Beruflich Dokumente
Kultur Dokumente
Research Scholar, Department of Computer Science and Engineering, Jntuh, Hyderabad, Telangana, India
2
3
Principal, University College of Engineering, Acharya Nagarjuna University, Andra Pradesh, India
ABSTRACT
Recently wave atoms have been introduced, as a variant of 2D wavelet packets obeying the parabolic scaling law
support size ~sqrt (wavelength), particularly well suited for representing oscillatory patterns in images. Wave atoms offer
an effective solution to the problems associated with image denoising using wavelets and Curvelets. Finger prints possess
the unique properties of distinctiveness, oscillatory patterns and persistence. However, their image contrast is poor due to
mixing of complex type of noise. In this paper an attempt has been made to present the results of denoising of such images
using wavelets, curvelets and wave atoms. The results obtained demonstrate that the wave atoms based reconstructions are
visually sharper than the wavelet and curvelet reconstructions. The recovery of edges and oscillatory features is of
particularly superior quality. The results obtained are in accordance with the expected predictions of the existing theory of
wave atom transforms.
www.tjprc.org
editor@tjprc.org
70
Curvelet transform is considered as the best multiscale geometric transforms [6, 7, 8, 9, 10 & 11]. In this
transform, the frame elements are indexed with scale, orientation and location parameters. It is designed to represent edges
and the singularities along curved paths more efficiently than the wavelet methods.
Recently wave atoms [12] have been introduced, as a variant of 2D wavelet packets obeying the parabolic scaling
law support size ~sqrt (wavelength), particularly well suited for representing oscillatory patterns in images.
psnr = 10 log10
M 1 M 2 max( f (i, j )) 2
M1 M 2
dB
(1)
i =1 j =1
Here size of images are M1 and M2. F (i, j) and f (i, j) are original ands denoised image.
71
is 4 to 5dB and wave atom denoised images is 5 to 6dB for Salt & Pepper and Gaussian types of noise. In case of Salt &
Pepper noise,the three transforms perform better as compared to Gaussian noise. The results presented in Figure 1, Figure 2
and Figure 3 display that the wave atom transform out performs the wavelet transform and curvelet transform, in
generating the reconstructed images.
From the analysis, it was found that denoising using the Wave atom transform reconstructs the original image
from the noisy one using lesser coefficients than denoising using the wavelet and Curvelet transforms. The Wave atom
frame transform technique was found to be theoretically simpler, faster and fewer redundant than the present techniques.
In terms of PSNR values the wave atom transform dominates Curvelet transform and Wavelet transform.
The Wave atom denoised images appear more closer to the original image than the Wavelet and Curvelet denoised images.
The Wave atom transform provides high PSNR values, irrespective of standard deviation, than the Wavelet and Curvelet
transforms in case of Salt & Pepper noise and Gaussian noise.
Table 1
S. No
S. D
1
2
3
4
5
6
0.40
0.41
0.42
0.43
0.44
0.45
PSNR in Db
Salt & Pepper Noise
Wavelets Curvelets Waveatoms
16.0277
19.3289
20.2240
15.9348
19.1431
20.1458
15.8918
18.9908
20.1358
15.8412
18.8541
20.1183
15.7953
18.6393
19.9644
15.7251
18.5012
19.9517
Table 2
S. No
S. D.
1
2
3
4
5
6
0.40
0.41
0.42
0.43
0.44
0.45
PSNR in dB
Gaussian Noise
Wavelets Curvelets Waveatoms
13.5221
15.0546
15.7575
13.2774
14.7336
15.3764
13.0242
14.4186
15.0043
12.7864
14.1072
14.6494
12.5576
13.8273
14.3271
12.3359
13.5321
13.9962
Figure 1 and Figure 2 Show the Images for Salt & Pepper Noise and Gaussian Types of Noise Respectively for
Standard Deviation (S. D.) Value of 0.45
a
www.tjprc.org
b
editor@tjprc.org
72
(c)
(d)
(e)
Figure 1: Images Corresponding to S. D. of 0.45 for Salt & Pepper Noise (a) Original Fingerprint image (b) Noisy
Image (c) Wavelet Reconstruction (d) Curvelet Reconstruction (e) Wave Atom Reconstruction
(a)
(c )
Impact Factor (JCC): 6.8785
(b)
( d)
Index Copernicus Value (ICV): 3.0
73
(e)
Figure 2: Images Corresponding to S. D. of 0.45 for Gaussian Noise (a) Original Fingerprint Image (b) Noisy Image
(c) Wavelet Reconstruction (d) Curvelet Reconstruction (e) Wave Atom Reconstruction
Figure 3 shows plots of PSNR vs SD corresponding to the Wavelet, Curvelet and Wave atom transforms, for the
Salt & Pepper and Gaussian types of noise.
editor@tjprc.org
74
REFERENCES
1.
