SlideShare a Scribd company logo
International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-3, Issue-2, Feb- 2017]
https://dx.doi.org/10.24001/ijaems.3.2.19 ISSN: 2454-1311
www.ijaems.com Page | 111
Mathematical Model for Image Restoration Based
on Fractional Order Total Variation
Akhtar Rasool1
, Noor Badshah2
, Asmat Ullah3
2, 1
Basic Sciences Department, University of Engineering and Technology KP Peshawar (Pak).
3
Department of Engineering Mechanics, Hohai University, No.1 Xikang Road, Nanjing, Jiangsu 210098, PR China.
Abstract—This paper addresses mathematical model for
signal restoration based on fractional order total variation
(FOTV) for multiplicative noise. In alternating minimization
algorithm the Newton method is coupled with time-marching
scheme for the solutions of the corresponding PDEs related to
the minimization of the denoising model. Results obtained
from experiments show that our model can not only reduce the
staircase effect of the restored images but also better improve
the PSNR as compare to other existed methods.
Keywords— Total Variation (TV), Fractional Order
Derivative, Speckle Noise, Image Restoration.
I. INTRODUCTION
Image denoising has various applications in different fields
including pattern recognition, remote sensing, preprocessing
for computer vision and medical images. It is still a valid
challenge for researchers. Two important types of image
denoising are additive noise and multiplicative noise [1, 2]. In
additive noise removal model, the original image 𝑢 is
considered to be contaminated by Gaussian additive noise 𝑛.
Our goal is to restore 𝑢 from g = u + n. The total variation
(TV) has been developed by Rudin, Osher and Fatemi. TV
based method is edge preserving [1]. We discuss
multiplicative noise in this paper. Consider the noisy image
𝑔: Ω → R, the original image 𝑢 is contaminated by some
noise 𝜈 satisfying the Gamma law with mean 1. This type of
noise mostly exists in ultrasound imaging, sonar and laser
imaging and synthetic aperture radar images. Multiplicative
noise are more complex as compared to additive noise. Rudin
et al early approached a variational model [3]. Second
approach for multiplicative noise has been proposed by
Aubert and Aujol, see [4] for more details. Keeping in mind
the statistical properties of multiplicative noise 𝑛, we
approximate 𝑔 by treating the following constrained
minimization problem
min
𝑢
∫ |𝐷𝑢|Ω
𝑑𝑥𝑑𝑦 (1)
Subject to: ∫
𝑔
𝑢
𝑑𝑥𝑑𝑦 = 1,Ω
∫ (
𝑔
𝑢
− 1)
2
𝑑𝑥𝑑𝑦 = 𝜎2
.Ω
The mean of noise is 1 with standard deviation 𝜎2
as listed
above. TV regularization methods remove noise and
simultaneously preserve sharp intensity boundaries but
sometimes textures are smoothed with noise in noise removing
process [1-3]. The staircase effects can be reduced by
generalizing the regularization term which can be performed in
two ways. First approach is based on high order derivative; see
[5-7].
Secondly, the regularization term can be generalized by using
fractional-order derivatives. In recent past, some general
fractional-order additive noise models have been treated in
[8,9]. Chambolle used the fitting term ||𝑔 − 𝑢||2
2
in ROF
model and hence introduced an efficient projection method
for additive noise [10]. Obtained results indicate that the
performance of fractional-order TV are better than integer high
order TV. Fractional order models have competitive
performance with other integer order based models in noise
removing as well as staircase reduction. Moreover Aubert and
Aujol developed a multiplicative noise model “AA-model”,
which is given by
min
𝑢
∫ (𝑙𝑜𝑔𝑢 +
𝑔
𝑢
) 𝑑𝑥𝑑𝑦 + 𝜆 ∫ |𝐷𝑢|𝑑𝑥𝑑𝑦ΩΩ
(2)
Recently Zhang Jun and Wei Zhihui introduced fractional
order AA Model by replacing TV term in AA model with
fractional order TV.
As the function (logu +
g
u
) is not convex everywhere,
therefore Haung et al [11] introduced the auxiliary variable
w = logu and introduced a strictly convex function
𝑚𝑖𝑛 𝑤,𝑢 ∫ (𝑤 + 𝑔𝑒−𝑤)Ω
𝑑𝑥𝑑𝑦 + 𝛼1|𝑤 − 𝑢| 𝐿2
2
+
𝛼2 ∫ |𝐷𝑢|𝑑𝑥𝑑𝑦Ω
, (3)
with two regularization parameters 𝛼1 and 𝛼2. As 𝑢 = 𝑒−𝑤
is
strictly convex for all value of 𝑤 which ensures that the
solution of the variational problem is unique. TV based
models have better performance in preserve sharp edges,
reducing the noise but often these models cause the staircase
effect. We develop fractional-order TV based method for
multiplicative noise in Sect. 2. In Sect. 3, we use alternating
minimization algorithm to find minimizer of our proposed
problem .In Sect. 4, we give some numerical results to validate
International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-3, Issue-2, Feb- 2017]
https://dx.doi.org/10.24001/ijaems.3.2.19 ISSN: 2454-1311
www.ijaems.com Page | 112
the performance of the proposed methodology. Finally, the
paper is concluded in the last Section.
II. THE PROPOSED APPROACH
We introduce a convex function for restoring images
contaminated by multiplicative noise in this section. Our new
model is based on fractional Order AA model and HNW
algorithm. Replacing TV term in model (3) with Fractional
order TV, we introduce the following model.
𝑚𝑖𝑛 𝑤,𝑢 ∫ (𝑤 + 𝑔𝑒−𝑤
)𝑑𝑥𝑑𝑦
Ω
+𝛼1||𝑤 − 𝑢|| 𝐿2
2
+ (𝛼2 + 𝛼3 𝑢2
)|𝐷 𝛼
𝑢|, (4)
with three positive regularization parameters 𝛼1, 𝛼2 and 𝛼3.
The fractional-order total variation norm provides a better
performance in preserving image edges. It also reduces
staircase effect. Moreover 𝑢 and 𝑤 preserve edges at the same
point; see [9]. The use of fractional-order TV to 𝑤 is quite
better for preserving sharp edges and reducing staircase effect.
The main advantage is that u = ew
is positive for all 𝑤, even
though 𝑢 in the objective function (2) required to be positive.
In our proposed model if we assign 𝑤 a large negative number,
𝑢 is still a small non zero positive number.
III. NUMERICAL SCHEME
We use an alternating minimization algorithm to solve the
FOTV problem. We split eqn. (4) in the following two
subproblems.
(i) First fix 𝑢 and find the solution of
min
w
∫ (w + ge−w
)dxdyΩ
+ α1||𝑤 − 𝑢||L2
2
. (5)
(ii) Then fix 𝑤, find the solution of
min
u
α1||w − u||L2
2
+(α2 + α3u2) ∫ |Dα
u|dxdy.Ω
(6)
Applying discrete setting, the first problem gets the form:
min
w
∑ (𝑤𝑖 + 𝑔𝑒−𝑤 𝑖) + 𝛼1||𝑤 − 𝑢||2
2
.𝑛2
1 (7)
The minimizer of the above problem can be obtained by
solving the following nonlinear system
1 − 𝑔𝑒−𝑤 𝑖 + 2𝛼1(𝑤𝑖 − 𝑢𝑖) = 0, 𝑖 = 1,2, . . . 𝑛2
. (8)
The 2nd derivative w.r.t 𝑤 of the first term in (5) is 𝑔𝑒−𝑤
,
which is positive for positive value of 𝑔. Thus the strictly
convex function (5) has a unique minimum which we compute
with Newton method as
wi+1 = wi −
1 − ge−wi + 2α1(wi − u)
ge−wi + 2α1
, i = 0,1,2, . ..
The Euler-Lagrange equation of FOTV model (6) is
𝜕𝑢
𝜕𝑡
=
2𝛼1
𝛼2+𝛼3 𝑢2 (𝑤𝑖,𝑗
𝑛
− 𝑢𝑖,𝑗) + (−1) 𝛼
𝑑𝑖𝑣 𝛼 ∇ 𝛼 𝑢 𝑖,𝑗
|∇ 𝛼 𝑢 𝑖,𝑗|
(9)
𝜕𝑢
𝜕𝑛
= 0 𝑜𝑛 𝜕Ω, u(0) = u0.
The numerical scheme for the steady-state equation (9) is
given by
𝑢𝑖,𝑗
𝑛+1
= 𝑢𝑖,𝑗
𝑛
+ 𝜂𝜏|𝑤𝑖,𝑗 − 𝑢𝑖,𝑗| + (−1) 𝛼
𝑑𝑖𝑣 𝛼
∇ 𝛼
𝑢𝑖,𝑗
|∇ 𝛼 𝑢𝑖,𝑗|
,
𝜂 =
2𝛼1
𝛼2+𝛼3 𝑢2 with B. Conditions
𝑢0,𝑗
𝑛
= 𝑢1,𝑗
𝑛
, 𝑢 𝑁,𝑗
𝑛
= 𝑢 𝑁−1,𝑗
𝑛
,
𝑢𝑖,0
𝑛
= 𝑢𝑖,1
𝑛
, 𝑢𝑖,𝑁
𝑛
= 𝑢𝑖,𝑁−1
𝑛
.
Algorithm
We take initial guess 𝑢(0)
to be the noisy image.
(i) First fix𝑢(𝑘−1)
, compute by Newton's algorithm 𝑤(𝑘)
=
arg min
w
∑ (𝑤𝑖 + 𝑔𝑛2
𝑘=0 𝑒−𝑤 𝑖) + 𝛼1||𝑤 − 𝑢 𝑘−1||2
2
(ii) Then fix 𝑤(𝑘)
, compute 𝑢(𝑘)
by using
𝜕𝑢
𝜕𝑡
=
2𝛼1
𝛼2 + 𝛼3 𝑢2
||𝑤(𝑘)
− 𝑢|| + (−1) 𝛼
𝑑𝑖𝑣 𝛼
(
∇ 𝛼
𝑢
|∇ 𝛼 𝑢|
)
(iii) Exit with an approximate denoised image 𝑢 𝑘
= 𝑒 𝑤(𝑘)
if
the stopping criteria
||𝑢 𝑘+1−𝑢 𝑘||
||𝑢 𝑘+1||
2
< 10−4
is satisfied, otherwise
