T
E
L
KO
M
NI
K
A
,
V
ol
.
16,
N
o.
2,
A
pr
i
l
20
18,
pp.
629
~
6
34
I
S
S
N
:
1
693
-
6
930
,
ac
c
r
edi
t
ed A
b
y
D
I
K
T
I
,
D
ec
r
ee N
o:
58/
D
I
K
T
I
/
K
ep/
201
3
D
O
I
:
10.
129
28/
T
E
L
K
O
MN
I
K
A
.
v
1
6i
2.
45
09
6
29
R
ec
ei
v
ed
O
c
t
ober
18
,
201
7
;
R
ev
i
s
ed
J
anu
ar
y
21,
201
8
;
A
c
c
ept
e
d
F
e
br
uar
y
10
,
20
1
8
S
t
r
eet
M
ar
k
Det
ect
io
n
U
s
i
ng
R
a
s
pbe
r
r
y
P
I
f
or
S
elf
-
d
r
i
vi
n
g
S
ys
t
e
m
Su
m
a
rd
i
*
,
M
u
h
a
m
m
a
d
T
a
u
fi
q
u
r
r
a
h
m
a
n
,
M
u
n
aw
ar
A
R
i
ya
d
i
D
epar
t
m
ent
of
E
l
e
c
t
r
i
c
a
l
E
ng
i
n
eer
i
ng
D
i
po
neg
or
o U
ni
v
er
s
i
t
y
J
l
n.
P
r
of
.
S
u
dhar
t
o,
T
em
b
al
a
n
g,
S
em
ar
an
g,
I
n
done
s
i
a
*
C
or
r
es
po
ndi
ng a
ut
hor
,
e
-
ma
i
l
:
s
um
ar
di
.
und
i
p@
g
m
a
il
.
c
o
m
A
b
st
r
act
S
el
f
dr
i
v
i
ng
i
s
an
au
t
onom
o
us
v
eh
i
c
l
e
t
hat
c
a
n
f
ol
l
ow
t
h
e r
o
ad w
i
t
h
le
s
s
h
um
an i
nt
er
v
ent
i
on.
Th
e
dev
e
l
opm
en
t
of
s
el
f
dr
i
v
i
ng
u
t
i
l
i
z
es
v
ar
i
o
us
m
et
h
od
s
s
u
c
h
a
s
r
adar
,
l
i
dar
,
G
P
S
,
c
am
er
a,
or
c
om
bi
n
at
i
on
o
f
t
hem
.
I
n t
hi
s
r
es
ear
c
h
,
s
t
r
eet
m
ar
k
det
e
c
t
i
o
n s
y
s
t
em
w
as
d
es
i
g
ned
us
i
ng
w
ebc
am
and
r
as
pb
er
r
y
-
pi
m
i
n
i
c
om
put
er
f
or
pr
oc
e
s
s
i
ng t
he
i
m
age
.
T
he
i
m
ag
e
wa
s
pr
o
c
es
s
ed
b
y
HS
V
c
ol
or
f
i
l
t
er
i
n
g
m
et
hod
.
T
h
e
pr
oc
e
s
s
i
ng
r
at
e
of
t
h
i
s
al
gor
i
t
h
m
w
as
1
37.
9
8 m
s
c
or
r
es
pond
i
ni
g
t
o
7
.
2
F
PS.
T
he
s
e
l
f
-
d
r
iv
ing
p
r
o
t
ot
y
pe
wa
s
f
ound
t
o
be
w
or
k
i
ng
opt
i
m
al
l
y
f
or
“
hu
e”
t
hr
es
hol
d of
0
-
1
7
9,
“
s
at
ur
a
t
i
on”
t
hr
es
h
ol
d
of
0
-
30,
a
nd “
v
a
l
ue”
t
hr
es
hol
d
of
200
-
255.
S
t
r
eet
m
ar
k
det
ec
t
i
on
ha
s
be
en
o
bt
a
i
ned
f
r
om
t
he
c
oor
di
n
at
e
s
of
s
t
r
eet
m
ar
k
ob
j
ec
t
w
hi
c
h
ha
d r
a
nge
4
-
16
7
o
n
x
ax
i
s
a
nd
4
-
1
39
on
y
ax
is
.
As
a
r
es
u
l
t
, w
e
ha
v
e
s
u
c
c
e
s
s
f
ul
l
y
bui
l
t
t
he
s
t
r
e
et
m
ar
k
det
e
c
t
i
o
n
by
C
O
G
m
et
hod
m
or
e
e
ffe
c
ti
v
e
ly
an
d
s
m
oot
hl
y
i
n
d
et
ec
t
i
o
n
i
n
c
om
pa
r
i
s
on
wi
t
h
H
ough
t
r
ans
f
or
m
m
et
hod
.
Ke
y
w
o
rd
s
:
St
re
e
t
m
a
rk
det
e
c
t
i
on,
H
S
V
c
ol
or
f
i
l
t
er
i
ng,
C
O
G
,
Se
l
f
-
d
r
iv
in
g c
ar
.
C
o
p
yr
i
g
h
t
©
201
8 U
n
i
v
er
s
i
t
as A
h
m
ad
D
ah
l
an
.
A
l
l
r
i
g
h
t
s r
es
er
v
ed
1.
I
n
tr
o
d
u
c
ti
o
n
S
e
lf
-
dr
i
v
i
ng i
s
o
ne t
y
pe
of
c
ar
c
ont
r
ol
t
hat
e
nab
l
e
s
t
o
dr
i
v
e
w
i
t
h l
es
s
hum
an i
nt
er
v
e
nt
i
o
n.
S
el
f
dr
i
v
i
ng
i
s
he
l
pf
ul
,
f
or
ex
am
pl
e
w
h
en t
he
dr
i
v
er
s
uf
f
er
s
c
er
t
ai
n c
o
nd
i
t
i
ons
t
hat
ne
ed
t
o
r
el
i
n
qu
i
s
h t
h
e s
t
eer
,
f
or
i
n
j
ur
i
es
or
f
ai
nt
i
ng
et
c
.
T
he dev
el
opm
ent
of
s
el
f
-
dr
i
v
i
ng r
es
ear
c
h
is
pr
edi
c
t
e
d
t
o
be
c
om
pl
et
e
and
r
ea
d
y
t
o
b
e
i
m
pl
em
ent
e
d
i
n
202
3
[
1]
.
D
e
v
e
l
opm
ent
of
t
he
r
es
ear
c
h on aut
onom
ous
dr
i
v
i
n
g c
ar
has
been d
on
e i
n m
a
n
y
w
a
ys
.
