Applied Mathematics, 2014, 5, 3001-3010 Published Online November 2014 in SciRes. http://www.scirp.org/journal/am http://dx.doi.org/10.4236/am.2014.519287
A Study on B-Spline Wavelets and Wavelet Packets Sana Khan, Mohammad Kalimuddin Ahmad Department of Mathematics, Aligarh Muslim University, Aligarh, India Email:
[email protected],
[email protected] Received 16 August 2014; revised 14 September 2014; accepted 5 October 2014 Copyright © 2014 by authors and Scientific Research Publishing Inc. This work is licensed under the Creative Commons Attribution International License (CC BY). http://creativecommons.org/licenses/by/4.0/
Abstract In this paper, we discuss the B-spline wavelets introduced by Chui and Wang in [1]. The definition for B-spline wavelet packets is proposed along with the corresponding dual wavelet packets. The properties of B-spline wavelet packets are also investigated.
Keywords B-Splines, Spline Wavelets, Wavelet Packets
1. Introduction Spline wavelet is one of the most important wavelets in the wavelet family. In both applications and wavelet theory, the spline wavelets are especially interesting because of their simple structure. All spline wavelets are linear combination of B-splines. Thus, they inherit most of the properties of these basis functions. The simplest example of an orthonormal spline wavelet basis is the Haar basis. The orthonormal cardinal spline wavelets in L2 ( ) were first constructed by Battle [2] and Lemarié [3]. Chui and Wang [4] found the compactly supported spline wavelet bases of L2 ( ) and developed the duality principle for the construction of dual wavelet bases [1] [5]. Wavelets are a fairly simple mathematical tool with a variety of possible applications. If 2 j 2ψ 2 j x − k , k ∈ is an orthonormal basis of L2 ( ) , then ψ is called a wavelet. Usually a wavelet is derived from a given multiresolution analysis of L2 ( ) . The construction of wavelets has been discussed in a great number of papers. Now, considerable attention has been given to wavelet packet analysis as an important generalization of wavelet analysis. Wavelet packet functions consist of a rich family of building block functions and are localized in time, but offer more flexibility than wavelets in representing different kinds of signals. The main feature of the wavelet transform is to decompose general functions into a set of approximation functions with different scales. Wavelet packet transform is an extension of the wavelet transform. In wavelet transformation signal de-
(
)
How to cite this paper: Khan, S. and Ahmad, M.K. (2014) A Study on B-Spline Wavelets and Wavelet Packets. Applied Mathematics, 5, 3001-3010. http://dx.doi.org/10.4236/am.2014.519287
S. Khan, M. K. Ahmad
composes into approximation coefficients and detailed coefficients, in which further decomposition takes place only at approximation coefficients whereas in wavelet packet transformation, detailed coefficients are decomposed as well which gives more wavelet coefficients for further analysis. For a given multiresolution analysis and the corresponding orthonormal wavelet basis of L2 ( ) , wavelet packets were constructed by Coifman, Meyer and Wickerhauser [6] [7]. This construction is an important generalization of wavelets in the sense that wavelet packets are used to further decompose the wavelet components. There are many orthonormal bases in the wavelet packets. Efficient algorithms for finding the best possible basis do exist. Chui and Li [8] generalized the concept of orthogonal wavelet packets to the case of nonorthogonal wavelet packets. Yang [9] constructed a scale orthogonal multiwavelet packets which were more flexible in applications. Xia and Suter [10] introduced the notion of vector valued wavelets and showed that multiwavelets can be generated from the component functions in vector valued wavelets. In [11], Chen and Cheng studied compactly supported orthogonal vector valued wavelets and wavelet packets. Other notable generalizations are biorthogonal wavelet packets [12], non-orthogonal wavelet packets with r-scaling functions [13]. The outline of the paper is as follows. In §2 , we introduce some notations and recall the concept of B-splines and wavelets. In §3 , we discuss the B-spline wavelet packets and the corresponding dual wavelet packets.
