where the series converge strongly.
Moreover, the
and
constitute two Riesz
bases, with
if and only if
This theorem tells us that under some technical conditions, we can
expand functions using the wavelets and reconstruct using their duals.The multiresolution formulations in
Chapter: A multiresolution formulation of Wavelet Systems can be revised as
where
If
[link] holds, we have
where
Although
is not the orthogonal complement to
in
as before, the dual space
plays the
much needed role. Thus we have four sets of spaces that form twohierarchies to span
.
In
Section: Further Properties of the Scaling Function and Wavelet , we have a list of properties of the scaling
function and wavelet that do not require orthogonality. The results forregularity and moments in
Chapter: Regularity, Moments, and Wavelet System Design can also be generalized to the
biorthogonal systems.
Comparisons
of orthogonal and biorthogonal wavelets
The biorthogonal wavelet systems generalize the classical orthogonal
wavelet systems. They are more flexible and generally easy todesign. The differences between the orthogonal and biorthogonal
wavelet systems can be summarized as follows.
- The orthogonal wavelets filter and scaling filter must be of the same
length, and the length must be even. This restriction has been greatlyrelaxed for biorthogonal systems.
- Symmetric wavelets and scaling functions are possible in the
framework of biorthogonal wavelets. Actually, this is one of the mainreasons to choose biorthogonal wavelets over the orthogonal ones.
- Parseval's theorem no longer holds in biorthogonal wavelet
systems; i.e., the norm of the coefficients is not the same as the norm ofthe functions being spanned. This is one of the main disadvantages of
using the biorthogonal systems. Many design efforts have been devoted tomaking the systems near orthogonal, so that the norms are close.
- In a biorthogonal system, if we switch the roles of the primary and
the dual, the overall system is still sound. Thus we can choose the bestarrangement for our application. For example, in image compression,
we would like to use the smoother one of the pair to reconstruct thecoded image to get better visual appearance.
- In statistical signal processing, white Gaussian noise remains white
afterorthogonal transforms. If the transforms are nonorthogonal, the noise
becomes correlated or colored. Thus, when biorthogonal wavelets are usedin estimation and detection, we might need to adjust the algorithm to
better address the colored noise.
Example families of biorthogonal systems
Because biorthogonal wavelet systems are very flexible, there are a wide variety
of approaches to design different biorthogonal systems. The key is todesign a pair of filters
and
that satisfy
[link] and
[link] and have other
desirable characteristics. Here we review several families of biorthogonalwavelets and discuss their properties and design methods.