T
he
r
es
ear
c
h
es
in
s
e
lf
d
r
iv
in
g
w
er
e
c
on
duc
t
e
d
i
n
v
ar
i
ous
f
oc
us
e
i
t
her
in
ac
t
ual
r
ea
l
i
z
at
i
o
n
[2
]
or
pr
ot
ot
y
p
e
s
c
al
e
d
[
3
]
.
S
ev
e
r
al
m
et
hod
s
f
or
c
ol
l
i
s
i
on
av
oi
danc
e
ap
pl
i
ed
l
i
da
r
or
c
am
er
a [
2]
,
[
4]
,
w
hi
l
e i
t
c
ou
l
d al
s
o
i
nc
l
u
ded
m
i
ni
c
om
put
er
l
i
k
e
r
as
pber
r
y
pi
w
hi
c
h
c
an
pr
oc
es
s
t
he
i
m
age
[
4
]
.
M
or
e
ov
er
,
t
her
e
ar
e
v
ar
i
ous
i
m
age
pr
oc
es
s
i
ng
f
or
pat
t
er
n
r
ec
og
ni
t
i
on
m
et
hod
l
i
k
e
i
n
f
i
l
t
er
i
n
g
i
m
age,
h
ough
m
et
hod,
and det
er
m
i
n
in
g
t
h
e t
r
ac
k
i
ng r
ef
er
enc
es
[5
-
7]
.
I
n t
er
m
s
of
d
r
iv
in
g
gu
i
da
nc
e,
s
e
v
er
a
l
m
et
hods
has
been
de
v
e
l
op
ed,
i
nc
l
ud
i
ng
t
he us
e
of
m
ar
k
ed
and
unm
ar
k
ed l
anes
[
8
-
1
1
].
T
he us
age of
s
t
r
eet
m
ar
k
f
or
gui
danc
e
i
s
po
pu
l
ar
,
e.
g i
n
[
10]
,
[
12
].
T
he
r
oad m
ar
k
det
ec
t
i
on
i
n
[
12]
w
as
us
i
ng
a
c
am
er
a
pr
oc
es
s
i
ng
i
m
ages
.
T
he
y
us
ed
I
nt
el
pr
oc
es
s
or
T
5750
2.
0
G
H
z
c
l
oc
k
s
peed
t
h
at
has
pr
oc
es
s
i
ng
t
i
m
e
be
l
o
w
14
m
s
f
or
s
i
ngl
e
pr
oc
es
s
i
ng
.
H
o
w
e
ve
r
,
t
h
e
y
f
ac
ed obs
t
ac
l
es
i
n
det
er
m
i
ni
n
g t
h
e bo
und
ar
y
l
i
ne
i
n
unf
av
or
a
bl
e t
ur
n c
o
nd
i
t
i
on
s
.
A
l
t
hou
gh
it
c
oul
d
be
o
v
er
c
om
e b
y
ap
pl
y
i
n
g
ei
t
her
a s
p
l
i
n
e or
s
et
o
f
l
i
ne
t
o
ap
pr
ox
i
m
at
e l
ane
b
or
der
,
b
ut
t
he
t
ur
ni
n
g r
oad
w
as
har
d t
o
det
ec
t
b
ec
aus
e t
h
e
y
us
ed
H
ough
t
r
ans
f
or
m
.
T
he us
age of
a
m
in
i
c
o
m
put
er
f
or
i
m
age
pr
oc
e
s
s
i
ng i
s
pr
ef
er
r
ed f
or
ef
f
ec
t
i
v
e
d
i
m
en
s
i
on,
c
os
t
,
an
d
per
f
or
m
anc
e
ac
c
or
di
ng
t
o
U
j
ai
ni
y
a e
t
a
l
[4
].
T
o t
hat
ex
t
end
,
w
e
c
ond
uc
t
r
es
ear
c
h o
n s
el
f
-
dr
i
v
i
ng
w
h
i
c
h c
an det
ec
t
l
a
ne of
t
he r
oad
or
t
he
s
t
r
eet
m
ar
k
.
W
e
pr
opos
e a
des
i
gn
of
1:
10 s
c
al
e
pr
ot
ot
y
p
e of
s
e
lf
-
dr
i
v
i
ng c
ar
w
i
t
h i
m
age
pr
oc
es
s
i
ng f
r
om
c
a
m
e
r
a i
n
r
as
pber
r
y
p
i
2
.
T
he pur
po
s
e f
r
o
m
t
hi
s
r
es
ear
c
h i
s
t
o
dep
l
o
y
s
t
r
eet
m
ar
k
det
ec
t
i
on
m
et
h
od
f
or
s
e
lf
-
dr
i
v
i
ng
s
i
s
t
em
i
n
t
he
pr
ot
ot
y
pe
.
It
det
er
m
i
n
e
s
t
he
c
oor
di
n
at
e
b
y
C
O
G
(
C
ent
er
of
G
r
af
i
t
y
)
m
et
hod of
det
ec
t
i
o
n ar
e
a
i
n f
i
l
t
er
ed
i
m
age r
es
ul
t
ac
q
ui
r
ed
b
y
s
i
n
gl
e
c
a
m
er
a
as
wa
s
us
e
d
i
n
[
12
]
.
T
hi
s
pap
er
do
es
not
pa
y
at
t
en
t
i
o
n
t
o
t
h
e
i
l
l
um
i
nat
i
o
n
ef
f
e
c
t
[7
]
,
but
onl
y
s
pec
i
f
y
t
he
f
i
l
t
er
par
am
et
er
s
w
i
t
h H
S
V
m
et
hod
t
o d
et
ec
t
s
t
r
eet
m
ar
k
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1
6
9
3
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
16
,
N
o
.
2
,
A
p
r
i
l
20
18
:
6
29
–
6
34
630
2.
R
e
sea
r
ch
M
et
h
o
d
T
he
s
t
r
eet
m
ar
k
det
ec
t
i
o
n s
y
s
t
em
des
i
gn
is
b
as
ed o
n t
he s
pec
i
f
i
c
at
i
on
f
or
s
el
f
-
d
r
iv
in
g
s
ys
t
e
m
.
T
he s
el
f
-
dr
i
v
i
n
g
s
y
s
t
em
r
equi
r
e
s
s
t
r
eet
m
ar
k
det
ec
t
i
on s
y
s
t
em
t
hat
us
es
a
v
i
s
i
on s
ens
or
f
or
aut
onom
ous
c
ar
t
r
ac
k
i
ng
s
y
s
t
em
.
