
w
as
s
ti
ll
co
m
p
eti
tiv
e,
alt
h
o
u
g
h
s
o
m
e
r
esu
lt
s
w
er
e
clo
s
est
to
a
n
o
th
e
r
,
w
it
h
litt
le
d
e
g
r
ad
atio
n
in
s
o
m
e
a
s
a
r
esu
lt
o
f
th
e
co
s
t
o
f
s
o
m
e
co
m
p
le
x
ca
l
cu
latio
n
s
.
T
h
e
d
etec
tio
n
p
r
ec
i
s
io
n
o
f
t
h
e
s
u
g
g
e
s
ted
alg
o
r
it
h
m
i
s
co
m
p
ar
ed
w
it
h
o
th
er
k
n
o
w
n
a
n
d
p
r
esen
t
m
et
h
o
d
s
.
T
h
e
co
m
p
ar
is
o
n
s
h
o
w
s
th
e
ef
f
icie
n
t
p
er
f
o
r
m
a
n
ce
o
f
t
h
e
p
r
o
p
o
s
ed
m
eth
o
d
o
n
s
o
m
e
o
f
th
e
s
elec
ted
f
r
a
m
es
s
h
o
w
n
i
n
Fig
u
r
e
6
.
W
e
co
m
p
ar
ed
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
w
i
th
t
h
e
m
o
s
t
r
ep
r
esen
tativ
e
alg
o
r
it
h
m
s
a
n
d
f
o
r
d
if
f
er
e
n
t
f
r
a
m
e
s
ize
s
a
n
d
s
etti
n
g
s
f
o
r
t
h
e
tes
ted
v
id
eo
s
;
w
e
u
s
ed
g
r
a
y
o
r
ch
r
o
m
atic
v
id
eo
s
eq
u
en
ce
s
.
T
h
e
r
esu
l
ts
w
er
e
co
m
p
ar
ab
le
to
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
w
it
h
o
th
er
alg
o
r
it
h
m
s
.
T
ab
le
1
.
P
er
f
o
r
m
a
n
ce
ex
ec
u
tio
n
ti
m
e
T
h
e
S
t
e
p
M
e
a
n
Ex
e
c
u
t
i
o
n
T
i
me
(
S
e
c
o
n
d
)
L
o
a
d
i
n
g
t
h
e
v
i
d
e
o
a
n
d
c
o
n
v
e
r
si
o
n
t
o
mat
r
i
x
9
P
r
e
p
r
o
c
e
ssi
n
g
a
n
d
f
r
a
me
r
e
a
d
i
n
g
5
.
2
F
o
r
e
g
r
o
u
n
d
d
e
t
e
c
t
i
o
n
1
2
.
2
N
o
i
se
R
e
mo
v
a
l
1
.
1
5
T
h
e
d
e
t
e
c
t
i
o
n
p
r
o
c
e
ss
6
.
2
T
r
a
c
k
i
n
g
S
y
st
e
m
4
T
o
t
a
l
3
7
.
7
5
T
ab
le
2
.
P
er
ce
n
tag
e
ac
cu
r
ac
y
%
C
o
mp
a
r
i
so
n
P
a
r
a
me
t
e
r
B
a
c
h
g
r
o
u
n
d
S
u
b
t
r
a
c
t
i
o
n
[
1
5
]
G
M
M
M
e
t
h
o
d
[
8
]
O
p
t
i
c
a
l
F
l
o
w
M
e
t
h
o
d
[
1
0
]
P
r
o
p
o
se
d
T
r
a
c
k
i
n
g
S
y
st
e
m
S
i
n
g
l
e
H
u
ma
n
1
0
0
1
0
0
90
1
0
0
S
p
e
e
d
D
i
v
e
r
si
t
y
80
80
10
85
V
i
e
w
P
o
i
n
t
D
i
f
f
e
r
e
n
c
e
85
90
90
90
F
i
x
e
d
O
b
j
e
c
t
s
75
80
70
90
M
u
l
t
i
p
l
e
O
b
j
e
c
t
s
90
85
20
95
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
10
,
No
.
2
,
A
p
r
il 2
0
2
0
:
1
3
1
7
-
1326
1324
T
o
ev
alu
ate
th
e
v
is
u
al
p
er
f
o
r
m
an
ce
o
f
th
e
p
r
o
p
o
s
ed
alg
o
r
it
h
m
,
w
e
co
m
p
ar
ed
th
e
p
r
o
p
o
s
e
d
alg
o
r
ith
m
to
3
alg
o
r
ith
m
s
.