we continue with 𝑘 = 𝑘 + 1 for next iteration.
IV. NUMERICAL RESULTS
This section describes some experimental results of image
denoising to claim the superiority of our method. We consider
some images contaminated by multiplicative noise. Figure 1
shows clean images. Results obtained from experiments
indicate the superiority of our proposed method. We use
PSNR to compare the capabilities of the restored images as
𝑃𝑆𝑁𝑅 = 20 log10
255
||𝑢 − 𝑢 𝑡𝑟𝑢𝑒||
Where 𝑢 𝑡𝑟𝑢𝑒 and 𝑢 denote the noise-free signal and recovered
signal respectively.
International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-3, Issue-2, Feb- 2017]
https://dx.doi.org/10.24001/ijaems.3.2.19 ISSN: 2454-1311
www.ijaems.com Page | 113
Original Images
Fig. 1: Face, Peppers, Barbara
(a) Face 0.3 (b) AA Method (c) Proposed Method
Fig.2: (a). Noisy Image, 𝜎2
= 0.3 (b) Restored signal “AA-method”, 𝜆 = 35 (c) Restored signal “Proposed method”, 𝛼1 =
190, 𝛼2 = 0.00002 𝑎3 = 490.
(a) Face, 𝜎 = 0.1(b) AA-Method (c) Proposed Method
Fig 3 (a) Noisy Signal, 𝜎2
= 0.1(b) Restored Signal, AA-Method, 𝜆 = 35 (c) Restored signal “Proposed method”, 𝛼1 =
190, 𝛼2 = 0.00002 𝑎3 = 490.
(a) Noisy Image, 𝜎 = 0.03 (b) AA Method (c) Proposed Method
Fig.4:(a) Noisy Signal, 𝜎2
= 0.03. (b) Restored Signal, AA-Method, 𝜆 = 35 (c) Proposed Method, 𝛼1 = 16, 𝛼2 = 0.0002, 𝛼3 =
150.
International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-3, Issue-2, Feb- 2017]
https://dx.doi.org/10.24001/ijaems.3.2.19 ISSN: 2454-1311
www.ijaems.com Page | 114
Test 1: (a) Noisy Image, 𝜎 = 0.03 (b) AA Method (c) Proposed Method
Fig.5: (a) Noisy Signal,𝜎2
= 0.01. (b) Restored Signal, AA-Method, 𝜆 = 34 (c) Proposed Method,𝛼1 = 12, 𝛼2 = 0.0002, 𝛼3 =
490.
Test 2.
Figure (a) Noisy Signal (b) AA Method (c) Proposed Method
Fig.7: (a) Noisy Signal,𝜎2
= 0.1. (b) Restored Signal, AA-Method, 𝜆 = 35 (c) Restored Signal, Proposed Method, 𝛼1 = 33, 𝛼2 =
0.00001, 𝛼3 = 600.
(a) Noisy Image (b) AA-Method (c) Proposed Method
Fig.8: (a) Noisy Signal,𝜎2
= 0.03. (b) Restored Signal, AA-Mode, 𝜆 = 35 (c) Proposed Method,𝛼1 = 200, 𝛼2 = 0.0002, 𝛼3 =
600.
(a) Noisy Image (b) AA-Method (c) Proposed Method
Fig.9: (a) Noisy Signal,𝜎2
= 0.01. (b) Restored Signal, AA-Method, 𝜆 = 35 (c) Restored Signal, Proposed Method with 𝛼1 =
200, 𝛼2 = 0.00001, 𝛼3 = 600.
International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-3, Issue-2, Feb- 2017]
https://dx.doi.org/10.24001/ijaems.3.2.19 ISSN: 2454-1311
www.ijaems.com Page | 115
Test 3.
(a)Barbara, 0.03 (b) AA Method (c) Proposed Method
Fig.10: (a) Noisy Signal, 𝜎2
= 0.03 (b) Restored Signal, AA-Method, 𝜆 = 35 (c) Restored Signal, Proposed Method with 𝛼1 =
450, 𝛼2 = 0.0001, 𝛼3 = 650.
(a) Barbara, 0.1 (b) AA Method (c) Proposed Method
Fig.11: (a) Noisy Signal, 𝜎2
= 0.1 (b) Restored Signal, AA Method, 𝜆 = 33 (c) Restored Signal, Proposed Method with 𝛼1 =
450, 𝛼2 = 0.00001, 𝛼3 = 650.
Table 1: Restoration results for face images
Image
Variance
"𝜎2
"
Fractional Order
𝛼
AA-Model
PSNR
Proposed Model
PSNR
Face
Size(256 × 256)
0.3 1.9 26.035 27.193
0.1 1.1 28.525 29.009
0.03 1.9 31.934 33.340
0.01 1.9 35.037 36.420
Table 2: Restoration results for Peppers images
Image
Variance
"𝜎2
"
Fractional Order
𝛼
AA-Model
PSNR
Proposed Model
PSNR
Peppers
Size(256 × 256)
0.1 1.1 26.152 26.301
0.03 1.1 29.228 29.355
0.01 1.1 32.142 32.201
Table 3: Restoration Results for Barbara Images.
Image
Variance
"𝜎2
"
Fractional Order
𝛼
AA-Model
PSNR
Proposed Model
PSNR
Barbara
Size(256 × 256)
0.03 1.1 28.162 28.211
0.1 1.1 25.821 25.910
In first experiment we test two dimensional signal “face”
256 × 256. We have compared the denoising capabilities of
the proposed algorithm with fractional order AA method.
Numerous values of 𝜆 have been tested for fractional order AA
algorithm and the best amongst them are mentioned in this
paper. As the regularization parameters perform well in
denoising so we have chosen the three regularization
parameters for increasing the PSNR. In Table 1 we have
mentioned the highest PSNR of fractional order AA method
and the proposed method. Peppers image of size 256 × 256 is
International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-3, Issue-2, Feb- 2017]
https://dx.doi.org/10.24001/ijaems.3.2.19 ISSN: 2454-1311
www.ijaems.com Page | 116
used to make the second experiment. Best results are chosen
among the obtained results of the performed experiements
through fractional order AA method. As the quality of the
restored image depends upon the regularization parameters so
we have chosen the suitable values for regularization
parameters. In table 2 we have shown the comparison of our
results with those obtained from fractional order AA
Algorithm. The PSNR values of the recovered peppers images
are listed in table 2. The values mentioned in table 2 clearly
shows that the PSNR values of the proposed method are
bigger than fractional order AA-Method. In third test we
consider the barbara image. The PSNR values shown in table
3 showing the better performance of the proposed algorithm.
V. CONCLUSION
This paper describes a new mathematical approach for image
denoising with fractional-order TV. Coupling the Newton
algorithmwith time- marching scheme for the solutions of
PDEs related to the minmization of denoising model based on
fractional order total variation, gives good results.The
proposed method can suppress noise very well while it can
preserve details of the recovered image. Results obtained from
experiements show that the efficiency of the proposed
approach are better than what obtained from other existed total
variation restoration methods.
REFERENCES
[1] Rudin L, Osher S, and Fatemi E, “Nonlinear Total
Variation Based Model Noise Removal Algorithm,”J.
Phys, 60:259-268(1992).
[2] A. Chambolle and P.-L. Lions, “Image Recovery via total
Variation Minimization and Related Problems,”
Numerische Mathematik, 76(2):167-188(1997).
[3] L. Rudin, P.-L. Lions, and S. Osher, “Multiplicative
Denoising and Deblurring: Theory and Algorithms,” in
Geometric Level Sets in Imaging, Vision and Graphics, S.
Osher and N. Paragios, Eds., Springer, New York, NY,
USA,103-119(2003).
[4] G. Aubert, J.-F. Aujol, “A Variational Approach to
Removing Multiplicative Noise,” SIAM Journal on
Applied Mathematics, 68(4):925-946(2008).
[5] You Y. L, Kaveh M.,“Fourth-Order Partial Differential
Equations for Noise Removal,”IEEE Transactions on
Image Processing, 9(10): 1723-1730(2000).
[6] Didas S., Weicket J. and burgeth B,“Properties of Higher
Order Nonlinear Diffusion Filtering,” Journal of
Mathematical Imaging and Vision, 35(3):208-226(2009).
[7] Lysaker M., Lundrevold A., Tai X.C., “Noise Removal
Using Fourth Order Partial Differential Equation with
Applications to Medical Magnetic Resonance Images in
Space and Time,” IEEE Transactions on Image
Processing, 12(12): 1579-1590(2003).
[8] Bai J., Feng X. C.,“Fractional-Order Anisotropic
Diffusion for Image Denoising,”IEEE Transactions on
Image Processing, 16(10): 2492-2502(2007).
[9] Zhang J. and Wei Z. H., “A Class of Fractional-Order
Multi-scale Variational Models and Alternating
Projection Algorithm for Image Denoising,” Applied
Mathematical Modeling, 35(5): 2516-2528(2011).
[10] Chambolle A., “An algorithm for Total Variation
Minimization and Applications,”Journal of Mathematical
Imaging and Vision, 20(1): 89-97(2004).
[11] Y.-M. Haung, M. K. Ng, and Y.-W. Wen, “A New Total
Variation Method for Multiplicative Noise Removal”,
SIAM Journal on Imaging Sciences, 2(1):20-40(2009).