S
pec
i
f
i
c
at
i
ons
t
o
be
ac
h
i
e
v
ed
b
y
pr
ot
ot
y
p
e
of
s
el
f
dr
i
v
i
n
g
s
y
s
t
em
,
na
m
el
y
:
a.
us
i
ng s
m
al
l
di
m
ens
i
on h
a
r
dw
ar
e
s
as
a pr
ot
ot
y
p
e
w
i
t
h t
h
e
1:
1
0 s
c
al
e c
ar
a
nd t
he
m
a
x
i
m
u
m
w
ei
ght
of
3
k
g.
b.
abl
e
t
o
s
ee
and c
apt
ur
e
t
h
e
i
m
age
f
r
o
m
t
he c
a
m
er
a
i
n r
eal
t
i
m
e
and
c
an
be r
ep
eat
e
d
c
ont
i
n
uous
l
y
.
c.
abl
e t
o
det
ec
t
t
he
s
t
r
eet
m
ar
k
as
w
h
i
t
e c
o
l
or
ar
oun
d t
h
e bl
ac
k
t
r
ac
k
.
d.
abl
e t
o
det
ec
t
c
oor
d
i
nat
es
of
s
t
r
eet
m
ar
k
f
r
o
m
t
he f
i
l
t
er
ed i
m
ages
.
T
h
is
pr
oj
ec
t
is
bui
l
t
u
s
in
g
m
i
ni
c
o
m
put
er
r
as
pber
r
y
p
i
2 t
hat
has
c
om
pac
t
s
i
ze
,
l
i
ght
w
ei
g
ht
a
nd
a
1
G
B R
AM
,
qu
ad
-
c
or
e pr
oc
es
s
or
.
T
hi
s
c
o
m
put
er
is
c
apab
l
e t
o
pr
oc
e
s
s
t
he
i
m
age
to
det
ec
t
t
he
s
t
r
eet
m
ar
k
.
T
o
det
ec
t
s
t
r
eet
m
ar
k
,
r
as
pb
er
r
y
p
i
2 as
m
i
ni
c
om
put
er
m
us
t
hav
e a
par
t
i
c
u
l
ar
m
o
del
of
i
m
age p
r
oc
es
s
i
ng t
h
at
c
an
per
f
or
m
c
ol
or
f
i
l
t
er
i
ng.
T
hi
s
pr
oj
ec
t
us
e
s
O
penC
V
as
an
i
m
age pr
oc
es
s
i
ng
l
i
br
ar
y
t
o r
u
n H
S
V
c
ol
or
f
i
l
t
er
i
ng
m
et
hod
.
T
he p
r
ot
ot
y
pe
of
de
s
i
gn
ed
aut
o
nom
ous
c
ar
i
s
s
ho
w
n
i
n F
i
g
ur
e 1.
F
i
gur
e 1.
A
ut
o
nom
ous
c
ar
pr
ot
ot
y
pe
R
a
s
pber
r
y
p
i
s
ho
ul
d
b
e
c
a
pab
l
e
t
o
t
ak
e
pi
c
t
ur
es
f
r
om
c
am
er
a,
r
un
t
h
e
l
i
br
ar
y
O
penC
V
f
or
c
ol
or
f
i
l
t
er
i
ng pr
oc
es
s
,
per
f
or
m
c
oor
di
nat
e c
al
c
ul
a
t
i
on
and s
en
d dat
a t
hr
o
ug
h s
er
i
al
.
T
h
es
e
t
as
k
s
ar
e
done
a
ut
om
at
i
c
al
l
y
an
d
r
e
pea
t
ed
l
y
us
i
ng
p
y
t
hon
pr
ogr
am
m
i
ng.
O
p
enC
V
i
s
an
i
m
age
pr
oc
es
s
i
ng
l
i
br
ar
y
t
hat
c
a
n per
f
or
m
c
ol
or
f
i
l
t
er
i
ng
p
r
oc
es
s
.
T
he
m
et
hod us
e
d
i
s
H
S
V
c
o
l
or
f
ilt
e
r
in
g
b
as
ed o
n
R
G
B
c
ol
o
r
s
pac
e
,
wh
i
c
h
is
s
im
p
le
r
t
h
an C
M
Y
K
.
H
S
V
c
ol
or
f
i
l
t
er
i
ng m
et
hod h
as
a
c
ol
or
s
pac
e
t
hat
i
s
m
apped
w
i
t
h
3
c
om
ponent
s
:
H
u
e,
s
at
ur
at
i
on
,
an
d
v
al
u
e.
T
he
t
hr
es
h
ol
d
is
det
er
m
i
ne
d
o
n
t
h
e
c
ol
or
s
p
ac
e
t
o
det
ec
t
t
he
s
t
r
e
et
m
ar
k
o
f
t
he s
ur
r
ound
i
n
g en
v
i
r
o
nm
ent
.
S
t
r
eet
m
ar
k
det
ec
t
i
o
n s
y
s
t
e
m
w
as
de
v
e
l
op
ed
us
i
n
g
p
y
t
hon
pr
ogr
am
m
i
ng l
ang
uag
e on
a
m
in
i
c
o
m
put
er
r
as
pber
r
y
pi
.
T
he
al
gor
i
t
hm
and f
l
ow
c
h
a
r
t
f
or
det
ec
t
i
on s
y
s
t
e
m
i
s
s
how
n
in
F
i
gur
e 2
,
c
ons
i
s
t
i
n
g of
t
he f
ol
l
o
w
i
ng t
as
k
s
:
a.
T
a
k
e pi
c
t
ur
es
b.
S
ep
ar
at
e c
om
pone
nt
s
of
H
S
V
c.
T
hr
es
hol
di
n
g
d.
T
ot
al
t
he r
es
ul
t
s
t
hr
es
ho
l
di
ng H
S
V
c
o
l
or
s
p
ac
e
e.
obj
ec
t
det
ec
t
i
o
n
f.
C
al
c
u
l
at
e
t
he
c
oor
d
i
nat
es
o
f
t
he obj
ec
t
g.
S
en
d t
h
e da
t
a
v
i
a t
h
e s
er
i
a
l
F
i
gur
e
3 s
ho
w
s
f
i
v
e
w
i
ndo
w
s
f
or
H
S
V
c
o
l
or
f
i
l
t
er
i
n
g i
n s
t
r
eet
m
ar
k
det
ec
t
i
o
n t
h
at
ar
e
H
ue
F
ilt
e
r
as
s
ho
w
n i
n
F
i
gur
e 3.
a
,
Sa
t
ur
at
i
on
F
i
lt
e
r
as
s
ho
w
n i
n F
i
gur
e 3
.
b
,
Va
l
ue
F
ilt
e
r
as
s
ho
w
n
F
i
gur
e 3.
c
w
hi
c
h
f
eat
ur
e
t
r
ac
k
bar
t
o s
et
v
ar
i
ab
l
e t
hr
es
hol
d m
i
ni
m
u
m
and m
ax
i
m
u
m
of
eac
h
c
o
m
ponent
f
i
l
t
er
i
ng
.
