Available Online at www.ijpret.com 103
INTERNATIONAL JOURNAL OF PURE AND
APPLIED RESEARCH IN ENGINEERING AND
TECHNOLOGY
A PATH FOR HORIZING YOUR INNOVATIVE WORK
CONVOLUTION THEOREM FOR TWO DIMENSIONAL FRACTIONAL MELLIN
TRANSFORM
V. D. SHARMA1, P. B. DESHMUKH2
1. Department of Mathematics, Arts, Commerce and Science College, Kiran Nagar, Amravati, India 2. Department of Engineering Mathematics, IBSS College of Engineering, Amravati, India
Accepted Date: 05/03/2015; Published Date: 01/05/2015
\
Abstract: The fractional Mellin transform is a generalization of the Mellin transform. It has many applications in the electronic world, such as computation of algorithm, cryptographic scheme, vowel recognition, pattern recognition. Signal processing and pattern recognition algorithms make extensive use of convolution. Convolution is an important tool because of its translation invariance property. Also convolution is a powerful way of characterizing the input-output relationship of time invariance system. In this paper we discussed the definition of two dimensional fractional Mellin transform (FRMT) which is extended to the distribution of compact support. The convolution theorem for FRMT is also proved.
Keywords: Fractional Mellin Transform, Mellin Transform, Testing Function space, Generalize function
Corresponding Author: MR.V. D. SHARMA
Access Online On:
www.ijpret.com
How to Cite This Article:
Available Online at www.ijpret.com 104
INTRODUCTION
The fractional integral transform is a backbone of today’s technology, because fractional integral transform has many applications which are very useful in electronic world.
Also Mellin transform has many applications such as navigation, radar system [1], in finding the stress distribution in an infinite wedge [1], also used in digital audio effects [1].
Mellin transform is also useful for segregating the information about the size and shape of vocal track by stabilizing the Mellin image [5]. For the past decades, the Mellin transform has received considerable attention from the optical image processing, radar and sonar signal processing and target classification [6]. Because of scale invariance property of Mellin transform the independent speech recognition of speaker is possible [6].
By using convolution theorem we can tabulate atomic scattering factors by working out the diffraction pattern of atoms place at origin. In pattern recognition convolution is an important tool because of its translation invariance property. It is useful for extraction of information in communication engineering [2].
In this paper we proposed a convolution theorem for two dimensional fractional Mellin transform. Also defined the testing function space and definition of two dimensional fractional Mellin transform.
The two dimensional fractional Mellin transform with parameter θ of f(x,y)denoted by
)} , ( {
FRMT f x y performs a linear operation, given by the integral transform
dxdy v u y x K y x f v u F y x f
FRMT{ ( , )} ( , ) ( , ) ( , , , )
0 0
(1)
where the kernel
] log log [ tan 1 sin 2 1 sin
2 2 2 2 2
) , , ,
( u v x y
i iv iu e y x v u y x
K
. 2
0 (2)
where, 𝐶1𝜃 = 2𝜋𝑖
𝑠𝑖𝑛𝜃, 𝐶2𝜃 = 𝜋
𝑡𝑎𝑛𝜃 (3)
For the convolution of two dimensional fractional Mellin transform.
Let, f, g be in W and denote their convolution h is as
ℎ(𝑥, 𝑦) = ∫ ∫ 𝑓̃(𝑟, 𝑠)0∞ 0∞ 1𝑟1𝑠𝑔̃(𝑥
𝑟, 𝑦
𝑠)𝑑𝑟𝑑𝑠
Available Online at www.ijpret.com 105
For any function 𝑓(𝑥, 𝑦)
𝑓̃(𝑥, 𝑦) = 𝑓(𝑥, 𝑦)𝑒𝑖𝐶2𝜃(𝑙𝑜𝑔2𝑥+𝑙𝑜𝑔2𝑦) (5)
Then for any two functions f & g the convolution operation * is defined as
ℎ(𝑥, 𝑦) = (𝑓 ∗ 𝑔)(𝑥, 𝑦)
= 𝑒−𝑖𝐶2𝜃(𝑙𝑜𝑔2𝑥+𝑙𝑜𝑔2𝑦)(𝑓̃ ∗ 𝑔̃)(𝑥, 𝑦) (6)
where * is the convolution operation for the FRMT as defined in (6).
1.1The test function space E
An infinitely differentiable complex valued function
x,y on nR belongs to
nR
E , if for each
compact set IS, KSbwhere
} 0 , , : {
x x R x a a
Sa Sb {y:yR, y b,b0}, n
R K
I,
) , ( )] , ( [ , , ,
, x y
sup
D x yq y x K y I x q
E
Thus E(Rn)will denote the space of all (x,y)E(Rn) with compact support contained in S&
b S .
Note that the space E is complete and therefore a Frechet space. Moreover, we say that f(x,y)
is a fractional Mellin transformable if it is a member of *
E , the dual of E.
