Volume 2-issue-6-2148-2154

ISSN: 2278 – 1323
International Journal of Advanced Research in Computer Engineering & Technology (IJARCET)
Volume 2, Issue 6, June 2013
2148
www.ijarcet.org
Image Denoising using Dual-Tree Complex DWT and
Double-Density Dual-Tree Complex DWT
Mr. R. K. Sarawale1
, Dr. Mrs. S.R. Chougule1
Abstract- Non-stationary signal processing applications
use standard non-redundant DWT (Discrete Wavelet
Transform) which is very powerful tool. But it suffers
from shift sensitivity, absence of phase information, and
poor directionality. To remove out these limitations,
many researchers developed extensions to the standard
DWT such as WP (Wavelet Packet Transform), and
SWT (Stationary Wavelet Transform). These extensions
are highly redundant and computationally intensive.
Complex Wavelet Transform (CWT) is also an
impressive option, complex-valued extension to the
standard DWT. There are various applications of
Redundant CWT (RCWT) in an image processing such
as Denoising, Motion estimation, Image fusion, Edge
detection, and Texture analysis. In this work, the
focused application is the image denoising using two
innovative techniques and the images are considered
which are corrupted by a random noise. In this paper,
first two sections explain about introduction to the topic
and regarding wavelet transform domain. Third section
gives an idea about basics concepts of the system. Forth
section illustrates the proposed systems. Last section
gives results and discussion. Here promising results are
compared with DWT extensions namely, Dual-Tree
Complex DWT (DTCWT) and Double-Density Dual-
Tree Complex DWT (DDDTCWT).
Index Terms- CWT, DWT, Dual-Tree Complex DWT,
Double-Density Dual-Tree Complex DWT.
I. INTRODUCTION
Image denoising is a technique which removes out
noise which is added in the original image. Noise
reduction is an important part of image processing
systems. An image is always affected by noise. Image
quality may get disturbed while capturing, processing
and storing the image. Noise is nothing but the real
world signals and which are not part of the original
signal. In images, noise suppression is a particularly
delicate task. In this task, noise reduction and the
preservation of actual image features are the main
focusing parts.
The wavelet transform provides a multiresolution
representation using a set of analyzing functions that
are dilations and translations of a few functions
(wavelets). It overcomes some of the limitations of
the Fourier transform with its ability to represent a
function simultaneously in the frequency and time
domains using a single prototype function (or
wavelet) and its scales and shifts [2].
The wavelet transform comes in several forms. The
critically-sampled form of the wavelet transforms
provides the most compact representation; however, it
has several limitations.
It lacks the shift-invariance property, and in multiple
dimensions it does a poor job of distinguishing
orientations, which is important in image processing.
For some applications, improvements can be obtained
by using an expansive wavelet transform in place of a
critically-sampled one [1]. A denoising method is
used to improve the quality of image corrupted by a
lot of noise due to the undesired conditions for image
acquisition. The image quality is measured by the
peak signal-to-noise ratio (PSNR) or signal-to noise
ratio (SNR). Traditionally, this is achieved by linear
processing such as Wiener filtering [3]. Recently
introduced Dual-Tree Complex DWT and Double-
Density Dual-Tree Complex DWT can give best
results in image denoising applications.
II. WAVELET TRANSFORM DOMAIN
A Fourier Transform (FT) is only able to retrieve the
global frequency content of a signal, the time
information is lost.
A multi-resolution analysis becomes possible by
using wavelet analysis. The Wavelet Transform (WT)
retrieves frequency and time content of a signal. The
basic types of wavelet transform are namely, i)
Continuous Wavelet Transform (CoWT) ii) Discrete
Wavelet Transform (DWT) iii) Complex Wavelet
Transform (CWT). A multi-resolution analysis is not
possible with Fourier Transform (FT) and Short Time
Fourier Transform (STFT) and hence there is a
restriction to apply these tools in image processing
systems; particularly in image denoising applications.
The multi-resolution analysis becomes possible by
using wavelet analysis. A Continuous Wavelet
Transform (CoWT) is calculated analogous to the
Fourier transform (FT), by the convolution between
the signal and analysis function. The Discrete Wavelet
Transform uses filter banks to perform the wavelet
analysis.
A. Complex wavelet transform
This is a newly introduced technique of DWT.
Orthogonal wavelet decompositions, based on
separable, multirate filtering systems have been
widely used in image and signal processing, largely
for data compression. Kingsbury introduced a very
elegant computational structure, the Dual-Tree
complex wavelet transform [5], which displays near-
shift invariant properties. Other constructions can be
found such as in [6]. Kingsbury [3] pointed out the
problems of Mallat-type algorithms. These algorithms
have the lack of shift invariance.

