山西大学学报(自然科学版
)32(
的:
566~571
,
2009
lournal
of
Shanxi University(Nat.
Sci.
Ed. )
Article
ID:
0253-2395 (2009 )04-0566-06
The Application
of
Wavelet
Transform
to
the
Image Segmentation
LI
Yue-e
,
LIU
Qing-fang
(College
of
Physics
and
Electronics
Engineering
,
Shanxi
University
,
Taiyuan
030006 ,
China)
Abstract: This article introduces some threshold segmentation algorithms based on
wavel
巳
t
transform and gives an
image segmentation method using Canny operator
and
wavelet transform based edge information fusion.
It
also ex
pounds a texture segmentation method. Firstly
, it uses wavelet transform
to
decompose the image and extract the
fea
ture,and then uses
fuzzy
c-mean clustering
to
do
the texture segmentation. Additionally the characteristic,scope and
their developing direction
of
many kinds
of
segmentation methods are investigated.
Key
words: image segmentation; wavelet transform; feature extraction;
fuzzy
c-mean clustering; threshold
CLC
number: TP391 Document
code:
A
1
Approaches
of
image
segmentation
Image
s
巳
gmentation
technique
is
generally
classified
as
two
types
of
methods
based
on
edge
detection
and
region
口
.2.3J.
Image
segmentation
methods
based
on
edge
detection
find
the
discontinuity
of
local
feature
in
an
image
by
using
local
window
operating
,
and
then
connect
these
discontinuous
image
elements
to
shape
borderlines.
These
borderlines
would
segment
the
image
into
different
locations.
Image
segmentation
meth-
ods
based
on
region
segment
an
image
into
several
regular
blocks
firstly
in
accordance
with
a
certain
criteri-
on
,
since
according
to
the
fundamental
of
attribute
coherence
,
then
s
巳
parate
inconsistent
image
blocks
ac-
cording
to
principle
of
the
same
property
again
and
again
and
amalgamate
adjacent
image
blocks
of
identical
property
,
finally
fromate
a
sheet
of
area
graph.
In
the
actual
application
,
research
personnel
put
forward
many
segmentation
methods
from
different
theoretical
angle.
These
algorithms
are
based
on
threshold
or
based
on
texture.
In
this
paper
,
getting
along
with
the
application
of
wavelet
transform
and
analysis
in
the
image
segmentation
,
we
discuss
the
following
methods
of
edge
information
fusion
including
gray
threshold
of
image
histogram
,
Mallat
algorithm
and
Canny
operato
r.
Moreover
we
also
research
the
image
segmenta-
tion
method
of
the
extracting
texture
feature
based
on
the
combining
Multi-band
wavelet
with
fuzzy
c
means
clustering
(FCM)
[4
J
•
Finally
we
have
proposed
the
analysis
and
comparison
of
the
feature
and
scope
of
many
kinds
of
segmentation
methods
and
segmentation
algorithms
2
Image
segmentation
based
on
wavelet
transform
and
analysis
Image
segmentation
based
on
wavelet
analysis
is
a
method
in
virtue
of
wavelet
which
is a
kind
of
math-
ematic
too
1.
Wavelet
transform
is
a
location
transform
from
airspace
to
frequency-space.
With
characters
of
compact
support
,
variable
focal
length
wavelet
functions
can
describe
local
information
of
a
signa
1.
Wavelet
functions
may
perform
multi-scale
thinning
analysis
to
signals
by
using
their
functions
of
dilations
and
translations
operator
,
provide
unified
accurate
frames
to
analyze
signal
character
in
different
scales.
In
view
of
image
segmentation
,
wavelet
decomposition
offers
completely
mathematical
description.
Wavelet
trans-
Received
date: 2009-08-25
Foundation item: Productivity Promotion Center Project
in
Shanxi Province
Biography: LI Yu
e-
e( 1955-) ,Female , born
in
Yuncheng , Shanxi, Professor, reserch direction: Signal and Image Processing