More Related Content

What's hot

Digital image processing
Digital image processing  Digital image processing
Digital image processing
kavitha muneeshwaran
 
Robust Image Denoising in RKHS via Orthogonal Matching Pursuit
Robust Image Denoising in RKHS via Orthogonal Matching PursuitRobust Image Denoising in RKHS via Orthogonal Matching Pursuit
Robust Image Denoising in RKHS via Orthogonal Matching Pursuit
Pantelis Bouboulis
 
Image Restitution Using Non-Locally Centralized Sparse Representation Model
Image Restitution Using Non-Locally Centralized Sparse Representation ModelImage Restitution Using Non-Locally Centralized Sparse Representation Model
Image Restitution Using Non-Locally Centralized Sparse Representation Model
IJERA Editor
 
Image Degradation & Resoration
Image Degradation & ResorationImage Degradation & Resoration
Image Degradation & Resoration
Sanjay Saha
 
Bayesian modelling and computation for Raman spectroscopy
Bayesian modelling and computation for Raman spectroscopyBayesian modelling and computation for Raman spectroscopy
Bayesian modelling and computation for Raman spectroscopy
Matt Moores
 
Image restoration recent_advances_and_applications
Image restoration recent_advances_and_applicationsImage restoration recent_advances_and_applications
Image restoration recent_advances_and_applications
M Pardo
 
Chapter 3 Image Processing: Basic Transformation
Chapter 3 Image Processing:  Basic TransformationChapter 3 Image Processing:  Basic Transformation
Chapter 3 Image Processing: Basic Transformation
Varun Ojha
 
Image sampling and quantization
Image sampling and quantizationImage sampling and quantization
Image sampling and quantization
BCET, Balasore
 
Free vibration analysis of composite plates with uncertain properties
Free vibration analysis of composite plates  with uncertain propertiesFree vibration analysis of composite plates  with uncertain properties
Free vibration analysis of composite plates with uncertain properties
University of Glasgow
 
Image Acquisition and Representation
Image Acquisition and RepresentationImage Acquisition and Representation
Image Acquisition and Representation
Amnaakhaan
 
Image Enhancement
Image Enhancement Image Enhancement
Image Enhancement
Deven Sahu
 
Image pre processing-restoration
Image pre processing-restorationImage pre processing-restoration
Image pre processing-restorationAshish Kumar
 
A MODIFIED DIRECTIONAL WEIGHTED CASCADED-MASK MEDIAN FILTER FOR REMOVAL OF RA...
A MODIFIED DIRECTIONAL WEIGHTED CASCADED-MASK MEDIAN FILTER FOR REMOVAL OF RA...A MODIFIED DIRECTIONAL WEIGHTED CASCADED-MASK MEDIAN FILTER FOR REMOVAL OF RA...
A MODIFIED DIRECTIONAL WEIGHTED CASCADED-MASK MEDIAN FILTER FOR REMOVAL OF RA...
cscpconf
 
Intro to ABC
Intro to ABCIntro to ABC
Intro to ABC
Matt Moores
 
Chapter 4 Image Processing: Image Transformation
Chapter 4 Image Processing: Image TransformationChapter 4 Image Processing: Image Transformation
Chapter 4 Image Processing: Image Transformation
Varun Ojha
 
Final PhD Seminar
Final PhD SeminarFinal PhD Seminar
Final PhD SeminarMatt Moores
 
2017-03, ICASSP, Projection-based Dual Averaging for Stochastic Sparse Optimi...
2017-03, ICASSP, Projection-based Dual Averaging for Stochastic Sparse Optimi...2017-03, ICASSP, Projection-based Dual Averaging for Stochastic Sparse Optimi...
2017-03, ICASSP, Projection-based Dual Averaging for Stochastic Sparse Optimi...
asahiushio1
 
Precomputation for SMC-ABC with undirected graphical models
Precomputation for SMC-ABC with undirected graphical modelsPrecomputation for SMC-ABC with undirected graphical models
Precomputation for SMC-ABC with undirected graphical models
Matt Moores
 
imageCorrectionLinearDiffusion
imageCorrectionLinearDiffusionimageCorrectionLinearDiffusion
imageCorrectionLinearDiffusionKellen Betts
 
2017-12, Keio University, Projection-based Regularized Dual Averaging for Sto...
2017-12, Keio University, Projection-based Regularized Dual Averaging for Sto...2017-12, Keio University, Projection-based Regularized Dual Averaging for Sto...
2017-12, Keio University, Projection-based Regularized Dual Averaging for Sto...
asahiushio1
 

What's hot (20)

Digital image processing
Digital image processing  Digital image processing
Digital image processing
 
Robust Image Denoising in RKHS via Orthogonal Matching Pursuit
Robust Image Denoising in RKHS via Orthogonal Matching PursuitRobust Image Denoising in RKHS via Orthogonal Matching Pursuit
Robust Image Denoising in RKHS via Orthogonal Matching Pursuit
 
Image Restitution Using Non-Locally Centralized Sparse Representation Model
Image Restitution Using Non-Locally Centralized Sparse Representation ModelImage Restitution Using Non-Locally Centralized Sparse Representation Model
Image Restitution Using Non-Locally Centralized Sparse Representation Model
 
Image Degradation & Resoration
Image Degradation & ResorationImage Degradation & Resoration
Image Degradation & Resoration
 
Bayesian modelling and computation for Raman spectroscopy
Bayesian modelling and computation for Raman spectroscopyBayesian modelling and computation for Raman spectroscopy
Bayesian modelling and computation for Raman spectroscopy
 