T
h
e
v
id
eo
s
ex
a
m
i
n
ed
co
n
tain
d
i
f
f
er
en
t
b
ac
k
g
r
o
u
n
d
s
ce
n
es,
a
n
d
m
u
ltip
le
m
o
v
i
n
g
o
b
j
ec
ts
b
o
th
o
u
td
o
o
r
s
an
d
in
d
o
o
r
s
(
p
ed
estr
ian
s
,
v
eh
ic
les,
etc.
)
.
W
e
h
av
e
ch
o
s
en
t
h
e
f
o
llo
w
i
n
g
m
o
s
t
m
eth
o
d
s
to
co
m
p
ar
e
w
it
h
o
u
r
p
r
o
p
o
s
ed
m
eth
o
d
:
(
1
)
T
h
e
B
ac
k
g
r
o
u
n
d
Su
b
tr
ac
tio
n
(
B
G
SUB
)
[
15
]
,
(
2
)
GM
M
m
et
h
o
d
[
8
]
,
an
d
(
3
)
Op
tical
Flo
w
(
O
F)
[
1
0
]
.
Vis
u
al
r
esu
lts
ar
e
s
h
o
w
n
o
n
th
e
v
id
eo
s
test
ed
in
Fig
u
r
e
6
.
I
n
d
iv
id
u
al
an
d
g
r
o
u
p
in
f
a
n
tr
y
,
s
m
al
l
d
y
n
a
m
ic
b
ac
k
g
r
o
u
n
d
,
a
n
d
m
u
l
tip
le
tr
af
f
ic
s
u
r
v
e
y
s
,
as
s
h
o
w
n
in
Fi
g
u
r
e
6
,
t
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
i
s
clo
s
es
t
to
Gr
o
u
n
d
T
r
u
th
(
GT
)
.
So
m
e
o
f
th
e
r
esu
lts
o
f
t
h
e
test
e
d
alg
o
r
ith
m
s
co
n
s
id
er
t
h
e
f
o
r
eg
r
o
u
n
d
o
b
j
ec
t
as
th
e
b
ac
k
g
r
o
u
n
d
.
T
h
e
m
ain
r
ea
s
o
n
is
t
h
at
t
h
e
p
ar
ts
o
f
t
h
e
o
b
j
ec
t
r
e
m
ai
n
s
tatic
in
t
h
e
v
id
eo
,
an
d
th
at
t
h
e
p
r
o
p
o
s
e
d
alg
o
r
ith
m
h
as o
v
er
co
m
e
t
h
i
s
ef
f
ec
t,
an
d
o
b
v
io
u
s
l
y
t
h
e
d
etec
t
io
n
ef
f
ec
t is b
etter
t
h
an
o
t
h
er
alg
o
r
ith
m
s
.
I
n
o
r
d
er
to
en
h
an
ce
t
h
e
r
o
b
u
s
t
n
es
s
an
d
ef
f
icie
n
c
y
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
te
m
,
a
s
ec
o
n
d
co
m
p
ar
is
o
n
w
as
m
ad
e
b
et
w
ee
n
Kal
m
a
n
Fil
ter
T
r
ac
k
er
an
d
Me
an
Sh
if
t
T
r
a
ck
er
[
2
4
]
as
s
h
o
w
n
in
Fig
u
r
e
7
.
T
h
e
Me
an
Sh
if
t
alg
o
r
ith
m
li
s
ts
t
h
e
n
o
n
-
p
ar
a
m
e
tr
ic
d
en
s
it
y
t
h
at
f
i
n
d
s
th
at
t
h
e
p
ictu
r
e
f
r
a
m
e
is
v
er
y
s
i
m
ilar
to
t
h
e
co
lo
r
h
is
to
g
r
a
m
o
f
th
e
o
b
j
ec
t
in
th
e
cu
r
r
en
t
f
r
a
m
e.
Me
an
-
S
h
i
f
t
tr
ac
k
in
g
f
r
eq
u
en
tl
y
in
cr
ea
s
e
s
ap
p
ea
r
an
ce
s
i
m
ilar
it
y
b
y
co
m
p
ar
i
n
g
th
e
o
b
j
ec
t
lay
o
u
t
a
n
d
th
e
w
i
n
d
o
w
la
y
o
u
t
ar
o
u
n
d
t
h
e
lo
ca
tio
n
o
f
th
e
f
ilter
o
b
j
ec
t.