ISSN: 2278 – 1323
2149
www.ijarcet.org
Complex wavelets have not been used widely in
image processing due to the difficulty in designing
complex filters which satisfy a perfect reconstruction
property. To overcome this, Kingsbury proposed a
Dual-Tree implementation of the CWT (DT CWT)
[7], which uses two trees of real filters to generate the
real and imaginary parts of the wavelet coefficients
separately. The DWT suffers from the following two
problems.
1] Lack of shift invariance - this results from the
down sampling operation at each level. When the
input signal is shifted slightly, the amplitude of the
wavelet coefficients varies so much.
2] Lack of directional selectivity - as the DWT filters
are real and separable the DWT cannot distinguish
between the opposing diagonal directions.
The first problem can be avoided if the filter outputs
from each level are not down sampled but this
increases the computational costs significantly and the
resulting undecimated wavelet transform still cannot
distinguish between opposing diagonals since the
transform is still separable.
To distinguish opposing diagonals with separable
filters the filter frequency responses are required to be
asymmetric for positive and negative frequencies. A
good way to achieve this is to use complex wavelet
filters which can be made to suppress negative
frequency components. The Complex DWT has
improved shift-invariance and directional selectivity
than the separable DWT [6]-[7].
III. BASIC CONCEPTS OF THE SYSTEM
A filter bank plays an important role in wavelet
transform applications. It consists of two banks
namely, analysis filter bank and synthesis filter bank.
The one dimensional filter bank is constructed with
analysis and synthesis filter bank which is shown in
Fig. 1.
Fig. 1 One dimensional filter bank
The analysis filter bank decomposes the input signal
x(n) into two sub band signals, c(n) and d(n). The
signal c(n) represents the low frequency part of x(n),
while the signal d(n) represents the high frequency
part of x(n). It uses filter banks to perform the wavelet
analysis. The DWT decomposes the signal into
wavelet coefficients from which the original signal
can be reconstructed again. The wavelet coefficients
represent the signal in various frequency bands. The
coefficients can be processed in several ways, giving
the DWT attractive properties over linear filtering.
A. A block schematic of wavelet based image
denoising technique
Image denoising means usually compute the soft
threshold in such a way that information present in
image is preserved. A block schematic of Wavelet
based image denoising technique is shown in Fig. 2.
Here the basic steps of wavelet based image
denoising are given below.
1. Decompose corrupted image by noise using
wavelet transform.
2. Compute threshold in wavelet domain and apply
to noisy coefficients.
3. Apply inverse wavelet transform to reconstruct
image.
Fig. 2 A block schematic of wavelet based image denoising
technique
B. Basic differences between the two DWT
extensions
The basic differences between the Dual-Tree DWT
and Double-Density DWT are given below.
1] The Dual-Tree and Double-Density DWTs are
implemented with totally different filter bank
structures.
2] The Dual-Tree DWT can be interpreted as a
complex-valued wavelet transform which is useful for
signal modeling and denoising (the Double-Density
DWT cannot be interpreted as such).
3] For the Dual-Tree DWT there are fewer degrees of
freedom for design, while for the Double-Density
DWT there are more degrees of freedom for design.
4] The Dual-Tree DWT can be used to implement
two-dimensional transforms with directional wavelets,
which is highly desirable for image processing [8].
IV. PROPOSED SYSTEMS
By introducing Complex wavelet transforms (CWT)
concept, we can achieve Dual-Tree Complex DWT
system. Also combining the Double-Density DWT
and Dual-Tree Complex DWT, we can achieve the
Double-Density Dual-Tree Complex DWT system.
Complex wavelet transforms (CWT) use complex-
valued filtering (analytic filter) that decomposes the
real/complex signals into real and imaginary parts in
transform domain. The real and imaginary