Image restoration recent_advances_and_applications
Image restoration recent_advances_and_applicationsImage restoration recent_advances_and_applications
Image restoration recent_advances_and_applications
 
Chapter 3 Image Processing: Basic Transformation
Chapter 3 Image Processing:  Basic TransformationChapter 3 Image Processing:  Basic Transformation
Chapter 3 Image Processing: Basic Transformation
 
Image sampling and quantization
Image sampling and quantizationImage sampling and quantization
Image sampling and quantization
 
Free vibration analysis of composite plates with uncertain properties
Free vibration analysis of composite plates  with uncertain propertiesFree vibration analysis of composite plates  with uncertain properties
Free vibration analysis of composite plates with uncertain properties
 
Image Acquisition and Representation
Image Acquisition and RepresentationImage Acquisition and Representation
Image Acquisition and Representation
 
Image Enhancement
Image Enhancement Image Enhancement
Image Enhancement
 
Image pre processing-restoration
Image pre processing-restorationImage pre processing-restoration
Image pre processing-restoration
 
A MODIFIED DIRECTIONAL WEIGHTED CASCADED-MASK MEDIAN FILTER FOR REMOVAL OF RA...
A MODIFIED DIRECTIONAL WEIGHTED CASCADED-MASK MEDIAN FILTER FOR REMOVAL OF RA...A MODIFIED DIRECTIONAL WEIGHTED CASCADED-MASK MEDIAN FILTER FOR REMOVAL OF RA...
A MODIFIED DIRECTIONAL WEIGHTED CASCADED-MASK MEDIAN FILTER FOR REMOVAL OF RA...
 
Intro to ABC
Intro to ABCIntro to ABC
Intro to ABC
 
Chapter 4 Image Processing: Image Transformation
Chapter 4 Image Processing: Image TransformationChapter 4 Image Processing: Image Transformation
Chapter 4 Image Processing: Image Transformation
 
Final PhD Seminar
Final PhD SeminarFinal PhD Seminar
Final PhD Seminar
 
2017-03, ICASSP, Projection-based Dual Averaging for Stochastic Sparse Optimi...
2017-03, ICASSP, Projection-based Dual Averaging for Stochastic Sparse Optimi...2017-03, ICASSP, Projection-based Dual Averaging for Stochastic Sparse Optimi...
2017-03, ICASSP, Projection-based Dual Averaging for Stochastic Sparse Optimi...
 
Precomputation for SMC-ABC with undirected graphical models
Precomputation for SMC-ABC with undirected graphical modelsPrecomputation for SMC-ABC with undirected graphical models
Precomputation for SMC-ABC with undirected graphical models
 
imageCorrectionLinearDiffusion
imageCorrectionLinearDiffusionimageCorrectionLinearDiffusion
imageCorrectionLinearDiffusion
 
2017-12, Keio University, Projection-based Regularized Dual Averaging for Sto...
2017-12, Keio University, Projection-based Regularized Dual Averaging for Sto...2017-12, Keio University, Projection-based Regularized Dual Averaging for Sto...
2017-12, Keio University, Projection-based Regularized Dual Averaging for Sto...
 

Similar to mathematical model for image restoration based on fractional order total variation

A Four Directions Variational Method For Solving Image Processing Problems
A Four Directions Variational Method For Solving Image Processing ProblemsA Four Directions Variational Method For Solving Image Processing Problems
A Four Directions Variational Method For Solving Image Processing Problems
Claudia Acosta
 
Df4201720724
Df4201720724Df4201720724
Df4201720724
IJERA Editor
 
Image denoising
Image denoisingImage denoising
Image denoising
Abdur Rehman
 
Gi2429352937
Gi2429352937Gi2429352937
Gi2429352937IJMER
 
OPTIMIZATION OF COST 231 MODEL FOR 3G WIRELESS COMMUNICATION SIGNAL IN SUBURB...
OPTIMIZATION OF COST 231 MODEL FOR 3G WIRELESS COMMUNICATION SIGNAL IN SUBURB...OPTIMIZATION OF COST 231 MODEL FOR 3G WIRELESS COMMUNICATION SIGNAL IN SUBURB...
OPTIMIZATION OF COST 231 MODEL FOR 3G WIRELESS COMMUNICATION SIGNAL IN SUBURB...
Onyebuchi nosiri
 
OPTIMIZATION OF COST 231 MODEL FOR 3G WIRELESS COMMUNICATION SIGNAL IN SUBURB...
OPTIMIZATION OF COST 231 MODEL FOR 3G WIRELESS COMMUNICATION SIGNAL IN SUBURB...OPTIMIZATION OF COST 231 MODEL FOR 3G WIRELESS COMMUNICATION SIGNAL IN SUBURB...
OPTIMIZATION OF COST 231 MODEL FOR 3G WIRELESS COMMUNICATION SIGNAL IN SUBURB...
Onyebuchi nosiri
 
An adaptive method for noise removal from real world images
An adaptive method for noise removal from real world imagesAn adaptive method for noise removal from real world images
An adaptive method for noise removal from real world imagesIAEME Publication
 
Fk35963966
Fk35963966Fk35963966
Fk35963966
IJERA Editor
 
An enhanced fletcher-reeves-like conjugate gradient methods for image restora...
An enhanced fletcher-reeves-like conjugate gradient methods for image restora...An enhanced fletcher-reeves-like conjugate gradient methods for image restora...
An enhanced fletcher-reeves-like conjugate gradient methods for image restora...
IJECEIAES
 
Speckle noise reduction using hybrid tmav based fuzzy filter
Speckle noise reduction using hybrid tmav based fuzzy filterSpeckle noise reduction using hybrid tmav based fuzzy filter
Speckle noise reduction using hybrid tmav based fuzzy filter
eSAT Publishing House
 
Implicit two step adam moulton hybrid block method with two off step points f...
Implicit two step adam moulton hybrid block method with two off step points f...Implicit two step adam moulton hybrid block method with two off step points f...
Implicit two step adam moulton hybrid block method with two off step points f...
Alexander Decker
 
Performance of MMSE Denoise Signal Using LS-MMSE Technique
Performance of MMSE Denoise Signal Using LS-MMSE TechniquePerformance of MMSE Denoise Signal Using LS-MMSE Technique
Performance of MMSE Denoise Signal Using LS-MMSE Technique
IJMER
 
C0363016021
C0363016021C0363016021
C0363016021
theijes
 
SinGAN for Image Denoising
SinGAN for Image DenoisingSinGAN for Image Denoising
SinGAN for Image Denoising
KhalilBergaoui
 
A New Approach for Speech Enhancement Based On Eigenvalue Spectral Subtraction
A New Approach for Speech Enhancement Based On Eigenvalue Spectral SubtractionA New Approach for Speech Enhancement Based On Eigenvalue Spectral Subtraction
A New Approach for Speech Enhancement Based On Eigenvalue Spectral Subtraction
CSCJournals
 
Improving the Efficiency of Spectral Subtraction Method by Combining it with ...
Improving the Efficiency of Spectral Subtraction Method by Combining it with ...Improving the Efficiency of Spectral Subtraction Method by Combining it with ...
Improving the Efficiency of Spectral Subtraction Method by Combining it with ...
IJORCS
 
Numerical Solution Of Delay Differential Equations Using The Adomian Decompos...
Numerical Solution Of Delay Differential Equations Using The Adomian Decompos...Numerical Solution Of Delay Differential Equations Using The Adomian Decompos...
Numerical Solution Of Delay Differential Equations Using The Adomian Decompos...
theijes
 
International journal of engineering and mathematical modelling vol2 no1_2015_1
International journal of engineering and mathematical modelling vol2 no1_2015_1International journal of engineering and mathematical modelling vol2 no1_2015_1
International journal of engineering and mathematical modelling vol2 no1_2015_1
IJEMM
 

Similar to mathematical model for image restoration based on fractional order total variation (20)

A Four Directions Variational Method For Solving Image Processing Problems
A Four Directions Variational Method For Solving Image Processing ProblemsA Four Directions Variational Method For Solving Image Processing Problems
A Four Directions Variational Method For Solving Image Processing Problems
 
Df4201720724
Df4201720724Df4201720724
Df4201720724
 
Image denoising
Image denoisingImage denoising
Image denoising
 
1873 1878
1873 18781873 1878
1873 1878
 
1873 1878
1873 18781873 1878
1873 1878
 
Gi2429352937
Gi2429352937Gi2429352937
Gi2429352937
 
OPTIMIZATION OF COST 231 MODEL FOR 3G WIRELESS COMMUNICATION SIGNAL IN SUBURB...
OPTIMIZATION OF COST 231 MODEL FOR 3G WIRELESS COMMUNICATION SIGNAL IN SUBURB...OPTIMIZATION OF COST 231 MODEL FOR 3G WIRELESS COMMUNICATION SIGNAL IN SUBURB...
OPTIMIZATION OF COST 231 MODEL FOR 3G WIRELESS COMMUNICATION SIGNAL IN SUBURB...
 