T
h
is
alg
o
r
ith
m
h
a
s
th
e
a
b
ilit
y
to
tr
ac
k
in
r
ea
l
ti
m
e,
th
e
n
o
n
lin
ea
r
m
o
v
in
g
o
b
j
ec
ts
,
an
d
h
as
g
o
o
d
ap
p
licab
ilit
y
w
it
h
o
b
j
ec
t
d
is
to
r
tio
n
an
d
r
o
tatio
n
.
Ho
w
e
v
er
,
t
h
e
Me
a
n
Sh
i
f
t
al
g
o
r
it
h
m
d
o
es
n
o
t u
s
e
th
e
o
b
j
ec
t'
s
d
ir
ec
tio
n
a
n
d
s
p
ee
d
i
n
f
o
r
m
atio
n
to
tr
ac
k
an
o
b
j
ec
t
,
an
d
it
is
ea
s
y
to
lo
s
e
an
o
b
j
ec
t
w
h
en
th
er
e
is
in
ter
f
er
en
ce
(
s
u
c
h
as
lig
h
t
an
d
s
ca
tter
in
g
)
in
th
e
s
u
r
r
o
u
n
d
i
n
g
e
n
v
ir
o
n
m
en
t
an
d
ca
n
n
o
t
d
ea
l
w
it
h
s
ca
le
a
n
d
clu
tter
d
if
f
er
e
n
ce
s
[
2
5
]
.
T
o
ac
h
iev
e
a
g
o
o
d
tr
ac
e
ef
f
ec
t,
it
is
e
x
tr
e
m
el
y
i
m
p
o
r
tan
t
to
i
m
p
r
o
v
e
t
h
e
p
er
f
o
r
m
a
n
ce
o
f
t
h
e
tr
ac
e
alg
o
r
it
h
m
b
y
ch
o
o
s
i
n
g
a
d
i
s
ti
n
ct
f
ea
tu
r
e
to
cr
ea
te
th
e
o
b
j
ec
t
m
o
d
el,
s
o
th
at
t
h
e
c
h
ar
ac
ter
is
tics
o
f
t
h
e
o
b
j
ec
t
an
d
th
e
b
ac
k
g
r
o
u
n
d
ar
e
clea
r
l
y
d
if
f
er
en
t.
T
ab
le
3
s
h
o
w
s
s
o
m
e
ad
v
a
n
ta
g
es a
n
d
d
is
ad
v
a
n
tag
e
s
o
f
b
o
th
m
et
h
o
d
s
.
(
a)
(
b
)
(
c)
(
d
)
(
e)
(
f
)
Fig
u
r
e
6
.
T
h
e
co
m
p
ar
is
o
n
r
es
u
lts
f
o
r
s
e
v
er
al
ex
p
er
i
m
e
n
t c
ap
t
u
r
ed
v
id
eo
: (
a)
I
n
p
u
t Fr
a
m
e
(
I
/
P
)
,
(
b
)
T
h
e
Gr
o
u
n
d
T
r
u
th
(
GT
)
,
(
c
)
T
h
e
b
ac
k
g
r
o
u
n
d
Su
b
tr
ac
tio
n
(
B
G
SUB
)
[
1
5
]
(
d
)
GM
M
Me
th
o
d
[
8
]
,
(
e)
Op
tical
Flo
w
(
OF)
[
1
0
]
,
an
d
(
f
)
T
h
e
P
r
o
p
o
s
ed
A
lg
o
r
ith
m
(
FP
C
P
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
Ob
jects d
etec
tio
n
a
n
d
tr
a
ck
in
g
u
s
in
g
fa
s
t p
r
in
cip
le
co
m
p
o
n
en
t p
u
r
is
t a
n
d
ka
lma
n
filt
er
(
H
a
d
ee
l N.
A
b
d
u
lla
h
)
1325
(
a)
(
b
)
(
c)
(
d
)
Fig
u
r
e
7
.
T
h
e
co
m
p
ar
is
o
n
r
es
u
lts
f
o
r
s
e
v
er
al
tr
ac
k
i
n
g
tec
h
n
iq
u
es:
(
a)
Or
ig
in
al
Fra
m
e,
(
b
)
o
b
j
ec
t
Dete
ctio
n
,
(
c)
KF T
r
ac
k
in
g
,
a
n
d
(
d
)
Me
an
Sh
i
f
t T
r
ac
k
in
g
[
2
4
]
T
ab
le
3
: Co
m
p
ar
is
o
n
b
et
w
ee
n
Kal
m
a
n
Fil
ter
a
n
d
Me
an
S
h
i
f
t
T
r
ac
k
er
T
e
c
h
n
i
q
u
e
A
d
v
a
n
t
a
g
e
s
D
i
sad
v
a
n
t
a
g
e
s
K
a
l
man
F
i
l
t
e
r
1
.