ISSN: 2278 – 1323
2150
www.ijarcet.org
coefficients are used to compute amplitude and phase
information.
A. Dual-Tree Complex DWT (DTCWT)
Kingsbury’s complex Dual-Tree DWT is based on
(approximate) Hilbert pairs of wavelets [8].
Kingsbury found that the Dual-Tree DWT is nearly
shift-invariant when the lowpass filters of one DWT
interpolate midway between the lowpass filters of the
second DWT [9].
The Dual-Tree Complex DWT can be implemented
using two critically-sampled DWTs in parallel as
shown in the Fig. 3. This transform gives 2N DWT
coefficients for an N-point signal. Hence this
transform is known as 2-times expansive.
Here the filters are designed in such a way that the
subband signals of the upper DWT can be interpreted
as the real part of a CWT and subbands signals of the
lower DWT can be interpreted as the imaginary part.
For specially designed sets of filters, the wavelet
associated with the upper DWT can be an
approximate Hilbert transform of the wavelet
associated with the lower DWT. In this manner, the
designed DTCWT is nearly shift-invariant than the
critically-sampled DWT [1]-[4].
Fig. 3 Design implementation of Dual-Tree Complex DWT
The DTCWT gives wavelets in six distinct
directions. In each direction, there are two wavelets.
In each direction, one of the two wavelets can be
interpreted as the real part and the other wavelet can
be interpreted as the imaginary part of a complex-
valued two dimensional (2D) wavelet. The DTCWT is
implemented as four critically sampled separable 2D
DWTs operating in parallel. However, different filter
sets are used along the rows and columns [1]-[4]. Fig.
4 indicates that a flowchart of Dual-Tree Complex
DWT. This illustrates the steps of implementation of
DTCWT.
Fig. 4 Flowchart of Dual-Tree Complex DWT
B. Double-Density Dual-Tree Complex DWT
(DDDTCWT)
Double-Density DWT performs superior than the
standard DWT in terms of enhancement In two
dimensions. But here all of the wavelets are not
directional. Though the Double-Density DWT utilizes
more wavelets, some lack a dominant spatial
orientation, which prevents them from being able to
isolate those directions [9]. A solution to this problem
is provided by the Double-Density Dual-Tree
Complex DWT, which combines the characteristics of
the Double-Density DWT and the Dual-Tree DWT.
The DDDTCWT is based on two scaling functions
and four distinct wavelets, each of which is
specifically designed such that the two wavelets of the
first pair are offset from one other by one half, and the
Start
Read an input image
Add noise to the input image and
compute PSNR
Use filter bank for first stage and
remaining stages
Set J (Number of stages) and T
(Threshold value)
Compute forward DTCWT
Compute inverse DTCWT
Extract output image and compute PSNR
End