OPTIMIZATION OF COST 231 MODEL FOR 3G WIRELESS COMMUNICATION SIGNAL IN SUBURB...
OPTIMIZATION OF COST 231 MODEL FOR 3G WIRELESS COMMUNICATION SIGNAL IN SUBURB...OPTIMIZATION OF COST 231 MODEL FOR 3G WIRELESS COMMUNICATION SIGNAL IN SUBURB...
OPTIMIZATION OF COST 231 MODEL FOR 3G WIRELESS COMMUNICATION SIGNAL IN SUBURB...
 
An adaptive method for noise removal from real world images
An adaptive method for noise removal from real world imagesAn adaptive method for noise removal from real world images
An adaptive method for noise removal from real world images
 
Fk35963966
Fk35963966Fk35963966
Fk35963966
 
An enhanced fletcher-reeves-like conjugate gradient methods for image restora...
An enhanced fletcher-reeves-like conjugate gradient methods for image restora...An enhanced fletcher-reeves-like conjugate gradient methods for image restora...
An enhanced fletcher-reeves-like conjugate gradient methods for image restora...
 
Speckle noise reduction using hybrid tmav based fuzzy filter
Speckle noise reduction using hybrid tmav based fuzzy filterSpeckle noise reduction using hybrid tmav based fuzzy filter
Speckle noise reduction using hybrid tmav based fuzzy filter
 
Implicit two step adam moulton hybrid block method with two off step points f...
Implicit two step adam moulton hybrid block method with two off step points f...Implicit two step adam moulton hybrid block method with two off step points f...
Implicit two step adam moulton hybrid block method with two off step points f...
 
Performance of MMSE Denoise Signal Using LS-MMSE Technique
Performance of MMSE Denoise Signal Using LS-MMSE TechniquePerformance of MMSE Denoise Signal Using LS-MMSE Technique
Performance of MMSE Denoise Signal Using LS-MMSE Technique
 
C0363016021
C0363016021C0363016021
C0363016021
 
SinGAN for Image Denoising
SinGAN for Image DenoisingSinGAN for Image Denoising
SinGAN for Image Denoising
 
A New Approach for Speech Enhancement Based On Eigenvalue Spectral Subtraction
A New Approach for Speech Enhancement Based On Eigenvalue Spectral SubtractionA New Approach for Speech Enhancement Based On Eigenvalue Spectral Subtraction
A New Approach for Speech Enhancement Based On Eigenvalue Spectral Subtraction
 
Improving the Efficiency of Spectral Subtraction Method by Combining it with ...
Improving the Efficiency of Spectral Subtraction Method by Combining it with ...Improving the Efficiency of Spectral Subtraction Method by Combining it with ...
Improving the Efficiency of Spectral Subtraction Method by Combining it with ...
 
Numerical Solution Of Delay Differential Equations Using The Adomian Decompos...
Numerical Solution Of Delay Differential Equations Using The Adomian Decompos...Numerical Solution Of Delay Differential Equations Using The Adomian Decompos...
Numerical Solution Of Delay Differential Equations Using The Adomian Decompos...
 
International journal of engineering and mathematical modelling vol2 no1_2015_1
International journal of engineering and mathematical modelling vol2 no1_2015_1International journal of engineering and mathematical modelling vol2 no1_2015_1
International journal of engineering and mathematical modelling vol2 no1_2015_1
 

Recently uploaded

J.Yang, ICLR 2024, MLILAB, KAIST AI.pdf
J.Yang,  ICLR 2024, MLILAB, KAIST AI.pdfJ.Yang,  ICLR 2024, MLILAB, KAIST AI.pdf
J.Yang, ICLR 2024, MLILAB, KAIST AI.pdf
MLILAB
 
Governing Equations for Fundamental Aerodynamics_Anderson2010.pdf
Governing Equations for Fundamental Aerodynamics_Anderson2010.pdfGoverning Equations for Fundamental Aerodynamics_Anderson2010.pdf
Governing Equations for Fundamental Aerodynamics_Anderson2010.pdf
WENKENLI1
 
Immunizing Image Classifiers Against Localized Adversary Attacks
Immunizing Image Classifiers Against Localized Adversary AttacksImmunizing Image Classifiers Against Localized Adversary Attacks
Immunizing Image Classifiers Against Localized Adversary Attacks
gerogepatton
 
Final project report on grocery store management system..pdf
Final project report on grocery store management system..pdfFinal project report on grocery store management system..pdf
Final project report on grocery store management system..pdf
Kamal Acharya
 
Gen AI Study Jams _ For the GDSC Leads in India.pdf
Gen AI Study Jams _ For the GDSC Leads in India.pdfGen AI Study Jams _ For the GDSC Leads in India.pdf
Gen AI Study Jams _ For the GDSC Leads in India.pdf
gdsczhcet
 
ML for identifying fraud using open blockchain data.pptx
ML for identifying fraud using open blockchain data.pptxML for identifying fraud using open blockchain data.pptx
ML for identifying fraud using open blockchain data.pptx
Vijay Dialani, PhD
 
RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...
RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...
RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...
thanhdowork
 
一比一原版(IIT毕业证)伊利诺伊理工大学毕业证成绩单专业办理
一比一原版(IIT毕业证)伊利诺伊理工大学毕业证成绩单专业办理一比一原版(IIT毕业证)伊利诺伊理工大学毕业证成绩单专业办理
一比一原版(IIT毕业证)伊利诺伊理工大学毕业证成绩单专业办理
zwunae
 
The Benefits and Techniques of Trenchless Pipe Repair.pdf
The Benefits and Techniques of Trenchless Pipe Repair.pdfThe Benefits and Techniques of Trenchless Pipe Repair.pdf
The Benefits and Techniques of Trenchless Pipe Repair.pdf
Pipe Restoration Solutions
 
Student information management system project report ii.pdf
Student information management system project report ii.pdfStudent information management system project report ii.pdf
Student information management system project report ii.pdf
Kamal Acharya
 
block diagram and signal flow graph representation
block diagram and signal flow graph representationblock diagram and signal flow graph representation
block diagram and signal flow graph representation
Divya Somashekar
 
English lab ppt no titlespecENG PPTt.pdf
English lab ppt no titlespecENG PPTt.pdfEnglish lab ppt no titlespecENG PPTt.pdf
English lab ppt no titlespecENG PPTt.pdf
BrazilAccount1
 
NO1 Uk best vashikaran specialist in delhi vashikaran baba near me online vas...
NO1 Uk best vashikaran specialist in delhi vashikaran baba near me online vas...NO1 Uk best vashikaran specialist in delhi vashikaran baba near me online vas...
NO1 Uk best vashikaran specialist in delhi vashikaran baba near me online vas...
Amil Baba Dawood bangali
 
road safety engineering r s e unit 3.pdf
road safety engineering  r s e unit 3.pdfroad safety engineering  r s e unit 3.pdf
road safety engineering r s e unit 3.pdf
VENKATESHvenky89705
 
一比一原版(SFU毕业证)西蒙菲莎大学毕业证成绩单如何办理
一比一原版(SFU毕业证)西蒙菲莎大学毕业证成绩单如何办理一比一原版(SFU毕业证)西蒙菲莎大学毕业证成绩单如何办理
一比一原版(SFU毕业证)西蒙菲莎大学毕业证成绩单如何办理
bakpo1
 
Top 10 Oil and Gas Projects in Saudi Arabia 2024.pdf
Top 10 Oil and Gas Projects in Saudi Arabia 2024.pdfTop 10 Oil and Gas Projects in Saudi Arabia 2024.pdf
Top 10 Oil and Gas Projects in Saudi Arabia 2024.pdf
Teleport Manpower Consultant
 
