Biorthogonal Wavelets

Biorthogonal Wavelets
Ref: Rao & Bopardikar, Ch.4

Jyun-Ming Chen
Spring 2001
Why is orthogonality useful

x 1a 1 2 a 2
x 1 1
a 1 2 1
a 2 1 2
1 x, a1 / a 1 , a 1 3 / 5
2 x, a 2 / a 2 , a 2 1 / 5
Orthonormal bases further simplify the
computation
Ortho v. Non-Ortho Basis
Sum of projection vectors !?
Dual Basis
a1 2 1
a 2 1 2
a 1 , b1
a 2 , b2
a1 , b 2
a 2 , b1
x 1 1
T
T
x 1a1 2 a 2
1 x, b1 / a 1 , b1 1 / 3
2 x, b 2 / a 2 , b 2 1 / 3
1
1
0
0
Dual Bases
b1 2 / 3 1 / 3
T
b 2 1 / 3 2 / 3
T
a1-a2 and b1-b2 are

biorthogonal
Dual Basis (cont)

a 1 1 1 2
a 2 0 1 0
b1 1 0 0
b 2 1 1 0
Verify
duality !
Dual basis may generate different spaces

Here: a1-a2 and b1-b2 generate two different 2D subspaces in Euclidean
3space.
Semiorthogonal:
For dual basis that generates the same subspace
Orthogonal:
Primal and dual are the same bases
Extend to Function Space

MRA types:
orthogonal, semiorthogonal, biorthognal
Extend the concept to using biorthogonal

MRA
More flexible design
Lifting scheme: a general design method for
biorthogonal wavelets
Alternative Wavelets:
Proposed by Cohen (1992)
Characteristics of Orthogonal Basis

Decomposition and
reconstruction filters are
FIR and have the same
length
Generally do not have
closed-form expressions
Usually not symmetric
(linear phase)
Haar wavelet is the only
real-valued wavelet that is
compactly supported,
symmetric and orthogonal
Higher-order filters (with

more coefficients) have
poor time-frequency
localization
Desired property: perfect
reconstruction FIR
symmetric (linear-phase)
filters
Not available in orthogonal
bases
The Need for Biorthogonal Basis

delegate the
responsibilities of
analysis and synthesis
to two different
functions (in the
biorthogonal case) as
opposed to a single
function in the
orthonormal case
more design freedom
compactly supported
symmetric analyzing
and synthesis wavelets
and scaling functions
Biorthogonal Scaling Functions
(t ) 2 h(n) (2t n)
n
~
(t ), (t k ) (k )
dual
~ ~
~
(t ) 2 h (n) (2t n)
n
~ k
(2 t ), (2 t n) 2 k (n)
k
~
Two sequences h(n) and h (n)serve as impulse
response of FIR filters

~
(t ) and (t ) generate
Two sets of scaling
functions
~
subspaces Vk and Vk respectively
The basis are orthogonal; the two MRAs are said
to be biorthogonal to each other
Dual MRA (cont)

Basis of
~
Vk and Vk
Translated copy of appropriate dilation of

~
(t ) and (t )
V0 V1 V2 V3
~
~ ~
~
V0 V1 V2 V3
Function approximation in
subspaces
Coarser approx f 0 (t )
a(0, n) (t n)
n
~
a (0, n) f (t ), (t n)
Finer approx f1 (t )
coefficients obtained by
~
projecting f (t ) on V0
a(1, n) (2t n)
n
~
a (1, n) 2 f (t ), (2t n)
~
~
f (t ), (2t n) a(1, n) (2t n), (2t n)
1
a(1, n)
2
Relation between Finer and

Coarser Coefficients
~
~
~
(t n) 2 h (m) (2t 2n m)
~
a (0, n) f (t ), (t n)
~
~
a (0, n) 2 h (m) f (t ), (2t 2n m)
m
~
a (1, n) 2 f (t ), (2t n)
~
a (1,2n m) 2 f (t ), (2t 2n m)
~
a (0, n) h (m)a (1,2n m)
m
~~
aa((00,,nn))
aa((11,,mm))hh((mm22nn))
n ( 2 n m )
mm
(t )dt 0
~ (t )dt 0
~ (t )
(
t
)
and
(t k ) : k Z spans W0 Two sets of wavelets