Ea
sy
t
o
i
m
p
l
e
me
n
t
,
P
o
i
n
t
t
r
a
c
k
i
n
g
(
D
i
st
a
n
c
e
a
n
d
V
e
l
o
c
i
t
y
)
.
2
.
C
o
n
t
i
n
u
o
u
s
t
r
a
c
k
i
n
g
d
e
sp
i
t
e
p
a
u
se
s
i
n
mo
v
i
n
g
v
i
d
e
o
s.
3
-
I
t
c
a
n
t
r
a
c
k
m
u
l
t
i
p
l
e
o
b
j
e
c
t
s
.
1
.
T
h
e
c
o
m
p
u
t
a
t
i
o
n
a
l
l
y
e
x
p
e
n
si
v
e
.
2
.
I
t
i
s l
i
n
e
a
r
t
r
a
c
k
e
r
.
M
e
a
n
S
h
i
f
t
1
.
I
t
t
r
a
c
k
s
b
a
se
d
o
n
c
o
l
o
r
o
f
o
b
j
e
c
t
2
.
I
t
i
s n
o
n
l
i
n
e
a
r
t
r
a
c
k
e
r
1
.
I
n
e
f
f
e
c
t
i
v
e
w
h
e
n
t
h
e
r
e
a
r
e
o
c
c
l
u
si
o
n
o
r
d
i
st
o
r
t
i
o
n
p
r
o
b
l
e
ms.
2
A
si
n
g
l
e
o
b
j
e
c
t
t
r
a
c
k
e
r
.
6.
CO
NCLU
SI
O
N
T
h
e
alg
o
r
ith
m
p
r
o
v
id
ed
w
a
s
im
p
le
m
e
n
ted
in
Ma
tlab
.
Fra
m
es
ar
e
p
r
o
ce
s
s
ed
in
d
if
f
er
en
t
s
izes
u
s
i
n
g
m
s
i
t
y
p
e
I
n
tel
C
o
r
e
i7
8
7
5
0
H
co
m
p
u
ter
.
T
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
s
h
o
w
s
its
ad
v
a
n
tag
e
o
v
e
r
ex
is
ti
n
g
o
b
j
ec
t
m
et
h
o
d
s
m
o
r
e
p
r
ec
is
ely
f
r
o
m
th
e
r
esu
lt
s
o
b
tain
ed
.
T
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
w
as
co
m
p
ar
ed
w
it
h
ex
i
s
ti
n
g
v
id
eo
f
ile
alg
o
r
it
h
m
s
w
h
er
e
th
e
r
es
u
lts
s
h
o
w
th
e
e
f
f
icien
c
y
o
f
th
e
p
r
o
p
o
s
ed
w
o
r
k
.
T
h
e
p
r
o
p
o
s
ed
s
y
s
te
m
ca
n
p
r
o
ce
s
s
g
r
a
y
an
d
co
lo
r
v
id
eo
s
at
f
e
w
s
ec
o
n
d
s
p
er
GP
U
-
Ma
tlab
ap
p
licatio
n
f
r
a
m
e.
T
h
e
p
r
o
p
o
s
ed
al
g
o
r
ith
m
ca
n
ad
ap
t
to
b
ac
k
g
r
o
u
n
d
ch
a
n
g
e
s
.
Ob
j
ec
t
d
etec
tio
n
an
d
tr
ac
k
i
n
g
ar
e
m
ai
n
an
d
af
f
r
o
n
t
m
i
s
s
io
n
in
m
a
n
y
co
m
p
u
ter
v
is
ib
ilit
y
i
m
p
le
m
en
ta
tio
n
s
,
s
u
c
h
as
m
o
n
ito
r
in
g
,
ca
r
s
alt
w
o
r
k
s
,
r
o
u
ti
n
g
,
an
d
au
to
m
atio
n
.
T
h
is
al
g
o
r
ith
m
p
r
ese
n
t
s
s
ev
er
a
l
b
en
ef
it
s
,
s
u
ch
a
s
m
u
ltip
le
o
b
ject
d
etec
tio
n
an
d
tr
ac
k
i
n
g
i
n
d
if
f
er
e
n
t
en
v
ir
o
n
m
e
n
ts
.