1.2Two dimensional fractional Mellin transform (FRMT)
The two dimensional fractional Mellin transform of f(x,y)E*(Rn)can be defined by ) , , , ( ), , ( ) , ( )} , (
{f x y F u v f x y K x y u v
FRMT
] log log [ tan 1 sin 2 1 sin
2 2 2 2 2
) , , ,
( u v x y
i iv iu e y x v u y x
K
Right hand side of equation (4) has a
meaning as the application of *
) ,
(x y E
f to K(x,y,u,v)E. It
can be extended to the complex space as an entire function given by ) , , , ( ), , ( ) , ( )} , (
{f x y F p k f x y K x y p k
FRMT The right hand side is meaningful because for each
n C k
Available Online at www.ijpret.com 106
II. Convolution Theorem of two dimensional fractional Mellin transform:
Let ℎ(𝑥, 𝑦) = (𝑓 ∗ 𝑔)(𝑥, 𝑦) & 𝐹𝜃, 𝐺𝜃, 𝐻𝜃 denotes the two dimensional fractional Mellin transform
of 𝑓, 𝑔, ℎ respectively then 𝐻𝜃(𝑢, 𝑣) = 𝑒−𝑖𝐶2𝜃(𝑢2+𝑣2)𝐹
𝜃(𝑢, 𝑣)𝐺𝜃(𝑢, 𝑣).
Proof- By using the definition of the two dimensional fractional Mellin transform
𝐻𝜃(𝑢, 𝑣) = ∫ ∫ ℎ(𝑥, 𝑦)𝑥𝐶1𝜃𝑢−1
∞
0
𝑦𝐶1𝜃𝑣−1
∞
0
𝑒𝑖𝐶2𝜃(𝑢2+𝑣2+𝑙𝑜𝑔2𝑥+𝑙𝑜𝑔2𝑦)𝑑𝑥𝑑𝑦
By using (6)
= ∫ ∫ (𝑓 ∗ 𝑔)(𝑥, 𝑦)𝑥𝐶1𝜃𝑢−1
∞
0
𝑦𝐶1𝜃𝑣−1
∞
0
𝑒𝑖𝐶2𝜃(𝑢2+𝑣2+𝑙𝑜𝑔2𝑥+𝑙𝑜𝑔2𝑦)𝑑𝑥𝑑𝑦
= ∫ ∫ 𝑒−𝑖𝐶2𝜃(𝑙𝑜𝑔2𝑥+𝑙𝑜𝑔2𝑦)
∞
0 ∞
0
(𝑓̃ ∗ 𝑔̃)(𝑥, 𝑦)𝑥𝐶1𝜃𝑢−1
𝑦𝐶1𝜃𝑣−1𝑒𝑖𝐶2𝜃(𝑢2+𝑣2+𝑙𝑜𝑔2𝑥+𝑙𝑜𝑔2𝑦)𝑑𝑥𝑑𝑦
By using (4)
= ∫ ∫ 𝑒−𝑖𝐶2𝜃(𝑙𝑜𝑔2𝑥+𝑙𝑜𝑔2𝑦)𝑥𝐶1𝜃𝑢−1
∞
0
𝑦𝐶1𝜃𝑣−1
∞
0
𝑒𝑖𝐶2𝜃(𝑢2+𝑣2+𝑙𝑜𝑔2𝑥+𝑙𝑜𝑔2𝑦)𝑑𝑥𝑑𝑦
∫ ∫ 𝑓̃(𝑟, 𝑠)0∞ 0∞ 1𝑟1𝑠𝑔̃(𝑥
𝑟, 𝑦
𝑠)𝑑𝑟𝑑𝑠
Putting 𝑥
𝑟= 𝑚, 𝑦 𝑠 = 𝑛
⇒ 𝑥
𝑚= 𝑟, 𝑦 𝑛 = 𝑠
Differentiate with respect to m & n respectively
−𝑥
𝑚2𝑑𝑚 = 𝑑𝑟,
−𝑦
Available Online at www.ijpret.com 107
𝐻𝜃(𝑢, 𝑣) = ∫ ∫ 𝑒−𝑖𝐶2𝜃(𝑙𝑜𝑔
2𝑥+𝑙𝑜𝑔2𝑦)
𝑥𝐶1𝜃𝑢−1
∞
0 ∞
0
𝑦𝐶1𝜃𝑣−1𝑒𝑖𝐶2𝜃(𝑢2+𝑣2+𝑙𝑜𝑔2𝑥+𝑙𝑜𝑔2𝑦)𝑑𝑥𝑑𝑦
∫ ∫ 𝑓̃ (𝑚𝑥,𝑦
𝑛) ∞
0 ∞ 0
𝑚 𝑥 𝑛
𝑦𝑔̃(𝑚, 𝑛)
(−𝑥
𝑚2) (
−𝑦
𝑛2) 𝑑𝑚𝑑𝑛
= ∫ ∫ 𝑥𝐶1𝜃𝑢−1
∞
0
𝑦𝐶1𝜃𝑣−1𝑒𝑖𝐶2𝜃(𝑢2+𝑣2)𝑑𝑥𝑑𝑦
∞
0
∫ ∫ 𝑚𝑛1 𝑓̃ (𝑥
𝑚, 𝑦 𝑛) ∞
0 ∞
0 𝑔̃(𝑚, 𝑛)𝑑𝑚𝑑𝑛
By using (5) = ∫ ∫ 𝑥∞ 𝐶1𝜃𝑢−1
0 𝑦
𝐶1𝜃𝑣−1
∞
0 𝑒
𝑖𝐶2𝜃(𝑢2+𝑣2)𝑑𝑥𝑑𝑦
∫ ∫ 𝑚𝑛1 𝑓 (𝑥
𝑚, 𝑦 𝑛) 𝑒
𝑖𝐶2𝜃(𝑙𝑜𝑔2 𝑥𝑚+𝑙𝑜𝑔2𝑦𝑛)
∞ 0 ∞ 0
𝑔(𝑚, 𝑛)𝑒𝑖𝐶2𝜃(𝑙𝑜𝑔2𝑚+𝑙𝑜𝑔2𝑛)𝑑𝑚𝑑𝑛
Putting 𝑚𝑥 = 𝑝 , 𝑦𝑛= 𝑞