Nuclear Power Economics and Structuring 2024
Nuclear Power Economics and Structuring 2024Nuclear Power Economics and Structuring 2024
Nuclear Power Economics and Structuring 2024
Massimo Talia
 
Fundamentals of Electric Drives and its applications.pptx
Fundamentals of Electric Drives and its applications.pptxFundamentals of Electric Drives and its applications.pptx
Fundamentals of Electric Drives and its applications.pptx
manasideore6
 
一比一原版(UofT毕业证)多伦多大学毕业证成绩单如何办理
一比一原版(UofT毕业证)多伦多大学毕业证成绩单如何办理一比一原版(UofT毕业证)多伦多大学毕业证成绩单如何办理
一比一原版(UofT毕业证)多伦多大学毕业证成绩单如何办理
ydteq
 
power quality voltage fluctuation UNIT - I.pptx
power quality voltage fluctuation UNIT - I.pptxpower quality voltage fluctuation UNIT - I.pptx
power quality voltage fluctuation UNIT - I.pptx
ViniHema
 

Recently uploaded (20)

J.Yang, ICLR 2024, MLILAB, KAIST AI.pdf
J.Yang,  ICLR 2024, MLILAB, KAIST AI.pdfJ.Yang,  ICLR 2024, MLILAB, KAIST AI.pdf
J.Yang, ICLR 2024, MLILAB, KAIST AI.pdf
 
Governing Equations for Fundamental Aerodynamics_Anderson2010.pdf
Governing Equations for Fundamental Aerodynamics_Anderson2010.pdfGoverning Equations for Fundamental Aerodynamics_Anderson2010.pdf
Governing Equations for Fundamental Aerodynamics_Anderson2010.pdf
 
Immunizing Image Classifiers Against Localized Adversary Attacks
Immunizing Image Classifiers Against Localized Adversary AttacksImmunizing Image Classifiers Against Localized Adversary Attacks
Immunizing Image Classifiers Against Localized Adversary Attacks
 
Final project report on grocery store management system..pdf
Final project report on grocery store management system..pdfFinal project report on grocery store management system..pdf
Final project report on grocery store management system..pdf
 
Gen AI Study Jams _ For the GDSC Leads in India.pdf
Gen AI Study Jams _ For the GDSC Leads in India.pdfGen AI Study Jams _ For the GDSC Leads in India.pdf
Gen AI Study Jams _ For the GDSC Leads in India.pdf
 
ML for identifying fraud using open blockchain data.pptx
ML for identifying fraud using open blockchain data.pptxML for identifying fraud using open blockchain data.pptx
ML for identifying fraud using open blockchain data.pptx
 
RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...
RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...
RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...
 
一比一原版(IIT毕业证)伊利诺伊理工大学毕业证成绩单专业办理
一比一原版(IIT毕业证)伊利诺伊理工大学毕业证成绩单专业办理一比一原版(IIT毕业证)伊利诺伊理工大学毕业证成绩单专业办理
一比一原版(IIT毕业证)伊利诺伊理工大学毕业证成绩单专业办理
 
The Benefits and Techniques of Trenchless Pipe Repair.pdf
The Benefits and Techniques of Trenchless Pipe Repair.pdfThe Benefits and Techniques of Trenchless Pipe Repair.pdf
The Benefits and Techniques of Trenchless Pipe Repair.pdf
 
Student information management system project report ii.pdf
Student information management system project report ii.pdfStudent information management system project report ii.pdf
Student information management system project report ii.pdf
 
block diagram and signal flow graph representation
block diagram and signal flow graph representationblock diagram and signal flow graph representation
block diagram and signal flow graph representation
 
English lab ppt no titlespecENG PPTt.pdf
English lab ppt no titlespecENG PPTt.pdfEnglish lab ppt no titlespecENG PPTt.pdf
English lab ppt no titlespecENG PPTt.pdf
 
NO1 Uk best vashikaran specialist in delhi vashikaran baba near me online vas...
NO1 Uk best vashikaran specialist in delhi vashikaran baba near me online vas...NO1 Uk best vashikaran specialist in delhi vashikaran baba near me online vas...
NO1 Uk best vashikaran specialist in delhi vashikaran baba near me online vas...
 
road safety engineering r s e unit 3.pdf
road safety engineering  r s e unit 3.pdfroad safety engineering  r s e unit 3.pdf
road safety engineering r s e unit 3.pdf
 
一比一原版(SFU毕业证)西蒙菲莎大学毕业证成绩单如何办理
一比一原版(SFU毕业证)西蒙菲莎大学毕业证成绩单如何办理一比一原版(SFU毕业证)西蒙菲莎大学毕业证成绩单如何办理
一比一原版(SFU毕业证)西蒙菲莎大学毕业证成绩单如何办理
 
Top 10 Oil and Gas Projects in Saudi Arabia 2024.pdf
Top 10 Oil and Gas Projects in Saudi Arabia 2024.pdfTop 10 Oil and Gas Projects in Saudi Arabia 2024.pdf
Top 10 Oil and Gas Projects in Saudi Arabia 2024.pdf
 
Nuclear Power Economics and Structuring 2024
Nuclear Power Economics and Structuring 2024Nuclear Power Economics and Structuring 2024
Nuclear Power Economics and Structuring 2024
 
Fundamentals of Electric Drives and its applications.pptx
Fundamentals of Electric Drives and its applications.pptxFundamentals of Electric Drives and its applications.pptx
Fundamentals of Electric Drives and its applications.pptx
 
一比一原版(UofT毕业证)多伦多大学毕业证成绩单如何办理
一比一原版(UofT毕业证)多伦多大学毕业证成绩单如何办理一比一原版(UofT毕业证)多伦多大学毕业证成绩单如何办理
一比一原版(UofT毕业证)多伦多大学毕业证成绩单如何办理
 
power quality voltage fluctuation UNIT - I.pptx
power quality voltage fluctuation UNIT - I.pptxpower quality voltage fluctuation UNIT - I.pptx
power quality voltage fluctuation UNIT - I.pptx
 

mathematical model for image restoration based on fractional order total variation