ISSN: 2278 – 1323
2151
www.ijarcet.org
other pair of wavelets form an approximate Hilbert
transform pair [9]. The DDDTCWT is 4-times
expansive. It yields two wavelets in the same
dominating orientations [11].
Fig. 5 Design implementation of the Double-Density Dual-
Tree Complex DWT
For each of the directions, one of the wavelets can
be interpreted as the real part and the other can be
interpreted as the imaginary part. This transform is
implemented by applying four 2D Double-Density
DWTs in parallel with distinct filter sets for the rows
and columns. Then the sum and difference of the
subband images is carried out. This operation gives
the 32 oriented wavelets [10].
Fig. 5 shows that design implementation of Double-
Density Dual-Tree Complex DWT. The DDDTCWT
is an over complete discrete wavelet transform
(DWT) designed to simultaneously possess the
properties of the Double-Density DWT and the Dual-
Tree DWT [7]. Fig. 6 shows that a flowchart of
Double-Density Dual-Tree Complex DWT. This
illustrates the steps of implementation of DDDTCWT.
V. RESULTS AND DISCUSSION
The implementation of this work has performed in
MATLAB (7.9) software. A Table I illustrates the
results for Tooth.jpg image with T=35. A Table II
illustrates the results for Tooth.jpg image with Noise
variance of 30. The performance of both the image
denoising methods can be compared by comparing
PSNR (Peak-Signal-to-Noise Ratio) value of each
method. Here as noise increases, Double-Density
Complex DWT gives better result than Dual-Tree
Complex DWT. Fig. 7 shows that plot of PSNR
versus Noise variance. Fig. 8 shows that plot of PSNR
versus Threshold value.
Fig. 9 and Fig. 10 indicate output of DTCWT and
DDDTCWT. Fig. 9(a) shows that original image
which is corrupted by a noise with variance of 30
which is shown in Fig. 9(b). This corrupted image is
applied to DTCWT and DDDTCWT. Fig. 9(c) depicts
output denoised image using DTCWT and Fig. 9(d)
output denoised image using DDDTCWT.
Similarly Fig. 10 shows that output images for noise
variance of 50.
Fig. 6 Flowchart of Double-Density Dual-Tree Complex
DWT
VI. CONCLUSION
The newly invented extensions of the DWT perform
best in image processing applications. In this paper,
the concept focused is wavelet based image denoising
methods of an image which is corrupted by additive
Gaussian noise. The techniques used are Dual-Tree
Complex DWT and Double-Density Dual-Tree
Complex DWT. These techniques give high
performance as compared to the existing basic DWT
methods. As noise increases Double-Density Dual-
Tree Complex DWT works superior than Dual-Tree
Complex DWT. The future work will be in the form
Read an input image
Add noise to the input image and compute
PSNR
Use filter bank for first stage and
remaining stages
Set J (Number of stages) and T (Threshold
value)
Compute forward DDDTCWT
Compute inverse DDDTCWT
Extract output image and compute PSNR
End
Start

ISSN: 2278 – 1323
2152
www.ijarcet.org
of application of both the methods to audio and video
signals.
Table I Results for Tooth.jpg image with T=35
Noise
PSNR (decibel)
Noisy
Image
DTCWT DDDTCWT
20 22.1360 31.5635 31.9626
25 20.2017 30.6586 31.7062
30 18.6247 29.5062 31.2922
35 17.2793 27.9968 30.5743
40 16.1089 26.3338 29.6936
45 15.1042 24.6170 28.6060
50 14.1952 22.9787 27.3532
55 13.3578 21.3644 25.8019
60 12.6091 19.9846 24.4681
65 11.8981 18.6467 22.9719
70 11.2730 17.5295 21.6661
Table II Results for Tooth.jpg image with Noise variance of
30
Threshold
value
(T)
PSNR (decibel)
Noisy
Image
DTCWT DDDTCWT
10 18.6347 22.4953 25.0989
15 18.6138 24.6411 28.1653
20 18.6297 26.6771 30.1356
25 18.6197 27.9968 30.5743
30 18.5950 29.0488 31.2199
35 18.6260 29.4638 31.2290
40 18.6087 29.5744 31.1469
45 18.6408 29.6180 31.0597
50 18.6304 29.5533 30.9814
55 18.6234 29.4834 30.8574
60 18.6296 29.5253 30.8638
Fig. 7 Plot of PSNR versus Noise variance
Fig. 8 Plot of PSNR versus Threshold value