2. Preliminaries In this Section, we introduce B-spline wavelets (or simply B-wavelets) and some notions used in this paper. Every mth order cardinal spline wavelet is a linear combination of the functions N 2( mm) ( 2 x − j ) . Here the function N m is the mth order cardinal B-spline. Each wavelet is constructed by spline multiresolution analysis. Let m be any positive integer and let N m denotes the mth order B-spline with knots at the set of integers such that
supp ( N m ) = [ 0, m ]. The cardinal B-splines N m are defined recursively by the equations
N1 ( x ) = χ[0,1] ( x ) ,
N m ( x ) =( N m −1 ∗ N1 ) ( x ) =∫0 N m −1 ( x − t ) dt , 1
m =2,3,.
We use the following convention for the Fourier transform, ∞ fˆ (ω ) = ∫−∞ f ( t ) e −iωt dt.
The Fourier transform of the scaling function N m is given by m
(
1 − e −iω Nˆ m (ω ) = . iω
)
(1)
For each j , k ∈ , we set = N m; j , k N m 2 j x − k , and for each j ∈ , let V jm denotes the L2 -closure of the algebraic span of { N m; j , k } . Then N m is said to generate spline multiresolution analysis if the following conditions are satisfied. 1) ⊂ V−m1 ⊂ V0m ⊂ V1m ⊂ 2) clos L2 ( ) V jm = L2 ( ) ; j∈ 3)
V jm = {0} , j∈
4) for each j , { N m; j , k } is a Riesz basis of V jm . Following Mallat [14], we consider the orthogonal complementary subspaces W−m1 , W0m , W1m , that is; 5) V jm+1 V jm ⊕W jm , ∀j ∈ . = 6) W jm ⊥ Wkm ,
∀k ≠ j .
7) L ( ) = ⊕ W jm . 2
j∈
3002
S. Khan, M. K. Ahmad
These subspaces W jm , j ∈ , are called the wavelet subspaces of L2 ( ) relative to the B-spline N m . Since N m ( x ) ∈ V jm and V jm ⊂ V jm+1 , we have
∑ pk N m ( 2 x − k ) ,
= Nm ( x ) where
{ pk }
(2)
k ∈
is some sequence in 2 . Taking the Fourier transform on both sides of (2), we obtain 1 ω Nˆ m (ω ) = ∑ pk e − ikω 2 Nˆ m . 2 k∈ 2
(3)
Substituting the value of Nˆ m (ω ) from (1) into (3), we have m
m
m m 1 − e −iω iω 2 1 + e −iω 2 m 1 − ikω 2 = p e = 2− m ∑ e − ikω 2 . ∑ = k − iω 2 2 k∈ i ω 2 k= 0 k 1− e
This gives
− m +1 m 2 for 0 ≤ k ≤ m pk = k 0 otherwise.
(4)
So, (2) can be written as
= Nm ( x )
m
m
k =0
∑ 2− m +1 k N m ( 2 x − k ) ,
which is called the two scale relation for cardinal B-splines of order m . Chui and Wang [1], introduced the following mth order compactly supported spline wavelet or B-wavelet
ψ m ( x) =
2m−2
1 2
m −1
∑ ( −1)
j
j =0
N 2 m ( j + 1) N 2( m) ( 2 x − j ) , m
(5)
with support [ 0, 2m − 1] that generates W0m and consequently all the wavelet spaces W jm , j ∈ . To verify that ψ m is in V1m , we need the spline identity m j m m N 2( m) ( x ) = ∑ ( −1) j N m ( x − j ) . j =0
(6)
So, substituting (6) into (5), we have the two scale relation 3m − 2
ψ m ( x) =
∑ qk N m ( 2 x − k ) ,
(7)
m N 2 m ( k − j + 1) . j =0 j
(8)
∑h− k N m ( 2 x − k ) ,
(9)
k =0
where,
qk =
( −1) 2
k
m −1
m
∑
Let
ψ m ( x ) = with the corresponding two scale sequence dual wavelet of ψ m such that
{h } . If −k
k
ψ m is a wavelet, then there exists another ψ m called the
ψ m (. − j ) ,ψ m (. − l = ) δ j ,l ,
∀j , l ∈ .