  • 1. International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-3, Issue-2, Feb- 2017] https://dx.doi.org/10.24001/ijaems.3.2.19 ISSN: 2454-1311 www.ijaems.com Page | 111 Mathematical Model for Image Restoration Based on Fractional Order Total Variation Akhtar Rasool1 , Noor Badshah2 , Asmat Ullah3 2, 1 Basic Sciences Department, University of Engineering and Technology KP Peshawar (Pak). 3 Department of Engineering Mechanics, Hohai University, No.1 Xikang Road, Nanjing, Jiangsu 210098, PR China. Abstract—This paper addresses mathematical model for signal restoration based on fractional order total variation (FOTV) for multiplicative noise. In alternating minimization algorithm the Newton method is coupled with time-marching scheme for the solutions of the corresponding PDEs related to the minimization of the denoising model. Results obtained from experiments show that our model can not only reduce the staircase effect of the restored images but also better improve the PSNR as compare to other existed methods. Keywords— Total Variation (TV), Fractional Order Derivative, Speckle Noise, Image Restoration. I. INTRODUCTION Image denoising has various applications in different fields including pattern recognition, remote sensing, preprocessing for computer vision and medical images. It is still a valid challenge for researchers. Two important types of image denoising are additive noise and multiplicative noise [1, 2]. In additive noise removal model, the original image 𝑢 is considered to be contaminated by Gaussian additive noise 𝑛. Our goal is to restore 𝑢 from g = u + n. The total variation (TV) has been developed by Rudin, Osher and Fatemi. TV based method is edge preserving [1]. We discuss multiplicative noise in this paper. Consider the noisy image 𝑔: Ω → R, the original image 𝑢 is contaminated by some noise 𝜈 satisfying the Gamma law with mean 1. This type of noise mostly exists in ultrasound imaging, sonar and laser imaging and synthetic aperture radar images. Multiplicative noise are more complex as compared to additive noise. Rudin et al early approached a variational model [3]. Second approach for multiplicative noise has been proposed by Aubert and Aujol, see [4] for more details. Keeping in mind the statistical properties of multiplicative noise 𝑛, we approximate 𝑔 by treating the following constrained minimization problem min 𝑢 ∫ |𝐷𝑢|Ω 𝑑𝑥𝑑𝑦 (1) Subject to: ∫ 𝑔 𝑢 𝑑𝑥𝑑𝑦 = 1,Ω ∫ ( 𝑔 𝑢 − 1) 2 𝑑𝑥𝑑𝑦 = 𝜎2 .Ω The mean of noise is 1 with standard deviation 𝜎2 as listed above. TV regularization methods remove noise and simultaneously preserve sharp intensity boundaries but sometimes textures are smoothed with noise in noise removing process [1-3]. The staircase effects can be reduced by generalizing the regularization term which can be performed in two ways. First approach is based on high order derivative; see [5-7]. Secondly, the regularization term can be generalized by using fractional-order derivatives. In recent past, some general fractional-order additive noise models have been treated in [8,9]. Chambolle used the fitting term ||𝑔 − 𝑢||2 2 in ROF model and hence introduced an efficient projection method for additive noise [10]. Obtained results indicate that the performance of fractional-order TV are better than integer high order TV. Fractional order models have competitive performance with other integer order based models in noise removing as well as staircase reduction. Moreover Aubert and Aujol developed a multiplicative noise model “AA-model”, which is given by min 𝑢 ∫ (𝑙𝑜𝑔𝑢 + 𝑔 𝑢 ) 𝑑𝑥𝑑𝑦 + 𝜆 ∫ |𝐷𝑢|𝑑𝑥𝑑𝑦ΩΩ (2) Recently Zhang Jun and Wei Zhihui introduced fractional order AA Model by replacing TV term in AA model with fractional order TV. As the function (logu + g u ) is not convex everywhere, therefore Haung et al [11] introduced the auxiliary variable w = logu and introduced a strictly convex function 𝑚𝑖𝑛 𝑤,𝑢 ∫ (𝑤 + 𝑔𝑒−𝑤)Ω 𝑑𝑥𝑑𝑦 + 𝛼1|𝑤 − 𝑢| 𝐿2 2 + 𝛼2 ∫ |𝐷𝑢|𝑑𝑥𝑑𝑦Ω , (3) with two regularization parameters 𝛼1 and 𝛼2. As 𝑢 = 𝑒−𝑤 is strictly convex for all value of 𝑤 which ensures that the solution of the variational problem is unique. TV based models have better performance in preserve sharp edges, reducing the noise but often these models cause the staircase effect. We develop fractional-order TV based method for multiplicative noise in Sect. 2. In Sect. 3, we use alternating minimization algorithm to find minimizer of our proposed problem .In Sect. 4, we give some numerical results to validate
  • 2. International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-3, Issue-2, Feb- 2017] https://dx.doi.org/10.24001/ijaems.3.2.19 ISSN: 2454-1311 www.ijaems.com Page | 112 the performance of the proposed methodology. Finally, the paper is concluded in the last Section. II. THE PROPOSED APPROACH We introduce a convex function for restoring images contaminated by multiplicative noise in this section. Our new model is based on fractional Order AA model and HNW algorithm. Replacing TV term in model (3) with Fractional order TV, we introduce the following model. 𝑚𝑖𝑛 𝑤,𝑢 ∫ (𝑤 + 𝑔𝑒−𝑤 )𝑑𝑥𝑑𝑦 Ω +𝛼1||𝑤 − 𝑢|| 𝐿2 2 + (𝛼2 + 𝛼3 𝑢2 )|𝐷 𝛼 𝑢|, (4) with three positive regularization parameters 𝛼1, 𝛼2 and 𝛼3. The fractional-order total variation norm provides a better performance in preserving image edges. It also reduces staircase effect. Moreover 𝑢 and 𝑤 preserve edges at the same point; see [9]. The use of fractional-order TV to 𝑤 is quite better for preserving sharp edges and reducing staircase effect. The main advantage is that u = ew is positive for all 𝑤, even though 𝑢 in the objective function (2) required to be positive. In our proposed model if we assign 𝑤 a large negative number, 𝑢 is still a small non zero positive number. III. NUMERICAL SCHEME We use an alternating minimization algorithm to solve the FOTV problem. We split eqn. (4) in the following two subproblems. (i) First fix 𝑢 and find the solution of min w ∫ (w + ge−w )dxdyΩ + α1||𝑤 − 𝑢||L2 2 . (5) (ii) Then fix 𝑤, find the solution of min u α1||w − u||L2 2 +(α2 + α3u2) ∫ |Dα u|dxdy.Ω (6) Applying discrete setting, the first problem gets the form: min w ∑ (𝑤𝑖 + 𝑔𝑒−𝑤 𝑖) + 𝛼1||𝑤 − 𝑢||2 2 .𝑛2 1 (7) The minimizer of the above problem can be obtained by solving the following nonlinear system 1 − 𝑔𝑒−𝑤 𝑖 + 2𝛼1(𝑤𝑖 − 𝑢𝑖) = 0, 𝑖 = 1,2, . . . 𝑛2 . (8) The 2nd derivative w.r.t 𝑤 of the first term in (5) is 𝑔𝑒−𝑤 , which is positive for positive value of 𝑔. Thus the strictly convex function (5) has a unique minimum which we compute with Newton method as wi+1 = wi − 1 − ge−wi + 2α1(wi − u) ge−wi + 2α1 , i = 0,1,2, . .. The Euler-Lagrange equation of FOTV model (6) is 𝜕𝑢 𝜕𝑡 = 2𝛼1 𝛼2+𝛼3 𝑢2 (𝑤𝑖,𝑗 𝑛 − 𝑢𝑖,𝑗) + (−1) 𝛼 𝑑𝑖𝑣 𝛼 ∇ 𝛼 𝑢 𝑖,𝑗 |∇ 𝛼 𝑢 𝑖,𝑗| (9) 𝜕𝑢 𝜕𝑛 = 0 𝑜𝑛 𝜕Ω, u(0) = u0. The numerical scheme for the steady-state equation (9) is given by 𝑢𝑖,𝑗 𝑛+1 = 𝑢𝑖,𝑗 𝑛 + 𝜂𝜏|𝑤𝑖,𝑗 − 𝑢𝑖,𝑗| + (−1) 𝛼 𝑑𝑖𝑣 𝛼 ∇ 𝛼 𝑢𝑖,𝑗 |∇ 𝛼 𝑢𝑖,𝑗| , 𝜂 = 2𝛼1 𝛼2+𝛼3 𝑢2 with B. Conditions 𝑢0,𝑗 𝑛 = 𝑢1,𝑗 𝑛 , 𝑢 𝑁,𝑗 𝑛 = 𝑢 𝑁−1,𝑗 𝑛 , 𝑢𝑖,0 𝑛 = 𝑢𝑖,1 𝑛 , 𝑢𝑖,𝑁 𝑛 = 𝑢𝑖,𝑁−1 𝑛 . Algorithm We take initial guess 𝑢(0) to be the noisy image. (i) First fix𝑢(𝑘−1) , compute by Newton's algorithm 𝑤(𝑘) = arg min w ∑ (𝑤𝑖 + 𝑔𝑛2 𝑘=0 𝑒−𝑤 𝑖) + 𝛼1||𝑤 − 𝑢 𝑘−1||2 2 (ii) Then fix 𝑤(𝑘) , compute 𝑢(𝑘) by using 𝜕𝑢 𝜕𝑡 = 2𝛼1 𝛼2 + 𝛼3 𝑢2 ||𝑤(𝑘) − 𝑢|| + (−1) 𝛼 𝑑𝑖𝑣 𝛼 ( ∇ 𝛼 𝑢 |∇ 𝛼 𝑢| ) (iii) Exit with an approximate denoised image 𝑢 𝑘 = 𝑒 𝑤(𝑘) if the stopping criteria ||𝑢 𝑘+1−𝑢 𝑘|| ||𝑢 𝑘+1|| 2 < 10−4 is satisfied, otherwise we continue with 𝑘 = 𝑘 + 1 for next iteration. IV. NUMERICAL RESULTS This section describes some experimental results of image denoising to claim the superiority of our method. We consider some images contaminated by multiplicative noise. Figure 1 shows clean images. Results obtained from experiments indicate the superiority of our proposed method. We use PSNR to compare the capabilities of the restored images as 𝑃𝑆𝑁𝑅 = 20 log10 255 ||𝑢 − 𝑢 𝑡𝑟𝑢𝑒|| Where 𝑢 𝑡𝑟𝑢𝑒 and 𝑢 denote the noise-free signal and recovered signal respectively.