ISSN: 2278 – 1323
2153
www.ijarcet.org
Fig. 9 Output for noise variance = 30
(a) Original image, (b) Noisy image, (c) Denoised image by
DTCWT and (d) Denoised image by DDDTCWT
Fig. 10 Output for noise variance = 50
(a) Original image, (b) Noisy image, (c) Denoised image by
DTCWT and (d) Denoised image by DDDTCWT

ISSN: 2278 – 1323
2154
www.ijarcet.org
REFERENCES
[1] I. W. Selesnick, R. G. Baraniuk, and N. G. Kingsbury,
―The dual-tree complex wavelet transform—A coherent
framework for multiscale signal and image processing,‖
IEEE Signal Procesin. Mag., vol. 22, no. 6, pp. 123–151,
Nov. 2005.
[2] Shan Lal, Mahesh Chandra, Gopal Krishna Upadhyay, Deep
Gupta, ―Removal of Additive Gaussian Noise by Complex
Double Density Dual Tree Discrete Wavelet Transform” MIT
International Journal of Electronics and Communication
Engineering, Vol. 1, No. 1, pp. 8-16 Jan 2011.
[3] Devanand Bhonsle and Sandeepa Dewangan ―Comparative
Study of dual-tree complex wavelet transform and double
density complex wavelet transform for Image Denoising
Using Wavelet-Domain‖ International Journal of Scientific
and Research Publications, vol. 2, Issue 7, pp. 1-5,July 2012.
[4] I. Bayram and I. W. Selesnick, ―A simple construction for the
M-band dual-tree complex wavelet transform,‖ in Proc. 12th
IEEE DSP Workshop, pp. 596–601, 2006.
[5] N. G. Kingsbury, ―The dual-tree complex wavelet transform:
a new technique for shift invariance and directional filters‖,
Proceedings of the IEEE Digital Signal Processing Workshop,
1998.
[6] J. Neumann and G. Steidl, ―Dual–tree complex wavelet
transform in the frequency domain and an application to
signal classification‖, International Journal of Wavelets,
Multiresolution and Information Processing IJWMIP, 2004.
[7] N. G. Kingsbury, ―A dual-tree complex wavelet transform
with improved orthogonality and symmetry properties,‖ in
Proc. IEEE Int. Conf. Image Process. (ICIP), 2000.
[8] I. W. Selesnick, ―The design of approximate Hilbert transform
pairs of wavelet bases,‖ IEEE Transaction on Signal
Processing, vol. 50, no. 5, pp. 1144–1152, May 2002.
[9] I. W. Selesnick, ―Hilbert transform pairs of wavelet bases,‖
IEEE Signal Process. Letter, vol. 8, no. 6, pp. 170–173 Jun.
2001.
[10] G. Sandhya, K. Kishore, ―Denoising of Images corrupted by
Random noise using Complex Double Density Dual Tree
Discrete Wavelet Transform‖, International Journal of
Engineering Research and Applications, vol. 2, Issue 3,
pp.079-087, May-Jun 2012.
[11] I. W. Selesnick, ―The double-density dual-tree DWT”, IEEE
Trans. on Signal Processing, 52(5): pp. 1304-1314, May
2004.
Mr. R. K. Sarawale1
received B.E. and pursuing M.E. in
Electronics and Telecommunication in Bharati Vidyapeeth’s
College of Engg., Kolhapur.
Dr. Mrs. S. R. Chougule2
received Ph.D. in Electronics Engg.
Currently she is working as a Principal in Bharati Vidyapeeth’s
College of Engg., Kolhapur.

Volume 2-issue-6-2148-2154

More Related Content

What's hot (19)

Similar to Volume 2-issue-6-2148-2154 (20)

More from Editor IJARCET (20)

Recently uploaded (20)

Volume 2-issue-6-2148-2154