T
he
w
i
nd
o
w
s
a
l
s
o s
ho
w
t
he r
es
ul
t
s
of
e
ac
h f
i
l
t
er
f
or
hu
e
,
s
at
ur
at
i
on
,
or
v
a
l
ue.
W
i
ndow
in
F
i
g
ur
e 3.
d
s
h
o
w
s
r
es
ul
t
s
f
r
o
m
t
he am
ount
of
t
h
e t
hr
ee
pr
oc
es
s
es
hue
,
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
K
A
I
S
S
N
:
1
693
-
6
930
S
t
r
eet
M
ar
k
D
e
t
ec
t
i
on U
s
i
n
g R
as
pb
er
r
y
P
I
F
or
S
e
l
f
-
dr
i
v
i
ng
S
y
s
t
em (
S
um
ar
di
)
631
s
at
ur
at
i
on,
and
v
a
l
u
e
t
o
t
h
e
r
es
ul
t
s
of
t
he
f
i
nal
f
i
l
t
er
t
hat
d
et
er
m
i
nes
t
he
H
S
V
ob
j
ec
t
i
s
det
ec
t
e
d
or
not
.
O
nc
e
t
he
obj
ec
t
c
an
be
det
ec
t
ed
f
r
o
m
H
S
V
f
i
l
t
er
,
t
hen
it
per
f
or
m
s
t
he
c
al
c
ul
at
i
o
n
of
t
he
c
oor
di
n
at
es
of
t
he
det
ec
t
e
d obj
ec
t
.
C
o
or
di
nat
e
d
at
a
t
hen
w
il
l
b
e
s
ent
t
o
t
he
an
ot
her
c
o
nt
r
o
l
l
er
(
m
i
c
r
oc
on
t
r
o
lle
r
)
v
i
a s
er
i
a
l
c
o
m
m
uni
c
at
i
on
.
F
i
g
ur
e 3.
e
s
ho
w
s
t
he r
es
u
l
t
of
s
t
r
eet
m
ar
k
t
r
a
ck
ed
f
r
om
t
he
or
i
gi
na
l
i
m
age
w
h
i
c
h
w
as
t
ak
en
b
y
t
h
e
c
am
er
a
w
i
t
h
t
h
e
a
ddi
t
i
o
n
of
a
s
q
uar
e
s
ha
pe
at
t
he or
i
gi
n of
t
h
e obj
ec
t
bei
n
g det
ec
t
ed
i
n c
a
l
c
ul
at
e
d c
o
or
di
n
at
es
.
T
he c
ar
pr
ot
ot
y
p
e
w
as
d
es
i
gned
w
i
t
h c
a
m
e
r
a
w
hi
c
h h
av
e
w
i
ndo
w
v
i
e
w
o
n
y
e
l
l
o
w
ar
ea
as
in
F
i
gur
e 4
.
T
he pr
ot
ot
y
p
e
has
bl
ank
s
pot
i
n f
r
ont
of
i
t
,
up
t
o 2
3
.
3 c
m
and has
40 de
gr
es
of
hor
i
z
ont
al
v
i
s
i
bi
l
i
t
y
.
A
c
c
or
di
ng t
o [
11
] [
12
]
,
t
he
c
apt
ur
e
d i
m
age
i
s
o
bt
a
i
ne
d f
r
om
l
i
ght
r
ef
l
ec
t
i
o
n t
o
c
a
m
er
a
l
ens
t
hat
i
s
pr
opor
t
i
on
al
w
i
t
h
t
he
d
i
s
t
anc
e.
T
her
ef
or
e,
t
he
poi
nt
c
oor
di
nat
es
i
s
det
ec
t
e
d
b
y
o
bt
a
i
n
i
ng
t
he
c
e
nt
er
of
det
ec
t
i
on
ar
ea
w
i
t
h
ge
om
e
t
r
y
m
et
hod.
T
he c
oor
d
i
nat
e po
i
nt
(
x
,
y
)
i
s
obt
a
i
ne
d b
y
c
a
lc
u
la
t
in
g
t
h
e c
ent
er
of
gr
a
v
i
t
y i
n
th
e
obj
ec
t
de
t
ec
t
i
ons
w
i
t
h
X
0=
M10/
M00
an
d
y0
=
M0
1/
M
00
[
13
].
S
om
e
r
es
ul
t
s
of
t
he
obj
ec
t
de
t
ec
t
i
on
an
d
C
O
G
poi
nt
ar
e
s
ho
w
n
in
F
i
gur
e
5
.
T
he
i
m
age
is
s
t
r
eam
ed
di
r
ec
t
l
y
,
an
d
t
h
e
C
O
G
i
s
s
i
m
u
l
t
an
eous
l
y
c
a
lc
u
la
t
ed
.
A
s
a
r
es
ul
t
,
t
he
s
e
l
f
dr
i
v
i
n
g c
ar
c
an a
dj
us
t
t
h
e c
oor
di
nat
e of
s
t
r
eet
t
o
f
ol
l
o
w
it
.
F
i
gur
e
2
.
F
l
o
w
c
har
t
of
s
t
r
eet
m
ar
k
det
ec
t
i
on s
y
s
t
em
(
a)
(b
)
(
c)
F
i
gur
e
3
.
T
he i
nt
er
f
ac
e of
c
ol
or
f
i
l
t
er
i
n
g pr
oc
es
s
i
n
s
t
r
e
et
m
ar
k
det
ec
t
i
on
S
y
s
t
em
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1
6
9
3
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
16
,
N
o
.
2
,
A
p
r
i
l
20
18
:
6
29
–
6
34
632
(d
)
(
e)
F
i
gur
e
3
.
T
he i
nt
er
f
ac
e of
c
ol
or
f
i
l
t
er
i
n
g pr
oc
es
s
i
n
s
t
r
e
et
m
ar
k
det
ec
t
i
on
S
y
s
t
em
F
i
gur
e
4
.
D
et
ec
t
i
on
ar
eas
i
n aut
onom
ous
c
ar
pr
ot
ot
y
p
e
:
(a
) t
o
p
v
i
e
w
,
(b
) s
i
d
e
v
ie
w
F
i
gur
e
5
.
D
et
ec
t
i
on
ar
eas
a
nd C
O
G
c
oor
d
i
n
at
e
s
3.
