Image Fusion Using Manual Segmentation
Image Fusion Using Manual Segmentation
International Journal of EmergingTrends & Technology in Computer Science(IJETTCS)
Web Site: www.ijettcs.org Email: [email protected], [email protected]
Volume 2, Issue 6, November December 2013 ISSN 2278-6856
Volume 2, Issue 6 November December 2013 Page 103
Abstract: In this paper, a new technique is introduced to
develop fused image with help of the spatial frequency. First,
the image is divided into several blocks by using manual
segmentation technique. The spatial frequencies of each
individual blocks is calculated. Place the blocks which are
having high spatial frequencies into corresponding blocks of
another image called fused image. Experimental results
indicate the superiority of the proposed method for
multifocus images.
Keywords: Image fusion, spatial frequency, discrete
wavelet transform
1. INTRODUCTION
Image fusion is one of the best processes for combining
images, obtained from sensors of different wavelengths
simultaneously viewing of the same scene, to form a
composite image. This composite image is known as
fused image. The resultant fused image is very clear and
perceptible to viewers [3]. The advantage of image is to
improve capability by complimentary information and
improving reliability by redundant information. To do
process on images, the images should be registered
[1].Register the images, which are taking from the
cameras to do operation image fusion.
The image fusion used to increase relevant information
related to particular task or information [1,2]. Image
fusion can be done by using different techniques. Image
fusion method can be categorized into two types, namely
as Spatial domain fusion method and Transform fusion
method. Spatial domain fusion method is a simplest
fusion method compare with the other methods [2]. This
method is directly deal with pixels. Here the source
images taken by pixel-by-pixel, but it had some
undesirable side effects such as reduced contrast. In
Transform domain technique, first the image transferred
into frequency domain and applied all fusion operations
on it, finally applied inverse forier transform to get
resultant fused image.
Image fusion reduced uncertainty and minimizes
redundancy in the output image [5]. In recent years a new
mathematical approach method was introduced for
getting accurate and cleared fused image with the help of
wavelet transform technique. Wavelet transforms used in
graphics, imagery and in medical images. Image fusion
algorithm based on wavelet transform, which faster
developed was a multi resolution analysis image fusion
method [4]. This technique had good timing frequency
characteristics. It was successfully applied in image
processing field.
The information was given in this paper, in section 2):
Brief introduction to image fusion. In section 3), the
spatial frequency. In section 4), proposed method to form
a fused image. In section 5) experimental results for
proposed method.
2. IMAGE FUSION
In recent days image fusion became more important in
image analysis and computer vision [6]. This image
fusion refers to image processing techniques that produce
a enhanced fused image by combining all images from
two or more sensors [7]. The resultant fused image
having accurate information which was related to all
source images very clear and perceptible to viewers.
Objectives of image fusion scheme:
- extract all complimentary information from all
source images
- do not introduce artifacts which will distract
human observers
An important step in image fusion is registration that
means corresponding pixel positions in the source images
must refer to the same location [6]. A lot of work has
been done in the area of multi-focus image fusion. Several
algorithms proposed for image fusion for various
applications [9].Image fusion can be done as simply
taking pixel-by-pixel average of source images, but the
resultant image having drawbacks like reduced contrast.
Basically image fusion has two methods, which are used
to produce an accurate fused image, are spatial domain
fusion and frequency domain fusion [8]. The fusion
methods such as Brovey method, principal component
analysis and IHS based methods are spatial domain
approaches. The main disadvantage of spatial domain
Image Fusion Using Manual Segmentation
Venkateswara Rao.N
1
, Veeraswamy.K
2
, Himabindu.CH
3
1
PG Student, Electronics and Communication Engineering Department,
QIS College of Engg & Tech.,Ongole, India
2
Professor & Principal, QIS College of Engg & Tech., Ongole, India
3
Assoc.Prof., QIS College of Engg & Tech., Ongole, India
International Journal of EmergingTrends & Technology in Computer Science(IJETTCS)
Web Site: www.ijettcs.org Email: [email protected], [email protected]
Volume 2, Issue 6, November December 2013 ISSN 2278-6856
Volume 2, Issue 6 November December 2013 Page 104
approach is that they produce spatial distortion in the
fused image. Some methods such as discrete wavelet
transform, laplacian pyramid based, curvlet transform
based methods show a better performance in spatial and
spectral quality of the fused image compared to the spatial
methods of fusion.