~
W
and
W
generate subspaces k
k
~
~
(t k ) : k Z spans W0 respectively
The basis are orthogonal; the two
MRAs are said to be
Require:
biorthogonal to each other
(t ),~ (t k ) (k ) Dual
~
(t ), (t n) 0 wavelet dual scaling fns
~ (t ), (t n) 0 dual wavelet scaling fns
Two-scale relations of wavelet:

primal and dual
(t ) 2 g (n) (2t n)
n
~
~
~
(t ) 2 g (n) (2t n)
n
Function Projection
detail fn : g 0 (t ) f1 (t ) f 0 (t ) b(0, n) (t n)
n
b(0, n) g 0 (t ),~ (t n) f1 (t ) f 0 (t ),~ (t n)

f1 (t ),~ (t n) f 0 (t ),~ (t n)
f (t ),~ (t n)
1
f1 (t ) a (1, m) (2t m)
m
~
~
(t n) 2 g~ (l ) (2t 2n l )
m=2n+l
f 0 (t ) a(0, n) (t n)
n
and ~ (t ), (t n) 0
~~(m 2n)
bb((00,,nn))
a
(
1
,
m
)
g
a(1, m) g (m 2n)
mm
Function Reconstruction
f1 (t ) f 0 (t ) g 0 (t )
a (0, l ) (t l ) b(0, l ) (t l )
l
2 a (0, l ) h(m) ( 2t 2l m) 2 b(0, l ) g (m) (2t 2l m)

m
Substituting n 2l m
f1 (t ) 2 a (0, l )h(n 2l ) (2t n) 2 b(0, l ) g (n 2l ) (2t n)
n
f1 (t ) a (1, n) (2t n)
n
Hence
a (1, n) 2 a (0, l )h(n 2l ) 2 b(0, l ) g (n 2l )
l
Filter Bank
Primal and Dual MRA

(biorthogonal)
~
VN
VN
VN-1
VN-2
VN-3 WN-3
WN-2
~
WN -1
~
VN -1
WN-1
~
VN -2
~
WN -2
~
~
VN -3 WN -3
~~ ~~
VVkk W
Wkk VVkk W
Wkk
Filter Relations
(between primal and dual)
~
(t ), (t n) (n)
~
~
4 h( p)h ( q ) (2t p ), (2t 2n q )
p
~
2 h( p)h ( q ) ( p 2n q )
~
2 h ( m 2 n ) h ( m )
m
~~
((nn))
hh((mm22nn))hh((mm))
22
mm
Similarly,
(t ),~ (t n) 0
((nn))
~
~
gg((mm22nn))gg((mm))
22
mm
only term left :

p 2n q 0
p 2n q
Filter Relations (cont)

(t ),~ (t n) 0
~
4 h( p) g~ (q ) (2t p ), (2t 2n q )
p
2 h( p) g~ (q ) ( p 2n q ) 2 h(m 2n) g~ (m)

p
~~(m) 0
h
(
m
2
n
)
g
h ( m 2n) g ( m) 0
mm
Similarly,
~
(t ), (t n) 0
~~
hh((mm22nn))gg((mm))00
mm
Design of Biorthogonal Wavelets

because there is quite a bit of freedom in
designing the biorthogonal wavelets, there
are no set steps in the design procedure.
Lifting (Sweldens 94): a scheme for customdesign biorthogonal wavelets
Special Cases:
orthogonal and semiorthogonal
Common property:
VN
Vk Wk Vk 1 Vk Wk
VN-1
Differences:
if orthogonal: scaling
functions (and wavelets) of
the same level are
orthogonal to each other
If semiorthogonal, wavelets
of different levels are
orthogonal (from nested
space)
WN-1
WN-2
VN-2
VN-3 WN-3
~
~
Vk Wk and Wk Wk
Dual and
primal are the
same

Biorthogonal Wavelets

Uploaded by

Copyright:

Available Formats

Biorthogonal Wavelets

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Biorthogonal Wavelets

Uploaded by

Copyright:

Available Formats

Biorthogonal Wavelets

Ref: Rao & Bopardikar, Ch.4

Why is orthogonality useful

Ortho v. Non-Ortho Basis

Sum of projection vectors !?

a1-a2 and b1-b2 are

Dual Basis (cont)

Dual basis may generate different spaces

Extend to Function Space

Extend the concept to using biorthogonal

Characteristics of Orthogonal Basis

Higher-order filters (with

The Need for Biorthogonal Basis

Biorthogonal Scaling Functions

response of FIR filters

Dual MRA (cont)

Translated copy of appropriate dilation of

Relation between Finer and

(t k ) : k Z spans W0 Two sets of wavelets

Two-scale relations of wavelet:

b(0, n) g 0 (t ),~ (t n) f1 (t ) f 0 (t ),~ (t n)

2 a (0, l ) h(m) ( 2t 2l m) 2 b(0, l ) g (m) (2t 2l m)

Primal and Dual MRA

only term left :

Filter Relations (cont)

2 h( p) g~ (q ) ( p 2n q ) 2 h(m 2n) g~ (m)

Design of Biorthogonal Wavelets

You might also like