车牌识别外文文献翻译中英文Word文件下载.docx
《车牌识别外文文献翻译中英文Word文件下载.docx》由会员分享,可在线阅读,更多相关《车牌识别外文文献翻译中英文Word文件下载.docx(14页珍藏版)》请在冰豆网上搜索。
paper,a
new
algorithm
based
improved
BP
(back
propagation)
neural
network
for
Chinese
vehicle
license
plate
recognition
(LPR)
is
described.The
proposed
approach
provides
a
solution
the
plates
(VLP)
which
were
degraded
severely.What
it
remarkably
differs
from
traditional
methods
application
of
prior
knowledge
to
procedure
location,segmentation
recognition.Color
collocation
used
locate
in
image.Dimensions
each
character
are
constant,which
segment
VLPs.The
Layout
VLP
an
important
feature,which
construct
classifier
recognizing.The
experimental
results
show
that
effective
under
condition
severely.
Index
Terms
recognition,prior
knowledge,vehicle
plates,neural
network.
I.INTRODUCTION
Vehicle
License-Plate
very
interesting
but
difficult
problem.It
number
applications
such
as
weight-and-speed-limit,red
traffic
infringement,road
surveys
park
security[1].VLP
system
consists
location,the
characters
segmentation,and
recognition.These
tasks
become
more
sophisticated
when
dealing
with
images
taken
various
inclined
angles
or
lighting,weather
cleanliness
plate.Because
problem
usually
real-time
systems,it
requires
not
only
accuracy
also
fast
processing.Most
existing
methods[2],[3],[4],[5]reduce
complexity
increase
rate
by
using
some
specific
features
local
VLPs
establishing
constrains
position,distance
camera
vehicles,and
angles.In
addition,neural
was
rate[6],[7]but
seldom
consider
VLPs.In
paper,we
learning
method
overcomes
low
speed
convergence
network[8]
remarkable
increases
especially
degrade
II.SPECIFIC
FEATURES
OF
CHINESE
VLPS
A.Dimensions
According
guideline
inspection[9],all
must
be
rectangular
have
dimensions
all7characters
written
single
line.Under
practical
environments,the
distance
vehicles
constant,so
all
fixed
width,and
between
medium
axes
two
adjoining
ratio
width
height
nearly
constant.Those
can
individual
character.B.Color
plate
There
four
kinds
color
plate.These
collocations
shown
table
I.
TABLE
I
Moreover,military
police
wagon
contain
red
belongs
set.This
feature
improve
rate.
C.Layout
The
criterion
defines
layout
plate.All
standard
characters,numbers
letters
Fig.1.The
first
one
abbreviation
provinces.The
second
letter
ranging
A
Z
except
I.The
third
fourth
ones
numbers.The
fifth
seventh
numbers
from0to9only.However
may
special
(as
Fig.1).
After
segmentation
process
extracted.Taking
advantage
knowledge,the
will
enter
classes:
abbreviations
provinces
set,letters
set,number
set,special
set.
(a)Typical
layout
(b)
Special
character
Fig.1The
III.THE
PROPOSED
ALGORITHM
This
modules:
location,character
segmentation,character
classification
recognition.The
main
steps
flowchart
LPR
Fig.2.
Firstly
located
input
image
segmented.Then
every
enters
decide
class
to,and
finally
decides
cha