⇒ 𝑥 = 𝑚𝑝 , 𝑦 = 𝑛𝑞
⇒ 𝑑𝑥 = 𝑚𝑑𝑝 , 𝑑𝑦 = 𝑛𝑑𝑞
𝐻𝜃(𝑢, 𝑣) = ∫ ∫ (𝑚𝑝)𝐶1𝜃𝑢−1
∞
0 (𝑛𝑞)
𝐶1𝜃𝑣−1
∞
0
𝑒𝑖𝐶2𝜃(𝑢2+𝑣2)∫ ∫ 1
𝑚𝑛𝑓(𝑝, 𝑞)𝑒
𝑖𝐶2𝜃(𝑙𝑜𝑔2𝑝+𝑙𝑜𝑔2𝑞)
∞ 0 ∞ 0
𝑔(𝑚, 𝑛)𝑒𝑖𝐶2𝜃(𝑙𝑜𝑔2𝑚+𝑙𝑜𝑔2𝑛)𝑚𝑑𝑝𝑛𝑑𝑞𝑑𝑚𝑑𝑛
= ∫ ∫ (𝑚)∞ 𝐶1𝜃𝑢−1
0 (𝑛)
𝐶1𝜃𝑣−1
∞
0 𝑔(𝑚, 𝑛)
𝑒𝑖𝐶2𝜃(𝑢2+𝑣2+𝑙𝑜𝑔2𝑚+𝑙𝑜𝑔2𝑛)𝑑𝑚𝑑𝑛 𝑒−𝑖𝐶2𝜃(𝑢2+𝑣2)∫ ∫ (𝑝)∞ 𝐶1𝜃𝑢−1 (𝑞)𝐶1𝜃𝑣−1
0 ∞ 0
𝑓(𝑝, 𝑞)𝑒𝑖𝐶2𝜃(𝑢2+𝑣2+𝑙𝑜𝑔2𝑝+𝑙𝑜𝑔2𝑞)𝑑𝑝𝑑𝑞 = 𝑒−𝑖𝐶2𝜃(𝑢2+𝑣2)𝐺
𝜃(𝑢, 𝑣)𝐹𝜃(𝑢, 𝑣)
= 𝑒−𝑖𝐶2𝜃(𝑢2+𝑣2)𝐹
Available Online at www.ijpret.com 108
III. Diagram:
CONCLUSION
In this paper we have defined distributional two dimensional fractional Mellin transform with compact support. Convolution theorem for two dimensional fractional Mellin transform is proved.
REFERENCES
1. V.D. Sharma, P.B.Deshmukh: “Inversion theorem of two dimensional fractional Mellin transform”, International Journal of Applied Mathematics & Mechanics, Vol.-3, No.-1 (2014), pp 33-39.
2. V.D. Sharma, P.B.Deshmukh: “A convolution theorem for the two dimensional fractional Fourier Transform in generalized sense”, 3rd International Conference of Emerging Trends in
Engineering & Technology, 2010, IEEE Computer Society.
3. Ahmed I Zayed, “A convolution & product theorem for the fractional Fourier transform”, IEEE Signal Processing letters, vol-5, No.-4, April 1998.
𝑪
𝟐𝜽=
𝝅
𝒕𝒂𝒏𝜽
Convolution for Two-Dimensional
Fractional Mellin Transform
𝒆𝒊𝑪𝟐𝜽(𝒍𝒐𝒈𝟐𝒙+𝒍𝒐𝒈𝟐𝒚)
f(x,y)
g(x,y)
C O N V O L U T I O N
Available Online at www.ijpret.com 109
4. V.N.Mahalle, A.S. Gudadhe, R.D.Taywade: “Generalization of linear Scale Invariant System in he fractional domain and some properties of Fractional Complex Mellin Transform”, International Journal of Contemp. Math Sciences, vol-6,2011, no.-2, 577-584.
5. Toshio Irino, Roy D. Patterson: “Segregating information about the size & shape of the vocal tract using a time domain auditory Model, The stbilised wavelet Mellin Transform”, Speech communication 36 (2002), pp 181-203.