In generally, depending on the stage at which the
combination mechanism takes place, image fusion can be
categorized into three types, namely as Pixel level,
Feature level, and Decision level [11]. Work on directly
with pixels of the output images which were obtained by
sensors. At same time in feature level fusion, on the other
hand works on the image features extracted from the
source images. In the decision level fusion, it involves at
high level, and merges the interpretations of different
images obtained after image understanding [10].
Related research fields of image fusion:
- computer vision
- automatic object detection
- image processing
- parallel and distributed processing
- robotics
- remote sensing
3. SPATIAL FREQUENCY
Spatial frequency is used to measure overall activity level
in image. The images divide into blocks having the size
as X Y. Spatial frequency was originated from human
visual system, indicates the overall active level in image.
The use of spatial frequency has led to an effective
objective quality index for image fusion [6,13].
Consider the image having the blocks of having size X
Y, where X is known as number of rows and Y is known
as number of columns. For Calculating spatial frequency
need to find row frequency and column frequency.
The row frequency can be calculated by using the
formula given below,
RF=
=
1
0
1
1
2
)] 1 , ( ) , ( [
1
X
x
Y
y
y x F y x F
XY
(3.1)
Where F(x, y) is the gray value of pixel at position
(x, y) of image F.
The column frequency can be calculated by using the
formula is given below,
CF=
=
1
0
1
1
2
)] , 1 ( ) , ( [
1
Y
y
X
x
y x F y x F
XY
(3.2)
The total spatial frequency of image is
SF=
2 2
) ( ) ( CF RF + (3.3)
4. PROPOSED METHOD
The fusion process can be done by using the following
steps.
Step 1: Consider two multi focus images A & B to be
fused
Step 2: Apply manual segmentation on respective images
A & B to compute spatial frequency of selective block of
the image SF
i
A
and SF
i
B.
Where i is number of blocks
Step 3: Compare the spatial frequency of each block of
image with the following condition
SF
i
=
<
>
B
i
A
i
B
i
B
i
A
i
A
i
SF SF B
SF SF A
( 4.1 )
Step 4: TheSF
i
gives the final fused image.
Figure 1: Schematic diagram for proposed image fusion
method
5. EXPERMENTAL RESULTS
Experiments have been conducted on three sets of images
using Mat lab 7.8.0 (R2009a) version. The consider
images are named as Pepsi, clock and disk images having
sizes like as 256 256, 128128 and 480640
respectively. In figs 2,3,4 (a) and (b) images are original
source images. (c) is the fused image by using the basic
average DWT transform technique (d) is the fused image
by proposed method.
The parameter, mutual information used to compare the
results obtained by basic average DWT transform with
proposed method. The mutual information is a metric
defined as the sum of mutual information between each
input and fused image. Considering the two input images
A and B , and a resulting fused image F
I
FA
( f , a ) =
a f
p
,
FA
( f , a ) log
) ( ) (
) , (
a p f p
a f p
A F
FA
( 5.1 )
I
FB
( f , b ) =
b f
p
,
FB
( f , b ) log
) ( ) (
) , (
b p f p
b f p
B F
FB
( 5.2 )
Thus the fused image fusion performance measure can be
defined as
MI
F
AB
=I
FA
(f, a ) +I
FB
( f , b ) (5.3)
This indicates that the proposed measure reflects the total
amount of information that the fused image F contains
that of A and B. for both criteria, the bigger the value, the
better is the fusion result. The values MI of figs: 2-4 are
listed in table 1. Based on the table, we can observe that
the proposed method provides better performance over
Image
A
Image
B
Spatial
frequency
Spatial
frequency
Compare
Fused
Image
Manual
segment-
ation
Manual
segment-
ation
International Journal of EmergingTrends & Technology in Computer Science(IJETTCS)
Web Site: www.ijettcs.org Email: [email protected], [email protected]
Volume 2, Issue 6, November December 2013 ISSN 2278-6856
Volume 2, Issue 6 November December 2013 Page 105
basic average wavelet transform technique in terms of
mutual information.
Figure2. a).Source image A. b).Source image B. c).fused
image by using basic average DWT. d). proposed method
Figure 3: a) source image A. b) Source image B. c)
fused image using basic average DWT. d) Proposed
method.
Figure 4 a).source image A. b).Source image B. c). fused
image by using basic average DWT. f). proposed method.