(10)
For the scaling function N m , we define its dual N m by = N m ( x )
∑ g − k N m ( 2 x − k ) ,
k ∈
3003
(11)
S. Khan, M. K. Ahmad
such that
N m (. − j ) , N m (. − l = ) δ j ,l ,
∀j , l ∈ .
(12)
Now, we have
= Nm ( x ) = N m ( x )
m
∑ pk N m ( 2 x − k ) ,
k =0
∑ g − k N m ( 2 x − k ) .
(13)
k ∈
Taking the Fourier transform of (13), we have ω ω Nˆ m (ω ) = Nˆ m , 2 2 ω ω Nˆ m (ω ) = Nˆ m . 2 2
(14)
1 m ∑ pk e−iωk , 2 k =0 1 (ω ) = ∑ g − k e − iω k . 2 k∈
(15)
where,
(ω ) =
A necessary and sufficient condition for the duality relationship (12) is that (ω ) and (ω ) are dual two scale symbols in the sense that
(ω ) (ω ) + ( −ω ) ( −ω= ) 1,
ω∈ .
(16)
A proof of this statement is given in ([15], Theorem 5.22). Also from (7) and (9), we have ω ˆ ω Nm , 2 2 ω ω ψˆ m (ω ) = Nˆ m . 2 2
(17)
1 3m − 2 − iω k ∑ qk e , 2 k =0 1 (ω ) = ∑h− k e −iω k . 2 k∈
(18)
ψˆ m (ω ) =
where,
(ω ) =
We observe that
1 (ω ) (ω ) + ( −ω ) ( −ω ) = (ω )
1 (ω ) + ( −ω ) ( −ω ) = 0 (ω ) (ω ) + ( −ω ) ( −ω ) = (ω ) (ω ) + ( −ω ) ( −ω= ) 0, ω ∈ .
(19)
See ([15], Section 5.3). If N m ( x ) is an orthogonal scaling function, then
N m (.) , N m (. −= l ) δ 0,l ,
∀l ∈ .
(20)
We say that ψ m ( x ) is orthogonal (o.n) B-wavelet function associated with orthogonal scaling function
N m ( x ) if
3004
N m (.) ,ψ m (. − l= ) 0,
S. Khan, M. K. Ahmad
∀l ∈ ,
(21)
and ψ m ( x − l ) , l ∈ is an orthonormal basis of W0m , so we have
ψ m (.) ,ψ m (. −= l ) δ 0,l ,
∀l ∈ .
(22)
Lemma 1 Let N m ( x ) ∈ L2 ( ) . Then N m ( x ) is an orthonormal family if and only if
∑Nˆ m (ω + 2πl ) Nˆ m (ω + 2πl ) =
1,
ω∈.
(23)
l
Proof See ([15], page no. 75]. Theorem 1 Let N m ( x ) defined by (13) is an orthonormal scaling function. Assume that ψ m ( x ) ∈ L2 ( ) whereas (ω ) and (ω ) are defined by (15) and (18) respectively. Then ψ m ( x ) is an orthonormal wavelet function associated with N m ( x ) if and only if (ω ) (ω ) + (ω + π ) (ω + π= ) 0, ω ∈ , (ω ) (ω ) + (ω + π ) (ω + π= ) 1,
(24)
ω ∈ .