2.
Raghuveer M. Rao and Ajit S. Bopardikar, Wavelet Transforms, Introduction to Theory and Applications.
3.
Shapiro, J. Embedded image coding using zerotrees of wavelet coefficients. IEEE Transactions on Signal
Processing December 1993; 41(12):3445-3462.
4.
Fodor, I. K. and C. Kamath. 2003. Denoising through Wavelet Shrinkage: An Empirical Study. SPIE Journal on
Electronic Imaging. Vol. 12(1):151-160.
5.
Malfaint, M. and Roose D.1997. Wavelet-based image denoising using a Markov random field a priorimodel.
IEEE Transactions on Image Processing. Vol. 6(4):549-565.
6.
E.
J.
Candes
and
D.
L.
Donoho,
Curvelets
[online]
Available:
http://www-
E. J. Candes and D. L. Donoho, Curvelets A Surprisingly Effective Nonadaptive Representation For Objects
with Edges, in Saint-Malo Proceedings
8.
9.
10. Jean-Luc Starck, Emmanuel J.Candes, and David L. Donohoo, The Curvelet Transform for Image Denoising, in
IEEE Transactions on Image Processing, June 2002.
11. R. Sivakumar, Denoising of Computer Tomography Images using Curvelet Transform, in ARPN Journal of
Engineering and Applied Sciences, February 2007.
12. J Rajeesh, R. S. Moni and S. S. Kumar performance analysis of wave atom transform in texture classification,
in Springer-verlag London limited, 15 may 2012.
AUTHORS DETAILS
Subba Reddy Borra Obtained his Bachelor of Technology (Computer Science and Engineering) from
BAPATLA ENGINEERING COLLEGE, from 1998 to 2002, and he received M. Tech (NEURAL NETWORKS) (CSE),
75
JNTK KAKINADA from 2009 to 2011 and he is doing Ph. D, JNTUH HYDERABAD. He is currently working as an
Assistant Professor in Department of Computer Science and Engineering at Dr Samuel George Institute of Engineering &
Technology Markapur, India. His areas of interest are Digital Image Processing, Machine Learning, pattern Recognition.
Date of Birth: 05-06-1978
Year of Registration: 2012
Dr .G. Jagadeeswar Reddy received B. Tech (Electronics & Communication Engg) from S.V. University in May
1995, M. Tech (Industrial Electronics) from Mysore University, S. J. C. E, Mysore in 1998, Ph. D from JNTU
HYDERABAD IN 2011. He has 17 years of teaching experience and Currently he is working as Principal, Narayana
Engineering College, Nellore, AP, India. He has Got Best paper award for the paper Lossy Compression and Curvelet
Thresholding for Image Denoising presented at International Conference on Electronic Design conducted at Malaysia in
the year 2008.He Got First Prize in the test in DSP short term course conducted by Sri Nidhi Institute of Technology,
Hyderabad in the year 2002.His areas of interest are Digital Image Processing, Digital Signal Processing,. He is Life
Member of ISTE.
Dr. E. Sreenivasa Reddy s received B. Tech(Electronics & Communication Engineering ) from Nagarjuna
University, M. S(Electronics & Control) from Birla Institute of Technology & Science, M. Tech(Computer Engg) from
Visveswaraya Technological University and Ph. D (Computer Science & Engineering ) from Acharya Nagarjuna
University, AP, India.. He has 23 years of teaching experience and Currently he is working as Principal, University College
of Engineering, Acharya Nagarjuna University, AP, India. He has published more than 20 research papers in National/
International Journals/ Conferences. His areas of research interest are Image processing, Biometrics, Security and Micro
controllers. He participate many International workshops held at South Korea, Malaysia. He is Member of IEEE, ISTE.
www.tjprc.org
editor@tjprc.org