Table 1: The fusion methods performance measures
based on mutual information
Mutual information
Techniq
ue
Pepsi Clock disk
DWT 3.0880 3.5707 3.2193
Propose
d
method
3.440 5.0521 5.5665
CONCLUSION
In this paper, a new technique proposed to obtain fused
image. The performance of proposed method is compared
with basic DWT technique. In Mutual Information of
proposed method is higher than basic method. The
proposed method is computationally efficient and simple.
The scope of this work can be extended with automatic
segmentation to get more efficient fused image.
Acknowledgement
This research work is carried with the facilities(QIS
College of Engineering and Technology, Ongole ) of
Research Promotion Scheme (RPS) grant by the AICTE,
India with ref no: 20/AICTE/RIFD/ RPS (policy III )
55/2012-13.
References
[1] A. Ardeshir Goshtasby. Fusion of multifocus images
to Maximize image information. Dayton, Oh45435.
[2] Deepak Kumar Sashu, M.p.parsai, Different Image
Fusion Techniquesvol.2, Issue. 5sep-Oct 2012 pp-
4298-4301 ISSN:2249-6645
[3] Deepali A.Godse, Dattatraya S.Bormane (2011)
wavelet based image fusion using pixel based
maximum selection rule International journal of
Engineering Science and Technology (IJEST), Vol.3
No 7 July 2011, ISSN: 0975-5462.
[4] Ankur Upadhyay, Ujwal Harode, Image fusing using
Transportation Approach, PIIT, New Panvel.
[5] A.Toet, Hierarchical image fusion, machine vision
and applications. 3(1990) 1-11
[6] Shutao Li, James T.Kwok, Yaonan Wang,
Combination of images with diverse focuses using
the spatial frequency 15 June 2001.
[7] J.K.Agarwal, Multi sensor image fusion for
computer vision, spinger-verlag, Berlin, Germany,
1993.
[8] Kelly E.Neriani, Alan R.Pinkus;
DavidW.Domett,An Investigation of image fusion
Algorithms using a visual performance-based image
evolution methodology. April 17, 2008.
[9] Bin yang, P Shah, Multi focus image fusion by
combining curvelet and wavelet transform, pattern
recognition letters vol.29,no.9, pp.1295-1301, 2008
[10] Shutao Li, James Tin-Yau Kwok,Ivor Wai-Hung
Tsang, and Yaonan Wang Fusing images with
different focuses using support vector machines,
nov-2004.
[11] P.K.Vardhney, Multi sensor data fusion, Electron.
Commiun.Eng. pp.245-253,Dec. 1997
[12] Bin Yang, Shutao Li, Multi-focus image fusion
based on spatial frequency and morphological
operators. Changsha 410082.2007.
[13] Marturi haribabu, Ch.Hima Bindu, Dr.K.Satya
Prasad Multimodal medical image fusion of MRI-
PET using wavelet transform.IEEE-2012 MNC
Applications.
International Journal of EmergingTrends & Technology in Computer Science(IJETTCS)
Web Site: www.ijettcs.org Email: [email protected], [email protected]
Volume 2, Issue 6, November December 2013 ISSN 2278-6856
Volume 2, Issue 6 November December 2013 Page 106
AUTHOR
N.venkateswara Rao is a PG student in QIS
College of Engineering & Technology, Ongole,
Andhra Pradesh, India.
K.VeeraSwamy is currently Professor in ECE
department and Principal of QIS College of
Engineering and Technology, Ongole, A.P, India.
He received his Ph.D from J NTUK, Kakinada.
He has fifteen years experience in teaching
under graduate and post graduate students. He published 66
papers in national/international conferences/journals. Presently
he is guiding 5 students for their PhD work. His research
interests are in the areas of image compression, image
watermarking, Face recognition, CBIR, and networking
protocols.
Ch.Hima bindu is currently working as
Associate Professor in ECE Department,
QIS College of Engineering & Technology,
ONGOLE, Andra Pradesh, India. She is
working towards her Ph.D. at JNTUK, Kakinada, India.
She received her M.Tech. from the same institute. She
has ten years of experience of teaching undergraduate
students and post graduate students. She has published 11
research papers in International journals and more than 9
research papers in National & International Conferences.
Her research interests are in the areas of image
Segmentation, image Feature Extraction and Signal