Proof Let us suppose that ψ m ( x ) is an orthonormal wavelet function associated with N m ( x ) . By Lemma 1 and (21), we have
0= ∑Nˆ m ( 2ω + 2πl )ψˆ m ( 2ω + 2πl ) l∈
= ∑ (ω + lπ ) Nˆ m (ω + lπ ) Nˆ m (ω + lπ ) (ω + lπ ) l∈
=
1
∑ (ω + ρ π ) ∑ Nˆ m (ω + ρ π + 2ν π ) Nˆ m (ω + ρ π + 2ν π ) (ω + ρ π )
ρ 0 =
ν ∈
= (ω ) (ω ) + (ω + π ) (ω + π ). Again by Lemma 1 and (22), we have
1= ∑ψˆ m ( 2ω + 2πl )ψˆ m ( 2ω + 2πl ) l∈
=
1
∑ (ω + η π ) ∑ Nˆ m (ω + η π + 2ν π ) Nˆ m (ω + η π + 2ν π ) (ω + η π )
η 0 =
ν ∈
= (ω ) (ω ) + (ω + π ) (ω + π ). On the other hand, let (24) holds. Now,
1 N m (.) ,ψ m (. − l ) = Nˆ m (ω ) , e − iωlψˆ m (ω ) 2π 1 ∞ ˆ = N m (ω )ψˆ m (ω ) eiωl dω 2π ∫−∞ 2π (ν +1) 1 = ∑∫ Nˆ m ( 2ω )ψˆ m ( 2ω ) e 2iωl dω 2π π ν ∈ ν 1 π ˆ = ∑ N m ( 2ω + 2ν π )ψˆ m ( 2ω + 2ν π ) e2iωl dω π ∫0 ν ∈ = 0. Also,
ψ m (.) ,ψ= m (. − l )
1 π ) e2iωl dω δ 0,l ∑ψˆ m ( 2ω + 2ν π )ψˆ m ( 2ω + 2ν π= π ∫0 ν ∈
[ by Lemma 1].
Thus, N m ( x ) and ψ m ( x ) are orthogonal and ψ m ( x ) is an orthonormal wavelet function associated with
3005
S. Khan, M. K. Ahmad
Nm ( x ) .
3. B-Spline Wavelet Packets and Their Duals Following Coifman and Meyer [6] [7], we introduce two sequences of L2 functions fined by
∑ζ k(
2 n + λ= ,m ( x )
k ∈
λ)
n,m
n , m ( 2 x − k )= , λ 0,1
, λ ∑ γ −(λk ) n,m ( 2 x − k )=
2 n + λ= ,m ( x )
{ }
and
{ } n,m
de(25) (26)
0,1
k ∈
where n = 0,1, When λ = 0 and n = 0 , we have 0, m N m = ζ k( ) p= k, 0
0, m N m , = γ −( 0k) g= −k ,
and for λ = 1 and n = 0 , we have = ζ k( ) q= 1, m ψ m , k, 1
= γ −(1k) h= 1, m ψ m . −k ,
We call {n , m } the sequence of B-spline wavelet packets induced by the wavelet ψ m and its corresponding scaling function N m whereas n , m denotes the corresponding sequence of dual wavelet packets. By applying the Fourier transformation on both sides of (25), we have
{
}
ω λ ω ˆ2 n + λ , m (ω ) = ( ) ˆn , m 2 2
(27)
where, (
λ)
1
(ω ) = ∑ζ k( λ ) e−ikω 2
(28)
k ∈
= ( ) (ω ) = (ω ) , ( ) (ω ) (ω ) . 0
1
(29)
So, (24) can be written as
(ω ) (1) (ω ) + ( 0) (ω + π ) (1) (ω + π=) 1 1 1 1 ( ) (ω ) ( ) (ω ) + ( ) (ω + π ) ( ) (ω + π= ) (
0)
0, 1,
ω ∈ , ω ∈ .
(30)
Similarly, taking the Fourier transformation on both sides of (26), we have ω ˆ λ ω ˆ 2 n + λ , m (ω ) = ( ) n , m , 2 2
(31)
where, (
λ)
1
(ω ) = ∑ γ −( λk ) e−ikω 2
(32)
k ∈
= ( ) (ω ) = (ω ) , ( ) (ω ) (ω ) . 0
1
(33)
Using these conditions we can write
(
λ)
−ω ) δ λ , µ , ω ∈ , = λ, µ (ω ) ( µ ) (ω ) + ( λ ) ( −ω ) ( µ ) (=
0,1.
(34)
We are now in a position to investigate the properties of B-spline wavelet packets. Theorem 2 Let 0,m ( x ) be any orthonormal scaling function and {n , m ( x )} its corresponding family of
3006
S. Khan, M. K. Ahmad
B-spline wavelet packets. Then for each n ∈ + , we have
n , m (.) , n , m (. − k= ) δ 0,k ,
k ∈ .