R
esu
l
t
s
a
n
d
A
n
al
ysi
s
3.
1.
C
o
lo
r
F
ilt
e
r
in
g
T
est
T
he
di
g
i
t
a
l
i
m
age pr
oc
es
s
i
n
g
w
as
b
ui
l
t
on R
as
pber
r
y
p
i
2
b
as
ed on
p
y
t
hon
pr
ogr
a
m
m
i
ng
l
an
gua
ge
a
nd
O
p
enC
V
lib
r
a
r
y
.
T
o
obt
ai
n
t
he
c
apa
bi
l
i
t
i
es
of
c
ol
or
f
i
l
t
er
pr
oc
es
s
,
t
es
t
i
ng
of
c
ol
or
f
ilt
e
r
in
g
wa
s p
e
r
f
or
m
ed
.
T
h
e t
es
t
w
as
pur
por
t
ed t
o
de
t
ec
t
t
he pr
es
enc
e of
wh
i
t
e
s
t
r
eet
m
ar
k
f
r
o
m
bl
ac
k
bac
k
gr
ound
.
T
he t
es
t
i
ng
wa
s
do
ne b
y
c
al
i
br
at
i
ng
t
he t
hr
es
h
ol
d of
c
ol
or
f
i
l
t
er
i
ng c
om
ponent
(
i.
e
hue
,
s
at
ur
at
i
on,
a
nd
v
a
l
ue
)
in
t
h
e
i
m
age
c
ol
or
s
pa
c
e
t
o
t
he
H
S
V
c
o
l
or
di
s
t
r
i
b
ut
i
o
n
m
et
hod.
T
hi
s
t
es
t
al
s
o pa
y
a
t
t
en
t
i
o
n t
o t
he i
nf
l
u
enc
e of
t
he i
nt
e
n
s
i
t
y
of
l
i
ght
or
i
l
l
um
i
nat
i
on p
r
ov
i
ded at
t
h
e
t
i
m
e of
t
es
t
i
ng c
ol
or
f
i
l
t
er
i
n
g.
I
l
l
um
i
nat
i
on r
ec
ei
v
e
d b
y
t
he c
am
er
a af
f
ec
t
s
t
he out
c
om
e o
f
i
m
ages
t
hat
c
an c
han
ge t
h
e c
ol
or
s
pac
e c
oo
r
di
nat
es
s
o t
hat
t
h
e t
hr
es
ho
l
d f
i
l
t
er
i
ng c
ol
or
s
us
ed
m
i
ght
not
be
c
or
r
ec
t
.
T
he i
m
age dat
a
is
bas
e
d
on
t
he
i
npu
t
c
am
er
a and
c
ol
or
-
f
ilt
e
r
ed
w
i
t
h H
S
V
m
et
ho
d.
T
r
ac
k
bar
i
s
des
i
gned
t
o
b
e
us
ed
i
n
t
h
e
c
al
i
br
at
i
on
pr
o
c
es
s
of
t
hr
es
hol
d
f
i
l
t
er
i
n
g
c
om
ponent
s
of
hue,
s
at
ur
at
i
on,
and
v
a
l
u
e.
D
at
a
w
i
dt
h
of
8
bi
t
s
w
as
a
ppl
i
ed
w
i
t
h
v
al
u
e
of
0
as
m
i
ni
m
um
v
al
ue
and
v
a
l
ue
of
255 as
t
he
m
a
x
i
m
u
m
v
al
ue
t
hat
r
epr
e
s
ent
s
t
he c
oor
di
n
at
es
b
as
ed on
t
he
hue
,
s
at
ur
at
i
on,
a
nd
v
a
l
ue of
a c
ol
or
.
I
n t
he O
p
enC
V
l
i
br
ar
y
,
s
at
ur
at
i
on an
d v
al
ue
i
s
r
ep
r
es
ent
ed
b
y
0
as
t
he
mi
n
i
mu
m
and 255
a
s
t
he
ma
x
i
mu
m
.
T
he
ma
x
i
mu
m
va
l
u
e
of
s
at
ur
at
i
on
i
s
t
he c
l
e
ar
c
ol
or
s
of
r
ed,
gr
een,
or
bl
ue
.
C
on
v
er
s
el
y
,
f
or
s
m
al
l
er
s
at
ur
at
i
on
v
al
ue
,
t
he c
ol
or
is
f
aded t
o
w
hi
t
e
y
of
t
he or
i
gi
nal
c
o
l
or
of
r
ed,
g
r
een,
or
b
l
ue.
Mea
n
w
h
il
e
,
f
or
t
he gr
eat
er
det
ec
t
ed
c
ol
or
v
al
ue
,
t
h
e
c
ol
or
s
ar
e br
i
gh
t
, w
h
il
e
s
m
a
ll
v
a
lu
e
t
ur
n
t
o b
l
ac
k
.
T
he f
i
l
t
er
c
om
ponent
s
of
hue r
epr
es
en
t
t
h
e c
o
or
di
n
at
es
of
t
h
e or
i
g
i
na
l
c
o
l
or
of
r
ed,
gr
een,
or
b
l
u
e t
o b
l
en
d t
h
em
i
n ac
c
or
danc
e w
i
t
h t
h
e
3D
c
ol
or
s
pac
e
d
ia
g
r
am
.
A
m
i
x
o
f
r
ed,
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
K
A
I
S
S
N
:
1
693
-
6
930
S
t
r
eet
M
ar
k
D
e
t
ec
t
i
on U
s
i
n
g R
as
pb
er
r
y
P
I
F
or
S
e
l
f
-
dr
i
v
i
ng
S
y
s
t
em (
S
um
ar
di
)
633
gr
een,
a
nd
b
l
ue
(
R
G
B
)
v
a
l
ues
ar
e
r
epr
es
ent
ed
i
n
de
g
r
ee
c
i
r
c
l
e
of
0
-
360
de
gr
ees
.
T
he
r
ange
of
hue v
a
l
ue i
s
r
epr
es
ent
ed
b
y
t
he
v
al
ue 0
as
a m
i
ni
m
u
m
v
al
ue
and
a m
ax
i
m
um
v
al
ue 1
79.
T
r
ac
k
bar
is
us
ed
i
n
pr
oc
es
s
c
ol
or
f
i
l
t
er
i
ng.
T
he
v
ar
i
ab
l
e
o
f
t
hr
es
hol
d
hu
e
f
i
l
t
er
i
ng
has
a
r
ange
of
v
a
l
ues
f
r
om
0
-
179.
