Connexions module: m10764 1
Haar Wavelet Basis
∗Roy Ha Justin Romberg
This work is produced by The Connexions Project and licensed under the Creative Commons Attribution License†
Abstract
This module gives an overview of wavelets and their usefulness as a basis in image processing. In particular we look at the properties of the Haar wavelet basis.
1 Introduction
Fourier series1is a useful orthonormal representation2on L2([0, T ])especiallly for inputs into LTI systems.
However, it is ill suited for some applications, i.e. image processing (recall Gibb's phenomena3).
Wavelets, discovered in the last 15 years, are another kind of basis for L2([0, T ]) and have many nice properties.
2 Basis Comparisons
Fourier series - cn give frequency information. Basis functions last the entire interval.
∗Version 2.7: Jul 19, 2006 4:38 pm GMT-5
†http://creativecommons.org/licenses/by/1.0
1"Fourier Series: Eigenfunction Approach" <http://cnx.org/content/m10496/latest/>
2"Orthonormal Basis Expansions" <http://cnx.org/content/m10760/latest/>
3"Gibbs's Phenomena" <http://cnx.org/content/m10092/latest/>
Connexions module: m10764 2
Figure 1: Fourier basis functions
Wavelets - basis functions give frequency info but are local in time.
Connexions module: m10764 3
Figure 2: Wavelet basis functions
In Fourier basis, the basis functions are harmonic multiples of eiω0t
Connexions module: m10764 4
Figure 4
Basis functions {ψj,k(t)} are indexed by a scale j and a shift k.
Let ∀, 0 ≤ t < T : (φ (t) = 1) Then n
φ (t) , 2j2ψ 2jt − k |j ∈ Z ∧ k = 0, 1, 2, . . . , 2j− 1o
Connexions module: m10764 5
Figure 5
ψ (t) =
1if 0 ≤ t < T2
−1if 0 ≤ T2 < T (1)
Connexions module: m10764 6
Figure 6
Let ψj,k(t) = 22jψ 2jt − k
Figure 7
Larger j → "skinnier" basis function, j = {0, 1, 2, . . . }, 2j shifts at each scale: k = 0, 1, . . . , 2j− 1 Check: each ψj,k(t)has unit energy
Connexions module: m10764 7
Figure 8
Z
ψj,k2(t) dt = 1 ⇒ k ψj,k(t) k2= 1 (2)
Any two basis functions are orthogonal.
Connexions module: m10764 8
Also, {ψj,k, φ}span L2([0, T ])
3 Haar Wavelet Transform
Using what we know about Hilbert spaces5: For any f (t) ∈ L2([0, T ]), we can write Synthesis
f (t) =X
j
X
k
(wj,kψj,k(t))
!
+ c0φ (t) (3)
Analysis
wj,k= Z T
0
f (t) ψj,k(t) dt (4)
c0= Z T
0
f (t) φ (t) dt (5)
note: the wj,k are real
The Haar transform is super useful especially in image compression Example 1
This demonstration lets you create a signal by combining Haar basis functions, illustrating the synthesis equation of the Haar Wavelet Transform. See here6 for instructions on how to use the demo.
This is an unsupported media type. To view, please see http://cnx.org/content/m10764/latest/HaarSyn.llb
5"Inner Products" <http://cnx.org/content/m10755/latest/>
6"How to use the LabVIEW demos" <http://cnx.org/content/m11550/latest/>