(35)
Proof Since 0,m ( x ) = N m ( x ) satisfies (35) for n = 0 . We may proceed to prove (35) by induction. Suppose that (35) holds for all n , where 0 ≤ n < 2r , r a positive integer and 2r ≤ n < 2r +1 . We have n 2r −1 ≤ < 2r , where [ x ] denote the largest integer not exceeding x . By induction hypothesis and Lemma 1, 2 we have
[ n 2], m (.) , [ n 2], m (. − k ) = δ 0, k ⇔
∑ˆ[n 2],m (ω + 2ν π ) ˆ[n 2],m (ω + 2ν π ) = 1.
(36)
ν ∈
By using (27), (30) and (36), we obtain
∑ˆn,m ( 2ω + 2ν π ) ˆn,m ( 2ω + 2ν π )
ν ∈
=
∑(λ ) (ω +ν π ) ˆ[n 2],m (ω +ν π ) ⋅ ˆ[n 2],m (ω +ν π ) (λ ) (ω +ν π )
ν ∈
=
1
λ λ ∑( ) (ω + ρ π ) ∑ˆ[n 2],m (ω + ρ π + 2kπ ) ⋅ ˆ[n 2],m (ω + ρ π + 2kπ ) ( ) (ω + ρ π )
ρ 0 =
= (
k ∈
λ)
(ω ) ( λ ) (ω ) + ( λ ) (ω + π= ) ( λ ) (ω + π )
1.
Hence, by Lemma 1, (35) follows. Theorem 3 Let {n , m ( x )} be a B-spline wavelet packet with respect to the orthonormal scaling function N m ( x ) = 0, m ( x ) . Then for every n ∈ + , we have
2 n , m (.) , 2 n +1, m (. − k ) = 0, k ∈ .
(37)
Proof By (27), (30) and (36), for k ∈ we have
1 2 n , m (.) , 2 n +1, m (. − k ) =∫ ˆ2 n , m (ω ) ˆ2 n +1, m (ω ) eiω k dω 2π 1 ω 1 ω ω 0 ω ( ) ˆn , m ( ) ˆn , m eiω k dω 2π ∫ 2 2 2 2 1 1 0 = ∫( ) (ω ) ˆn , m (ω ) ˆn , m (ω ) ( ) (ω ) e 2iω k dω π 2π (ν +1) 1 0 1 ( ) (ω ) ˆn , m (ω ) ˆn , m (ω ) ( ) (ω ) e 2iω k dω = ∑∫ π ν ∈ 2πν
=
}
{
1 2π ( 0 ) 1 (ω ) ∑ˆn , m (ω + 2ν π ) ˆn , m (ω + 2ν π ) ( ) (ω ) e 2iω k dω ∫ 0 π ν ∈ 1 π ( 0) 1 0 1 (ω ) ( ) (ω ) + ( ) (ω + π ) ( ) (ω= = + π ) e 2iω k dω 0. π ∫0 =
For the family of B-spline wavelet packets 0,m = N m , consider the family of subspaces
{ ( x )} n,m
(
corresponding to some orthonormal scaling function
)
τ nj , m : clos L2 ( ) 2 j 2 Wn , m 2 j x − k : k ∈ , = generated by
{ }
where V jm
j ∈ , n ∈ +
{ } . We observe that n,m
= τ 0,j m V jm ,
j ∈
= τ 1,j m W jm ,
j ∈ ,
is the MRA of L2 ( ) generated by 0,m = N m , and
3007
{W } m j
is the sequence of orthogonal
S. Khan, M. K. Ahmad
complementary (wavelet) subspaces generated by the wavelet 1,m = ψ m . Then the orthogonal decomposition = V jm+1 V jm ⊕W jm ,
∀j ∈
may be written as = τ 0,j +m1 τ 0,j m ⊕τ 1,j m ,
∀j ∈ .
A generalization of the above result for other values of n can be written as
τ nj +, m1 τ 2j n, m ⊕τ 2j n +1, m , =
∀j ∈ .