T
he
v
ar
i
ab
l
e
of
s
at
ur
at
i
on
f
ilt
e
r
in
g
has
a
r
ang
e
v
al
u
e
of
0
-
25
5
.
F
i
l
t
er
i
ng
and
v
ar
i
ab
l
e t
hr
es
hol
d
v
a
l
ue
ha
s
a v
a
l
u
e r
an
ge of
0
-
25
5.
a.
H
u
e
F
i
l
te
r
i
n
g
T
he
t
es
t
i
ng
pr
oc
es
s
i
s
d
one
b
y
c
han
gi
ng
t
h
e
hu
e
f
i
l
t
er
i
ng
t
hr
es
ho
l
d
v
al
ue
on
t
he
t
r
ac
k
bar
.
T
es
t
i
ng
i
s
do
ne
b
y
f
i
nd
i
ng
t
he
t
hr
es
ho
l
d
v
al
u
e
of
t
he
m
i
ni
m
u
m
t
o
m
ax
i
m
u
m
and
f
r
o
m
ma
x
i
mu
m
t
o
mi
n
i
mu
m.
T
abl
e 1.
T
he
R
es
ul
t
of
H
ue
T
hr
es
hol
d C
a
l
i
br
at
i
on
C
al
i
br
at
i
on
Mi
n
.
V
al
ue
M
a
x.
V
al
ue
W
hi
t
e
O
b
j
ec
t
B
l
ac
k
O
bj
ec
t
M
i
ni
m
al
to
ma
x
i
m
al
0
74
N
ot
D
et
e
c
t
ed
D
et
ec
t
ed
M
a
x
im
a
l t
o
m
i
ni
m
al
78
179
D
et
ec
t
ed
N
ot
D
et
e
c
t
ed
T
abl
e 1
r
ev
e
al
s
t
he t
hr
es
ho
l
d f
i
l
t
er
i
ng c
om
ponent
f
or
c
ol
or
hu
e
.
I
t
c
an be
e s
een
t
hat
t
he
det
ec
t
i
on
of
bl
ac
k
and
w
h
i
t
e i
n
hu
e c
om
ponent
i
s
n
ot
s
ig
n
if
ic
a
n
t
.
T
hi
s
i
s
b
ec
a
us
e bl
ac
k
and
w
hi
t
e
ar
e
not
b
as
i
c
c
ol
or
s
,
but
a m
i
x
t
ur
e of
r
ed,
gr
e
en,
or
bl
u
e.
B
l
ac
k
and w
hi
t
e ar
e v
a
l
i
d o
n t
h
e
v
a
l
ue of
an
y
hu
e.
b.
S
a
tu
r
a
ti
o
n
F
i
l
t
e
r
i
n
g
S
at
ur
at
i
on t
es
t
i
n
g pr
oc
es
s
f
i
l
t
er
i
ng
i
s
do
ne
b
y
c
ha
n
gi
n
g t
h
e t
hr
es
ho
l
d
v
a
l
ue
on t
he
t
r
ac
k
bar
.
T
es
t
i
ng
i
s
do
ne
b
y
f
i
nd
i
ng
t
he
t
hr
es
ho
l
d
v
al
u
e
of
t
he
m
i
ni
m
u
m
t
o
m
ax
i
m
u
m
and
f
r
o
m
m
a
x
i
m
u
m
t
o
m
i
ni
m
u
m
.
i
s
i
nput
f
r
om
a v
i
de
o c
am
er
a
i
n r
eal
t
i
m
e on t
he t
es
t
i
n
g pr
oc
es
s
f
i
l
t
er
s
at
ur
at
i
on.
F
r
om
t
he t
es
t
i
n
g
of
t
hr
es
hol
d f
i
l
t
er
i
ng
f
or
c
o
m
ponent
c
ol
or
s
at
ur
a
t
i
on
as
s
een
in
T
abl
e 2
,
th
e
te
s
t
s
ho
w
s
t
hat
i
m
ages
of
bl
ac
k
c
ol
or
c
an be d
et
ec
t
e
d at
m
ax
i
m
u
m
s
at
ur
at
i
on
t
hr
es
hol
d
v
a
l
ues
a
nd
w
hi
t
e
c
ol
or
de
t
ec
t
i
on
w
i
t
h
m
i
ni
m
a
l
s
at
ur
at
i
o
n t
hr
es
h
ol
d v
al
ue
.
T
abl
e
2
.
T
he
R
es
ul
t
of
S
at
u
r
at
i
on
T
hr
es
hol
d C
al
i
br
a
t
i
o
n
C
al
i
br
at
i
on
S
at
ur
at
i
on
(
m
in
)
S
at
ur
at
i
on
(
m
ax
)
W
hi
t
e
O
b
j
ec
t
B
l
ac
k
O
bj
ec
t
M
i
ni
m
al
to
m
ax
i
m
al
0
30
D
et
ec
t
ed
N
ot
D
et
e
c
t
ed
M
a
xi
m
a
l
to
M
i
ni
m
al
25
255
N
ot
D
et
e
c
t
ed
D
et
ec
t
ed
c.
V
a
lu
e
F
ilt
e
r
in
g
T
he t
es
t
i
ng pr
oc
es
s
v
a
l
u
e
f
i
l
t
er
i
ng
i
s
don
e b
y
c
h
an
gi
n
g t
he t
hr
es
ho
l
d v
al
ue
o
n t
he
t
r
ac
k
bar
.
T
es
t
i
ng
i
s
do
ne
b
y
f
i
nd
i
ng
t
he
t
hr
es
ho
l
d
v
al
u
e
of
t
he
m
i
ni
m
u
m
t
o
m
ax
i
m
u
m
and
f
r
o
m
ma
x
i
mu
m t
o
mi
n
i
mu
m
.
F
r
om
t
he t
es
t
i
ng
of
t
hr
es
ho
l
d v
al
ue
f
or
c
om
ponent
c
o
l
or
f
i
l
t
er
i
n
g
as
s
ho
w
n i
n
T
abl
e 3
,
i
t
i
s
f
ou
nd t
ha
t
bl
ac
k
c
ol
or
c
an be det
ec
t
e
d at
a m
i
ni
m
u
m
v
al
u
e t
hr
es
ho
l
d
v
a
l
ue
and
t
he
w
h
i
t
e c
o
l
or
d
et
ec
t
i
on t
hr
es
ho
l
d
v
a
l
ue m
ax
i
m
u
m
v
al
ue.
T
abl
e
3
.