T
h
e
d
is
ad
v
a
n
tag
e
s
o
f
t
h
i
s
tech
n
iq
u
e
u
s
i
n
g
o
n
e
m
et
h
o
d
w
il
l
n
o
t
p
r
o
d
u
ce
p
er
f
ec
t
r
esu
l
ts
b
ec
au
s
e
it
s
ac
cu
r
ac
y
is
i
n
f
l
u
en
ce
d
b
y
d
i
f
f
er
e
n
t
o
p
er
ato
r
s
s
u
ch
as
th
e
lo
w
r
eso
l
u
tio
n
o
f
ca
p
tu
r
ed
v
id
eo
,
ch
an
g
e
in
w
ea
t
h
er
.
E
tc.
I
n
th
e
f
u
tu
r
e,
w
e
h
o
p
e
to
ex
p
an
d
o
u
r
s
co
p
e
o
f
d
etec
tio
n
a
n
d
tr
a
ck
in
g
o
f
o
b
j
ec
ts
in
o
v
er
cr
o
w
d
ed
s
ce
n
er
y
o
r
t
h
e
ap
p
ea
r
an
ce
o
f
s
e
v
er
e
co
n
tr
ast
in
lig
h
ti
n
g
a
n
d
r
ea
l
-
ti
m
e
s
ce
n
es
.
RE
F
E
R
E
NC
E
S
[1
]
Ch
e
n
g
,
J.,
Ya
n
g
,
J.
,
Z
h
o
u
,
Y.
a
n
d
Cu
i,
Y
.
,
“
F
lex
ib
le
b
a
c
k
g
ro
u
n
d
m
ix
tu
re
m
o
d
e
ls
f
o
r
f
o
re
g
ro
u
n
d
se
g
m
e
n
tatio
n
,
”
Ima
g
e
a
n
d
Vi
si
o
n
C
o
mp
u
ti
n
g
,
24
(
5
):
4
7
3
-
4
8
2
,
2
0
0
6
.
[2
]
W
u
,
Y.,
L
im
,
J.
a
n
d
Ya
n
g
,
M
.
H.,
“
On
li
n
e
o
b
jec
t
trac
k
in
g
:
A
b
e
n
c
h
m
a
rk
,
”
In
Pro
c
e
e
d
in
g
s o
f
th
e
IEE
E
c
o
n
fer
e
n
c
e
o
n
c
o
mp
u
ter
v
isio
n
a
n
d
p
a
tt
e
rn
re
c
o
g
n
it
i
o
n
(p
p
.
2
4
1
1
-
2
4
1
8
)
.
2
0
1
3
.
[3
]
Na
e
e
m
,
H.,
A
h
m
a
d
,
J.
a
n
d
T
a
y
y
a
b
,
M
.
,
“
Re
a
l
-
ti
m
e
o
b
jec
t
d
e
tec
ti
o
n
a
n
d
trac
k
in
g
,”
In
INM
IC
(p
p
.
1
4
8
-
1
5
3
).
I
EE
E
,
De
c
e
m
b
e
r
2
0
1
3
.
[4
]
C
y
g
a
n
e
k
,
B.
,
“
Ob
jec
t
d
e
tec
ti
o
n
a
n
d
re
c
o
g
n
it
io
n
i
n
d
ig
it
a
l
im
a
g
e
s: t
h
e
o
ry
a
n
d
p
ra
c
ti
c
e
,”
J
o
h
n
W
il
e
y
&
S
o
n
s,
2
0
1
3
.
[5
]
T
a
n
g
,
S
.
L
.
,
Ka
d
i
m
,
Z.
,
L
ian
g
,
K.
M
.
a
n
d
L
im
,
M
.
K.,
“
H
y
b
rid
b
l
o
b
a
n
d
p
a
rti
c
le
f
il
ter
trac
k
in
g
a
p
p
r
o
a
c
h
f
o
r
ro
b
u
st
o
b
jec
t
trac
k
in
g
,
”
Pro
c
e
d
ia
Co
m
p
u
ter
S
c
i
e
n
c
e
,
1
(1
),
2
5
4
9
-
2
5
5
7
,
2
0
1
0
.
[6
]
Zh
a
n
g
,
H.Y.,
“
M
u
l
ti
p
le
m
o
v
in
g
o
b
jec
ts
d
e
tec
ti
o
n
a
n
d
trac
k
in
g
b
a
se
d
o
n
o
p
t
ica
l
f
lo
w
in
p
o
l
a
r
-
lo
g
ima
g
e
s,”
In
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
M
a
c
h
in
e
L
e
a
rn
i
n
g
a
n
d
Cy
b
e
rn
e
ti
c
s
(v
o
l.