Theorem 4 For the B-spline wavelet packets, the following two scale relation
(
{
)
(
)
(
1 ∑ pl − 2k 2n,m 2 j x − k + ql − 2k 2n +1,m 2 j x − k 2 k
= n , m 2 j +1 x − l
)}
(38)
holds for all l ∈ . Proof In order to prove the two scale relation, we need the following identity, see ([15], Lemma 7.9)
2δ r ,l . ∑ { pr − 2k pl − 2k + qr − 2k ql − 2k } =
(39)
k
Taking the right-hand side of (38), and applying the identity (39), we have
{
(
(
)
)}
)
(
1 1 2j x − k ∑ pl − 2k 2n,m 2 j x − k + ql − 2k 2n +1,m= ∑∑ { pl pl − 2k + ql ql − 2k } n,m 2 j +1 x − 2k − l 2 k 2 k l 1 =∑∑ { pt − 2 k pl − 2 k + qt − 2 k ql − 2 k } n , m 2 j +1 x − t 2 k t
(
)
1 = ∑ 2 ∑ [ pt − 2k pl − 2k + qt − 2k ql − 2k ] n,m 2 j +1 x − t t k
(
=
(
)
(40)
∑δ t ,l n,m ( 2 j +1 x − t ) . t
)
t= l ⇒ n , m 2 j +1 x − l This completes the proof of the theorem. Next, we discuss the duality properties between the wavelet packets Lemma 2 For all k , l ∈ and n ∈ + , n , m (. − k ) , n , m (. −= l ) δ k ,l ,
{W } n,m
{
}
and Wn , m .
k ∈ .
(41)
Proof We will prove (41) by induction on n . The case n = 0 is the same as our assumption (12) on the dual scaling functions 0,m = N m and 0,m = N m . Suppose that (41) holds for all n where 0 ≤ n < 2r , n where r is a positive integer. Then for 2r ≤ n < 2r +1 , we can write = n 2 + λ for some λ ∈ {0,1} , 2 according to the proof of Theorem 7.24 in [15]. From the Fourier transform formulations of equations (25) and (26) and using (34) we have
1 ˆ i l −k ω n , m (. − k ) , n , m (. − l ) = ∫ ˆn , m (ω ) n , m (ω ) e ( ) dω 2π =
1 ω ˆ ω i l −k ω λ ω λ ω ( ) ( ) ˆ n n e ( ) dω ∫ , , m m 2π 2 2 2 2 2 2
=
1 4π i ( l − k )ω ( λ ) ω ( λ ) ω ω ˆ ω e ∑ˆ n + 2πj ⋅ n + 2πj dω. 2π ∫0 2 2 j∈ 2 , m 2 2 , m 2
n n Since ≤ < 2r , it follows from the induction hypothesis that 2 2 k , l ∈ , and this is equivalent to
3008
n
2 ,m
(. − k ) , n ,m (. − l ) 2
= δ k ,l for all
ω
ˆ ω 2
S. Khan, M. K. Ahmad
1 a.e.. ∑ˆ n ,m 2 + 2πj n ,m 2 + 2πj = j∈
2
(42)
Thus, we have
1 4π i l − k ω λ ω λ ω n , m (. − k ) , n , m (. − l ) = ∫ e ( ) ( ) ( ) dω 2π 0 2 2 1 2π i ( l − k )ω e 2π ∫0
= =
(λ ) ω (λ ) ω (λ ) ω (λ ) ω 2 2 + − 2 − 2 dω
1 2π i ( l − k )ω e dω = δ k ,l . 2π ∫0
This shows that (41) also holds for 2r ≤ n < 2r +1 . Lemma 3 For all k , l ∈ and n ∈ + , and λ , µ ∈ {0,1} , with λ ≠ µ ,
2 n + λ , m (. − k ) , 2 n + λ , m (. − l )= 0,
k ∈ .
(43)
Proof By applying the Fourier transform formulations of Equations (25) and (26) and using (42) and (34), we have as in the proof of Lemma 2 that
1 ˆ i l −k ω 2 n + λ , m (. − k ) , 2 n + λ , m (. − l ) = ∫ˆ2 n + λ , m (ω ) 2 n + λ , m (ω ) e ( ) dω 2π = = = = =
1 ω ˆ ω i l −k ω λ ω λ ω ( ) ( ) ˆn , m n , m e ( ) dω ∫ 2π 2 2 2 2 1 4π i ( l − k )ω ( λ ) ω ( λ ) ω ω ˆ ω e ∑ˆn , m + 2πj n , m + 2πj dω ∫ 0 2π 2 2 j∈ 2 2 1 4π i ( l − k )ω ( λ ) ω ( λ ) ω e dω 2π ∫0 2 2 1 2π i ( l − k )ω e 2π ∫0
(λ ) ω (λ ) ω (λ ) ω (λ ) ω 2 2 + − 2 − 2 dω
1 2π i ( l − k )ω e δ λ , µ dω = δ λ , µ δ k ,l = 0, 2π ∫0
λ ≠ µ , k = l.