T
he
R
es
ul
t
of
V
a
l
u
e T
hr
es
hol
d C
al
i
br
at
i
o
n
C
al
i
br
at
i
on
V
al
ue
o
f
m
in
V
al
ue
of
m
ax
W
hi
t
e
O
b
j
ec
t
B
l
ac
k
O
bj
ec
t
M
i
ni
m
al
to
ma
xi
m
al
0
203
N
ot
D
et
e
c
t
ed
D
et
ec
t
ed
M
a
xi
m
a
l
t
o m
i
ni
m
al
200
255
D
et
ec
t
ed
N
ot
D
et
e
c
t
ed
3.
2.
T
e
s
t o
f
o
b
j
e
c
t c
o
o
r
d
i
n
a
te
d
e
te
c
ti
o
n
T
es
t
of
obj
ec
t
c
oor
di
nat
e
det
ec
t
i
on
is
per
f
or
m
ed
t
o c
hec
k
t
he abi
l
i
t
y
of
t
h
e
s
ys
t
e
m
in
det
ec
t
i
ng
t
h
e pr
es
e
nc
e
of
obj
ec
t
s
i
n
t
h
e f
or
m
of
s
t
r
e
et
m
ar
k
bas
ed o
n t
he
c
oo
r
di
nat
es
of
t
h
e
c
a
m
er
a
i
m
age
f
i
l
t
er
.
C
o
or
d
i
nat
es
ar
e
ob
t
ai
ned
f
r
om
c
ent
r
al
p
oi
n
t
o
n
t
he
obj
ec
t
of
f
i
l
t
er
ed
ar
ea
m
ar
k
ed
b
y
a
gr
ee
n
s
quar
e
s
hape.
C
oor
d
i
nat
e
v
al
u
es
of
t
he
x
-
ax
i
s
and
y
-
a
x
is
is
0
o
r
m
in
im
a
l
in
t
he u
pper
l
ef
t
c
or
ner
i
n t
h
e i
m
age r
ec
ei
v
e
d b
y
t
h
e
c
a
m
er
a.
T
he t
es
t
dat
a c
or
r
es
pond
i
ng
t
o
c
oor
di
n
at
e
de
t
ec
t
i
on
is
s
ho
w
n i
n
T
abl
e 4
,
w
h
i
le
t
h
e
r
ef
e
r
enc
e
c
o
or
di
nat
e
ax
es
x
a
nd
y
ax
i
s
(
0,
0)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1
6
9
3
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
16
,
N
o
.
2
,
A
p
r
i
l
20
18
:
6
29
–
6
34
634
is
t
he up
per
l
ef
t
pi
x
e
l
of
t
he i
m
age.
T
hi
s
i
s
t
he i
m
age m
odel
ed i
nt
o a m
at
r
i
x
of
pi
x
el
s
pos
s
es
s
ed.
T
he
r
es
ol
ut
i
on
us
e
d
i
s
240
x
240
s
o
t
hat
t
he
m
ax
i
m
u
m
v
al
u
e
of
t
he
x
-
ax
i
s
i
s
240.
I
n
t
es
t
i
ng
t
he
v
a
l
ue
of
t
he
x
-
ax
i
s
an
d y
-
ax
i
s
m
ax
i
m
u
m
v
al
ue
i
s
o
nl
y
a
bou
t
1
60 f
or
d
et
ec
t
i
ng
a f
or
m
of
dat
a
t
ak
en
i
s
t
he
c
ent
er
of
t
he
o
bj
ec
t
.
W
hi
l
e
t
he
pr
oc
es
s
of
det
ec
t
i
o
n
of
t
he
p
i
x
el
ar
ea
i
s
l
i
m
i
t
ed
t
o
a
m
i
ni
m
u
m
v
al
ue
t
o
av
oi
d
det
ec
t
i
on
of
noi
s
e
.
T
abl
e
4
.
T
he
R
es
ul
t
of
c
oor
di
n
at
es
o
bj
ec
t
det
ec
t
i
o
n c
al
i
br
at
i
on
C
al
i
br
at
i
on
X
V
al
ue
Y
V
al
ue
X
pos
i
t
i
on
Y
po
s
i
t
i
on
1
94
86
C
ent
er
C
ent
er
2
167
70
R
i
ght
C
ent
er
3
7
73
Lef
t
C
ent
er
4
98
4
C
ent
er
T
op
5
89
139
C
ent
er
B
o
tt
o
m
6
4
6
Lef
t
T
op
7
166
129
R
i
ght
B
o
tt
o
m
4.
C
o
n
c
l
u
s
i
o
n
T
he
s
t
r
eet
m
ar
k
det
ec
t
i
on
has
been
su
cce
ssf
u
l
l
y
b
u
i
l
t
in
R
as
pber
r
y
-
pi
2
and
c
an
det
ec
t
s
t
r
eet
m
ar
k
s
and det
er
m
i
ni
ng t
he t
r
ac
k
i
ng c
oor
d
i
nat
e
w
i
t
h
C
O
G
m
et
hod.
P
r
oc
es
s
i
ng
r
at
e
of
t
h
i
s
al
g
or
i
t
hm
w
h
i
c
h r
uns
o
n 9
00MH
z
q
uad
-
c
or
e
A
R
M
C
or
t
ex
A
7
i
s
136
.
48 m
s
or
7.
2 F
P
S
.
T
he
r
es
ul
t
s
of
H
S
V
t
hr
es
h
ol
d
i
n
c
al
i
br
at
i
on pr
oc
es
s
i
s
hue
m
a
x
=
179,
hu
e m
i
n=
0,
s
at
ur
at
i
o
n
m
ax
=
30,
s
at
ur
at
i
on
m
i
n=
0,
V
al
ue
m
ax
=
25
5,
an
d t
he
v
al
ue m
i
n=
2
00.
T
he t
hr
es
ho
l
d
v
al
u
e c
an det
ec
t
a
s
t
r
eet
m
ar
k
ed w
i
t
h t
he c
o
or
di
n
at
es
w
i
t
h
a r
an
ge
of
4
-
1
67
o
n t
h
e x
ax
i
s
and
4
-
1
3
9
on
t
he
y
-
a
xi
s.
S
et
po
i
nt
of
c
oor
di
nat
es
we
r
e
obt
ai
ned
w
it
h
c
a
lc
u
la
t
i
on o
f
C
O
G
poi
nt
of
f
i
l
t
er
ed de
t
ec
t
ed ar
ea
i
n
(X
,
Y
).
T
he s
t
r
eet
m
ar
k
det
ec
t
i
o
n
c
an
pr
ov
i
d
e pr
e
di
c
t
i
on
of
t
he
s
t
r
e
et
m
ar
k
i
n f
r
ont
ei
t
h
er
i
n
s
t
r
ai
ght
l
a
ne
or
i
n t
ur
n
i
ng l
a
ne
.