3,
p
p
.
1
5
7
7
-
1
5
8
2
)
,
Ju
ly
2
0
1
0
.
[7
]
M
a
n
d
e
ll
o
s,
N.A
.
,
Ke
ra
m
it
so
g
lo
u
,
I.
a
n
d
Kira
n
o
u
d
is,
C.
T
.
,
“
A
b
a
c
k
g
ro
u
n
d
su
b
trac
ti
o
n
a
lg
o
rit
h
m
f
o
r
d
e
tec
ti
n
g
a
n
d
trac
k
in
g
v
e
h
icle
s,”
Exp
e
rt S
y
ste
m
s wit
h
A
p
p
l
ica
ti
o
n
s
,
38
(
3
):
1
6
1
9
-
1
6
3
1
,
2
0
1
1
.
[8
]
Ba
k
ti
,
R.
Y.,
A
re
n
i,
I.
S
.
a
n
d
P
ra
y
o
g
i,
A
.
A
.
,
“
V
e
h
icle
De
tec
ti
o
n
a
n
d
T
ra
c
k
in
g
u
sin
g
G
a
u
ss
ian
M
ix
tu
re
M
o
d
e
l
a
n
d
Ka
l
m
a
n
F
il
ter,”
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
C
o
mp
u
ta
ti
o
n
a
l
In
t
e
ll
ig
e
n
c
e
a
n
d
Cy
b
e
rn
e
ti
c
s
(
p
p
.
1
1
5
-
1
1
9
),
IE
EE
,
No
v
e
m
b
e
r
2
0
1
6
.
[9
]
Ka
le,
K.,
P
a
w
a
r,
S
.
a
n
d
D
h
u
lek
a
r,
P
.
,
“
M
o
v
in
g
o
b
jec
t
trac
k
in
g
u
sin
g
o
p
ti
c
a
l
f
lo
w
a
n
d
m
o
ti
o
n
v
e
c
to
r
e
stim
a
ti
o
n
,
”
In
4
t
h
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
Relia
b
il
it
y
,
In
f
o
c
o
m
T
e
c
h
n
o
l
o
g
i
e
s
a
n
d
O
p
ti
miza
t
io
n
(
ICRIT
O)(
T
r
e
n
d
s
a
n
d
F
u
t
u
re
Dire
c
ti
o
n
s)
(p
p
.
1
-
6
),
IEE
E
,
S
e
p
t
e
m
b
e
r
2
0
1
5
.
[1
0
]
A
sla
n
i,
S
.
a
n
d
M
a
h
d
a
v
i
-
Na
sa
b
,
H.,
“
Op
ti
c
a
l
f
lo
w
b
a
se
d
m
o
v
in
g
o
b
jec
t
d
e
tec
ti
o
n
a
n
d
trac
k
in
g
f
o
r
traff
ic
su
rv
e
il
lan
c
e
.
”
In
ter
n
a
t
io
n
a
l
J
o
u
rn
a
l
o
f
El
e
c
trica
l,
El
e
c
tro
n
ics
,
Co
mm
u
n
ica
ti
o
n
,
En
e
rg
y
S
c
ien
c
e
a
n
d
En
g
i
n
e
e
rin
g
,
7
(
9
):
7
8
9
-
7
9
3
,
2
0
1
3
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
10
,
No
.
2
,
A
p
r
il 2
0
2
0
:
1
3
1
7
-
1326
1326
[1
1
]
T
e
la
g
a
ra
p
u
,
P
.
,
Ra
o
,
M
.
N.
a
n
d
S
u
re
sh
,
G
.
,
“
A
n
o
v
e
l
traf
f
ic
-
tr
a
c
k
in
g
sy
ste
m
u
sin
g
m
o
rp
h
o
lo
g
ica
l
a
n
d
Blo
b
a
n
a
ly
sis,”
In
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
C
o
mp
u
t
in
g
,
C
o
mm
u
n
ic
a
ti
o
n
a
n
d
A
p
p
li
c
a
ti
o
n
s
(p
p
.
1
-
4
)
.
IE
EE
,
F
e
b
r
u
a
ry
2
0
1
2
.
[1
2
]
Hu
,
W
.
C.
,
C
h
e
n
,
C
.
H.,
C
h
e
n
,
T
.
Y.,
Hu
a
n
g
,
D.Y.
a
n
d
W
u
,
Z.
C.
