Information Technology Reference
In-Depth Information
Definition 3 (A Suspicious License Plate,
SLP
).
A license plate
LP
is deemed to be suspicious if it only occurs in a single
camera
ca
i
and does not emerge in its
n
−
1
neighbors
CA
=
{
ca
1
,ca
2
, ..., ca
n−
1
}
.
Otherwise, it should be captured by the neighboring cameras of
ca
i
with high
probability. We claim that
LP
is a suspicious plate, and the probability that it is
incorrectly recognized can also be lower bounded by:
n−
1
Pr
(
SLP
)
≥
1
−
p
i
(1
−
p
j
)
(2)
j
=1
Here,
p
i
indicates the precision of the camera
ca
i
∈ CA
.
Definition 4 (
k
-degree Neighbors of a Camera).
The vast cameras deployed all over the city can be modeled as a directed graph
G
=
,here
V
=
CA
is the set of all cameras, and
E
denotes the road
between cameras. For any two cameras
ca
i
,ca
j
,and
ca
i
,ca
j
∈
{
V,E
}
V
,if
ca
i
is con-
nected to
ca
j
directly by roads without crossing another
ca
k
,
E
.For
a
ca
i
,its
k
-degree neighbors are those cameras from which it takes
k
hops to
camera
ca
i
, plus those cameras to which it takes
k
hops from camera
ca
i
.Es-
pecially, 0-degree stands for current camera itself. Formally,
k
-degree neighbors
Nb
k
(
ca
i
)
of camera
ca
i
can be defined as follows:
ca
i
,ca
j
∈
Nb
k
(
ca
i
)=
{ca
j
| <ca
i
,ca
j
>∈ E∨ <ca
j
,ca
i
>∈ E} , k
=1
{ca
j
|
(
<ca
k
,ca
j
>∈ E∨ <ca
j
,ca
k
>∈ E
)
∧ ca
k
∈ Nb
k−
1
(
ca
i
)
} ,k>
1
(3)
Fig. 2.
k
-degree neighbors of node A
Fig. 2 is an abstract view of roads and cameras in urban area. In Fig. 2, the
nodes stand for cameras and the links stand for roads. The number
k
in nodes
stands for
k
-degree of node A. For example, there are 4, 6, 4 points corresponding
to 1,2,3-degree neighbors respectively.
3.2 A Heuristic License Plate Correction Method
A Heuristic License Plate Correction Method Framework.
As a formal
specification of the method has been introduced at the beginning of Section 3,
an overview of HelpMe framework is depicted in Fig. 3.