  • 3. International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-3, Issue-2, Feb- 2017] https://dx.doi.org/10.24001/ijaems.3.2.19 ISSN: 2454-1311 www.ijaems.com Page | 113 Original Images Fig. 1: Face, Peppers, Barbara (a) Face 0.3 (b) AA Method (c) Proposed Method Fig.2: (a). Noisy Image, 𝜎2 = 0.3 (b) Restored signal “AA-method”, 𝜆 = 35 (c) Restored signal “Proposed method”, 𝛼1 = 190, 𝛼2 = 0.00002 𝑎3 = 490. (a) Face, 𝜎 = 0.1(b) AA-Method (c) Proposed Method Fig 3 (a) Noisy Signal, 𝜎2 = 0.1(b) Restored Signal, AA-Method, 𝜆 = 35 (c) Restored signal “Proposed method”, 𝛼1 = 190, 𝛼2 = 0.00002 𝑎3 = 490. (a) Noisy Image, 𝜎 = 0.03 (b) AA Method (c) Proposed Method Fig.4:(a) Noisy Signal, 𝜎2 = 0.03. (b) Restored Signal, AA-Method, 𝜆 = 35 (c) Proposed Method, 𝛼1 = 16, 𝛼2 = 0.0002, 𝛼3 = 150.
  • 4. International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-3, Issue-2, Feb- 2017] https://dx.doi.org/10.24001/ijaems.3.2.19 ISSN: 2454-1311 www.ijaems.com Page | 114 Test 1: (a) Noisy Image, 𝜎 = 0.03 (b) AA Method (c) Proposed Method Fig.5: (a) Noisy Signal,𝜎2 = 0.01. (b) Restored Signal, AA-Method, 𝜆 = 34 (c) Proposed Method,𝛼1 = 12, 𝛼2 = 0.0002, 𝛼3 = 490. Test 2. Figure (a) Noisy Signal (b) AA Method (c) Proposed Method Fig.7: (a) Noisy Signal,𝜎2 = 0.1. (b) Restored Signal, AA-Method, 𝜆 = 35 (c) Restored Signal, Proposed Method, 𝛼1 = 33, 𝛼2 = 0.00001, 𝛼3 = 600. (a) Noisy Image (b) AA-Method (c) Proposed Method Fig.8: (a) Noisy Signal,𝜎2 = 0.03. (b) Restored Signal, AA-Mode, 𝜆 = 35 (c) Proposed Method,𝛼1 = 200, 𝛼2 = 0.0002, 𝛼3 = 600. (a) Noisy Image (b) AA-Method (c) Proposed Method Fig.9: (a) Noisy Signal,𝜎2 = 0.01. (b) Restored Signal, AA-Method, 𝜆 = 35 (c) Restored Signal, Proposed Method with 𝛼1 = 200, 𝛼2 = 0.00001, 𝛼3 = 600.
  • 5. International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-3, Issue-2, Feb- 2017] https://dx.doi.org/10.24001/ijaems.3.2.19 ISSN: 2454-1311 www.ijaems.com Page | 115 Test 3. (a)Barbara, 0.03 (b) AA Method (c) Proposed Method Fig.10: (a) Noisy Signal, 𝜎2 = 0.03 (b) Restored Signal, AA-Method, 𝜆 = 35 (c) Restored Signal, Proposed Method with 𝛼1 = 450, 𝛼2 = 0.0001, 𝛼3 = 650. (a) Barbara, 0.1 (b) AA Method (c) Proposed Method Fig.11: (a) Noisy Signal, 𝜎2 = 0.1 (b) Restored Signal, AA Method, 𝜆 = 33 (c) Restored Signal, Proposed Method with 𝛼1 = 450, 𝛼2 = 0.00001, 𝛼3 = 650. Table 1: Restoration results for face images Image Variance "𝜎2 " Fractional Order 𝛼 AA-Model PSNR Proposed Model PSNR Face Size(256 × 256) 0.3 1.9 26.035 27.193 0.1 1.1 28.525 29.009 0.03 1.9 31.934 33.340 0.01 1.9 35.037 36.420 Table 2: Restoration results for Peppers images Image Variance "𝜎2 " Fractional Order 𝛼 AA-Model PSNR Proposed Model PSNR Peppers Size(256 × 256) 0.1 1.1 26.152 26.301 0.03 1.1 29.228 29.355 0.01 1.1 32.142 32.201 Table 3: Restoration Results for Barbara Images. Image Variance "𝜎2 " Fractional Order 𝛼 AA-Model PSNR Proposed Model PSNR Barbara Size(256 × 256) 0.03 1.1 28.162 28.211 0.1 1.1 25.821 25.910 In first experiment we test two dimensional signal “face” 256 × 256. We have compared the denoising capabilities of the proposed algorithm with fractional order AA method. Numerous values of 𝜆 have been tested for fractional order AA algorithm and the best amongst them are mentioned in this paper. As the regularization parameters perform well in denoising so we have chosen the three regularization parameters for increasing the PSNR. In Table 1 we have mentioned the highest PSNR of fractional order AA method and the proposed method. Peppers image of size 256 × 256 is
  • 6. International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-3, Issue-2, Feb- 2017] https://dx.doi.org/10.24001/ijaems.3.2.19 ISSN: 2454-1311 www.ijaems.com Page | 116 used to make the second experiment. Best results are chosen among the obtained results of the performed experiements through fractional order AA method. As the quality of the restored image depends upon the regularization parameters so we have chosen the suitable values for regularization parameters. In table 2 we have shown the comparison of our results with those obtained from fractional order AA Algorithm. The PSNR values of the recovered peppers images are listed in table 2. The values mentioned in table 2 clearly shows that the PSNR values of the proposed method are bigger than fractional order AA-Method. In third test we consider the barbara image. The PSNR values shown in table 3 showing the better performance of the proposed algorithm. V. CONCLUSION This paper describes a new mathematical approach for image denoising with fractional-order TV. Coupling the Newton algorithmwith time- marching scheme for the solutions of PDEs related to the minmization of denoising model based on fractional order total variation, gives good results.The proposed method can suppress noise very well while it can preserve details of the recovered image. Results obtained from experiements show that the efficiency of the proposed approach are better than what obtained from other existed total variation restoration methods. REFERENCES [1] Rudin L, Osher S, and Fatemi E, “Nonlinear Total Variation Based Model Noise Removal Algorithm,”J. Phys, 60:259-268(1992). [2] A. Chambolle and P.-L. Lions, “Image Recovery via total Variation Minimization and Related Problems,” Numerische Mathematik, 76(2):167-188(1997). [3] L. Rudin, P.-L. Lions, and S. Osher, “Multiplicative Denoising and Deblurring: Theory and Algorithms,” in Geometric Level Sets in Imaging, Vision and Graphics, S. Osher and N. Paragios, Eds., Springer, New York, NY, USA,103-119(2003). [4] G. Aubert, J.-F. Aujol, “A Variational Approach to Removing Multiplicative Noise,” SIAM Journal on Applied Mathematics, 68(4):925-946(2008). [5] You Y. L, Kaveh M.,“Fourth-Order Partial Differential Equations for Noise Removal,”IEEE Transactions on Image Processing, 9(10): 1723-1730(2000). [6] Didas S., Weicket J. and burgeth B,“Properties of Higher Order Nonlinear Diffusion Filtering,” Journal of Mathematical Imaging and Vision, 35(3):208-226(2009). [7] Lysaker M., Lundrevold A., Tai X.C., “Noise Removal Using Fourth Order Partial Differential Equation with Applications to Medical Magnetic Resonance Images in Space and Time,” IEEE Transactions on Image Processing, 12(12): 1579-1590(2003). [8] Bai J., Feng X. C.,“Fractional-Order Anisotropic Diffusion for Image Denoising,”IEEE Transactions on Image Processing, 16(10): 2492-2502(2007). [9] Zhang J. and Wei Z. H., “A Class of Fractional-Order Multi-scale Variational Models and Alternating Projection Algorithm for Image Denoising,” Applied Mathematical Modeling, 35(5): 2516-2528(2011). [10] Chambolle A., “An algorithm for Total Variation Minimization and Applications,”Journal of Mathematical Imaging and Vision, 20(1): 89-97(2004). [11] Y.-M. Haung, M. K. Ng, and Y.-W. Wen, “A New Total Variation Method for Multiplicative Noise Removal”, SIAM Journal on Imaging Sciences, 2(1):20-40(2009).