References [1]
Chui, C.K. and Wang, J.Z. (1992) On Compactly Supported Spline Wavelets and a Duality Principle. Transactions of the American Mathematical Society, 330, 903-915. http://dx.doi.org/10.1090/S0002-9947-1992-1076613-3
[2]
Battle, G.A. (1987) A Block Spin Construction of Ondelettes, Part-I: Lemarié Functions. Communications in Mathematical Physics, 110, 610-615. http://dx.doi.org/10.1007/BF01205550
[3]
Lemarié, P.G. (1988) Ondelettes á localisation exponentielles, Journal de Mathématiques Pures et Appliquées, 67, 227-236.
[4]
Chui, C.K. and Wang, J.Z. (1991) A Cardinal Spline Approach to Wavelets. Proceedings of the American Mathematical Society, 113, 785-793. http://dx.doi.org/10.1090/S0002-9939-1991-1077784-X
[5]
Chui, C.K. and Wang, J.Z. (1992) A General Framework of Compactly Supported Splines and Wavelets. Journal of Approximation Theory, 71, 263-304. http://dx.doi.org/10.1016/0021-9045(92)90120-D
[6]
Coifman, R.R., Meyer, Y. and Wickerhauser, M.V. (1992) Wavelet Analysis and Signal Processing. In: Ruskai, M.B., et al., Eds., Wavelets and Their Applications, Jones and Barlett, Boston, 153-178.
[7]
Coifman, R.R., Meyer, Y. and Wickerhauser, M.V. (1992) Size Properties of Wavelet Packets. In: Ruskai, M.B., et al., Eds., Wavelets and Their Applications, Jones and Barlett, Boston, 453-547.
[8]
Chui, C.K. and Li, C. (1993) Nonorthogonal Wavelet Packets. SIAM Journal on Mathematical Analysis, 24, 712-738. http://dx.doi.org/10.1137/0524044
[9]
Yang, S. and Cheng, Z. (2000) A Scale Multiple Orthogonal Wavelet Packets. Mathematica Applicata, in Chinese, 13,
3009
S. Khan, M. K. Ahmad 61-65.
[10] Xia, X.G. and Suter, B.W. (1996) Vector Valued Wavelets and Vector Filter Banks. IEEE Transactions on Signal Processing, 44, 508-518. http://dx.doi.org/10.1109/78.489024 [11] Chen, Q. and Cheng, Z. (2007) A Study on Compactly Supported Orthogonal Vector Valued Wavelets and Wavelet Packets. Chaos, Solitons and Fractals, 31, 1024-1034. http://dx.doi.org/10.1016/j.chaos.2006.03.097 [12] Cohen, A. and Daubechies, I. (1993) On the Instability of Arbitrary Biorthogonal Wavelet Packets. SIAM Journal on Mathematical Analysis, 24, 1340-1354. http://dx.doi.org/10.1137/0524077 [13] Feng, Z. and Chen, G. (2001) Nonorthogonal Wavelet Packets with r-Scaling Functions. Journal of Computational Analysis and Applications, 3, 317-330. http://dx.doi.org/10.1023/A:1012046206504 [14] Mallat, S.G. (1989) Multiresolution Approximations and Wavelet Orthonormal Bases of L2 ( R ) . Transactions of the American Mathematical Society, 315, 69-87. http://dx.doi.org/10.2307/2001373 http://dx.doi.org/10.1090/S0002-9947-1989-1008470-5 [15] Chui, C.K. (1992) An Introduction to Wavelets. Academic Press, Inc., Boston.
3010