,
“
M
o
v
in
g
o
b
jec
t
d
e
tec
ti
o
n
a
n
d
trac
k
in
g
f
ro
m
v
id
e
o
c
a
p
tu
re
d
b
y
m
o
v
in
g
c
a
m
e
ra
,
”
J
o
u
rn
a
l
o
f
Vi
s
u
a
l
Co
mm
u
n
ica
ti
o
n
a
n
d
Ima
g
e
Rep
re
se
n
ta
ti
o
n
,
3
0
:
1
6
4
-
1
8
0
,
2
0
1
5
.
[1
3
]
Ra
y
,
K.S
.
a
n
d
C
h
a
k
ra
b
o
rty
,
S
.
,
“
Ob
jec
t
d
e
tec
ti
o
n
b
y
sp
a
ti
o
-
tem
p
o
ra
l
a
n
a
ly
sis
a
n
d
trac
k
in
g
o
f
th
e
d
e
tec
ted
o
b
jec
ts
in
a
v
id
e
o
w
it
h
v
a
riab
le
b
a
c
k
g
ro
u
n
d
,
”
J
o
u
rn
a
l
o
f
Vi
su
a
l
Co
mm
u
n
ica
ti
o
n
a
n
d
Ima
g
e
Rep
re
se
n
ta
t
io
n
,
58:
6
6
2
-
6
7
4
.
2
0
1
9
.
[1
4
]
M
a
h
a
li
n
g
a
m
,
T
.
a
n
d
S
u
b
ra
m
o
n
i
a
m
,
M
.
,
“
A
ro
b
u
st
si
n
g
le
a
n
d
m
u
lt
ip
le
m
o
v
in
g
o
b
jec
t
d
e
tec
ti
o
n
,
trac
k
in
g
a
n
d
c
las
si
f
ica
ti
o
n
,
”
Ap
p
li
e
d
C
o
mp
u
ti
n
g
a
n
d
In
f
o
rm
a
ti
c
s,
2
0
1
8
.
[1
5
]
Ca
o
,
W
.
,
W
a
n
g
,
Y.,
S
u
n
,
J.
,
M
e
n
g
,
D.,
Ya
n
g
,
C.
,
Cich
o
c
k
i,
A
.
a
n
d
X
u
,
Z.
,
“
T
o
tal
v
a
riatio
n
re
g
u
lariz
e
d
ten
so
r
R
P
CA
f
o
r
b
a
c
k
g
ro
u
n
d
su
b
trac
ti
o
n
f
ro
m
c
o
m
p
re
ss
iv
e
m
e
a
su
re
m
e
n
ts,”
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Im
a
g
e
Pro
c
e
ss
in
g
,
25
(
9
):
4
0
7
5
-
4
0
9
0
,
2
0
1
6
.
[1
6
]
S
u
p
re
e
th
,
H.
S
.
G
.
,
a
n
d
Ch
a
n
d
r
a
sh
e
k
a
r
M
.
P
a
ti
l
.
"
Ef
f
icie
n
t
m
u
lt
ip
le
m
o
v
in
g
o
b
jec
t
d
e
tec
ti
o
n
a
n
d
trac
k
in
g
u
sin
g
c
o
m
b
in
e
d
b
a
c
k
g
r
o
u
n
d
su
b
trac
ti
o
n
a
n
d
c
l
u
ste
rin
g
,
"
S
ig
n
a
l,
Ima
g
e
a
n
d
Vi
d
e
o
Pr
o
c
e
ss
in
g
,
1
2
(
6
)
:
1
0
9
7
-
1
1
0
5
,
2
0
1
8
.
[1
7
]
Bo
u
wm
a
n
s,
T
.
a
n
d
Zah
z
a
h
,
E.
H.,
“
Ro
b
u
st
P
CA
v
ia
p
rin
c
ip
a
l
c
o
m
p
o
n
e
n
t
p
u
rsu
it
:
A
re
v
i
e
w
f
o
r
a
c
o
m
p
a
ra
ti
v
e
e
v
a
lu
a
ti
o
n
i
n
v
id
e
o
s
u
rv
e
il
lan
c
e
,
”
Co
mp
u
ter
Vi
sio
n
a
n
d
Ima
g
e
Un
d
e
rs
ta
n
d
i
n
g
,
1
2
2
:
2
2
-
3
4
,
2
0
1
4
.
[1
8
]
Ro
d
rig
u
e
z
,
P
.
a
n
d
W
o
h
l
b
e
rg
,
B.