R
ef
er
en
ce
s
[1
]
S
. D
e
v
i
tt, S
.
Fl
a
n
n
e
r
y
, G
. L
o
c
r
a
ft,
A
.
W
o
od,
K
.
W
e
i
s
s
,
an
d A
.
S
c
he
nk
er
.
A
ut
ono
m
ou
s
C
ar
s
S
e
l
f
-
D
r
i
v
i
ng t
he
N
ew
A
ut
o I
nd
us
t
r
y
P
ar
adi
g
m
,
M
or
gan
S
t
anl
ey
,
20
13
:1
-
10
9
[2
]
H
. C
h
o
, Y
. S
e
o
, B
. V
.
K
.
V
.
K
um
ar
,
a
nd R
.
R
.
R
aj
k
u
m
ar
.
A
Mu
l
t
i
-
S
ens
or
F
u
s
i
o
n S
y
s
t
em
f
or
M
ov
i
ng
O
bj
ec
t
D
et
e
c
t
i
on and T
r
a
c
k
i
n
g i
n U
r
ban D
r
i
v
i
ng E
nv
i
r
onm
ent
s
,
I
E
E
E
I
nt
er
nat
i
ona
l
C
onf
er
enc
e on
R
obot
i
c
s
&
A
ut
o
m
at
i
on,
H
o
ng
k
ong,
2
014
:
183
6
–
1
843
.
[3
]
M
ohom
ed,
I
q
bal
.
Se
l
f
-
dr
i
v
i
ng
L
ego M
i
nds
t
or
m
s
R
obot
,
P
r
oc
.
P
y
t
hon
i
n s
c
i
en
c
e
C
on
f
.
(SC
I
P
Y
),
2012
.
[4
]
U
jja
in
iy
a
,
L
o
h
i
t,
M
.
K.
C
hak
r
a
v
ar
t
hi
.
R
as
pber
r
y
-
P
i
B
as
e
d
C
os
t
E
f
f
ec
t
i
v
e
V
ehi
c
l
e
C
ol
l
i
s
i
on
A
v
oi
dan
c
e
S
y
s
t
em
U
s
i
ng
I
m
a
ge P
r
o
c
e
s
s
i
ng,
A
RP
N J
.
En
g
.
A
ppl
.
Sc
i
,
2
015
;
1
0(
7)
[5
]
P
.
Z
hao,
J
.
C
hen,
Y
.
S
ong,
X
.
T
ao,
T
.
X
u,
and T
.
M
ei
.
D
es
i
gn
of
a C
ont
r
ol
S
y
s
t
em
f
or
an A
u
t
ono
m
ou
s
Ve
h
i
c
l
e B
a
s
ed
on A
dapt
i
v
e
-
P
ID
,
In
t. J
.
A
dv
.
R
ob
ot
ic
S
ys
t
.
,
I
NT
E
CH,
Ri
j
e
k
a
,
20
12
;
9(
44)
[6
]
T
ek
al
p,
A
.
M
ur
at
.
D
i
gi
t
a
l
v
i
de
o pr
oc
e
s
s
i
ng
.
P
r
ent
i
c
e
H
al
l
P
r
es
s
,
20
15.
[7
]
J
.
M
.
A
l
v
ar
ez
and A
.
M
.
Lopez
.
R
oad det
ec
t
i
on
bas
ed o
n i
l
l
u
m
i
na
nt
i
nv
ar
i
an
c
e,
I
E
E
E
T
r
an
s
.
I
nt
el
l
i
g
ent
T
r
ans
por
t
at
i
on
S
y
s
t
em
s
,
2010
.
[8
]
T
.
K
uhnl
dan J
.
F
r
i
t
s
c
h.
V
is
i
o
-
s
pat
i
al
r
oa
d boun
dar
y
de
t
ec
t
i
o
n f
or
unm
ar
k
ed ur
b
an and r
ur
al
r
oad
s
,
in
I
E
E
E
I
nt
el
l
i
g
ent
V
ehi
c
l
e
s
S
y
m
pos
i
um
P
r
oc
e
edi
ngs
,
20
14
:
125
1
–
12
56.
[9
]
D.
P
o
ns
a,
J
.
S
er
r
at
,
A
.
M
.
Lóp
ez
.
O
n
-
boar
d
i
m
ag
e
-
b
as
e
d
v
e
hi
c
l
e
d
et
e
c
t
i
o
n
a
nd
t
r
ac
k
i
ng,
Tr
a
n
s
.
I
n
s
t
.
M
eas
.
C
ont
r
ol
,
201
1;
33(
7)
783
–
805
[
10]
S
. F. X
. B
a
y
e
r
l
, T
. L
u
e
tte
l
, H
. W
u
e
n
s
c
h
e
.
F
ol
l
ow
i
n
g D
i
r
t
R
oa
ds
at
N
i
ght
-
T
i
m
e:
S
en
s
or
s
a
n
d F
eat
ur
e
s
f
or
La
ne R
e
c
og
ni
t
i
on
and
T
r
a
c
k
i
ng,
P
r
oc
e
edi
n
gs
I
E
E
E
/
R
S
J
I
nt
er
nat
i
onal
C
onf
er
en
c
e
on
I
nt
el
l
i
ge
n
t
R
obot
s
and
S
y
s
t
em
s
,
2015
;
11
7
–
12
2.
[
11]
S
t
ei
n,
G
i
deon
P
.
,
O
f
er
M
ano,
and
A.
S
has
hua,
V
is
io
n
-
b
as
ed
A
C
C
w
i
t
h
a
S
i
ngl
e
C
am
er
a:
B
ound
s
o
n
R
ange a
nd R
a
nge R
a
t
e A
c
c
ur
ac
y
,
I
E
E
E
I
V
2
003 I
nt
e
l
l
i
g
ent
V
ehi
c
l
e
s
S
y
m
po
s
i
u
m
,
U
SA,
2
0
0
3.
[
12]
Bu
cz
ko
w
ski
,
M
.
and S
t
as
i
ns
k
i
,
R
.
,
A
ut
om
at
i
c
L
ane D
et
e
c
t
i
o
n
,
P
W
T
20
12,
P
oz
na
n,
2
012
[
13]
Ard
e
s
h
i
r,
A,
I
m
a
ge R
eg
i
s
t
r
at
i
o
n:
P
r
i
n
c
i
p
l
es
,
T
o
ol
s
and
M
et
hods
,
S
pr
i
ng
er
,
Lo
ndo
n,
2
012.
Evaluation Warning : The document was created with Spire.PDF for Python.