,
“
F
a
st
p
ri
n
c
ip
a
l
c
o
m
p
o
n
e
n
t
p
u
rsu
i
t
v
ia
a
lt
e
r
n
a
ti
n
g
m
in
im
iz
a
ti
o
n
,
”
In
IEE
E
In
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
Im
a
g
e
Pro
c
e
ss
in
g
c
(
p
p
.
6
9
-
7
3
),
IE
EE
,
S
e
p
tem
b
e
r
2
0
1
3
.
[1
9
]
Zh
a
o
,
Z.
Q
.
,
Zh
e
n
g
,
P
.
,
X
u
,
S
.
T
.
a
n
d
W
u
,
X.,
“
Ob
jec
t
d
e
tec
ti
o
n
w
it
h
d
e
e
p
lea
rn
in
g
:
A
re
v
ie
w
,
”
IEE
E
tra
n
sa
c
ti
o
n
s
o
n
n
e
u
ra
l
n
e
two
rk
s a
n
d
lea
rn
i
n
g
sy
ste
ms
,
2
0
1
9
.
[2
0
]
G
irsh
ick
,
R.
,
"
F
a
st r
-
c
n
n
.
"
In
Pro
c
e
e
d
in
g
s o
f
th
e
IEE
E
i
n
ter
n
a
ti
o
n
a
l
c
o
n
fer
e
n
c
e
o
n
c
o
mp
u
ter
v
isio
n
(
p
p
.
1
4
4
0
-
1
4
4
8
)
.
2
0
1
5
.
[2
1
]
L
i,
Q.,
L
i,
R.
,
Ji,
K.
a
n
d
Da
i,
W
.
,
“
Ka
l
m
a
n
f
il
ter
a
n
d
it
s
a
p
p
li
c
a
ti
o
n
,
”
In
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
In
telli
g
e
n
t
Ne
two
rk
s a
n
d
In
tell
ig
e
n
t
S
y
ste
ms
(
ICINIS
)
(p
p
.
7
4
-
7
7
)
,
IEE
E
,
No
v
e
m
b
e
r
2
0
1
5
.
[2
2
]
S
a
h
o
,
K.,
“
Ka
lm
a
n
F
il
ter
f
o
r
M
o
v
in
g
Ob
jec
t
T
ra
c
k
in
g
:
P
e
rf
o
rm
a
n
c
e
A
n
a
l
y
sis
a
n
d
F
il
ter
De
sig
n
,
”
In
Ka
lma
n
Fi
lt
e
rs
-
T
h
e
o
ry
fo
r
Ad
v
a
n
c
e
d
A
p
p
l
ica
ti
o
n
s
.
In
tec
h
Op
e
n
,
2
0
1
7
.
[2
3
]
L
iu
,
X
.
,
W
e
n
,
Z.
a
n
d
Z
h
a
n
g
,
Y.,
"
L
i
m
it
e
d
m
e
m
o
r
y
b
lo
c
k
Kr
y
lo
v
su
b
sp
a
c
e
o
p
t
im
iza
ti
o
n
f
o
r
c
o
m
p
u
ti
n
g
d
o
m
in
a
n
t
sin
g
u
lar v
a
lu
e
d
e
c
o
m
p
o
siti
o
n
s,"
S
IAM
J
o
u
rn
a
l
o
n
S
c
ien
ti
fi
c
C
o
mp
u
ti
n
g
,
3
5
(3
)
:
A
1
6
4
1
-
A
1
6
6
8
,
2
0
1
3
.
[2
4
]
S
h
iv
h
a
re
,
A
.
a
n
d
Ch
o
u
d
h
a
r
y
,
V
.
,
“
Ob
jec
t
trac
k
in
g
in
v
id
e
o
u
sin
g
m
e
a
n
sh
if
t
a
l
g
o
rit
h
m
:
A
re
v
i
e
w
,”
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Co
m
p
u
ter
S
c
ien
c
e
a
n
d
In
fo
rm
a
ti
o
n
T
e
c
h
n
o
l
o
g
ies
,
2
0
1
5
.
[2
5
]
Ka
ra
su
lu
,
B.
a
n
d
Ko
r
u
k
o
g
lu
,
S
.
,
“
P
e
rf
o
rm
a
n
c
e
Ev
a
lu
a
ti
o
n
S
o
f
twa
re
:
M
o
v
in
g
Ob
jec
t
De
tec
ti
o
n
a
n
d
T
ra
c
k
in
g
in
V
id
e
o
s,
”
S
p
rin
g
e
r S
c
ien
c
e
&
Bu
sin
e
ss
M
e
d
ia
,
2
0
1
3
.
Evaluation Warning : The document was created with Spire.PDF for Python.