CS589-04 Digital Image Processing

Lecture 9. Wavelet Transform

Spring 2008
New Mexico Tech
 Wavelet Definition

“The wavelet transform is a tool that cuts up data, functions

or operators into different frequency components, and
then studies each component with a resolution matched to
its scale”
Dr. Ingrid Daubechies, Lucent, Princeton U.

08/10/22 2
 Fourier vs. Wavelet

► FFT, basis functions: sinusoids

► Wavelet transforms: small waves, called wavelet

► FFT can only offer frequency information

► Wavelet: frequency + temporal information

► Fourier analysis doesn’t work well on discontinuous,

“bursty” data
 music, video, power, earthquakes,…
08/10/22 3
 Fourier vs. Wavelet

► Fourier
 Loses time (location) coordinate completely
 Analyses the whole signal
 Short pieces lose “frequency” meaning

► Wavelets
 Localized time-frequency analysis
 Short signal pieces also have significance
 Scale = Frequency band

08/10/22 4
 Fourier transform

Fourier transform:

08/10/22 5
 Wavelet Transform

► Scale and shift original waveform

► Compare to a wavelet

► Assign a coefficient of similarity

08/10/22 6
 Scaling-- value of “stretch”
► Scaling a wavelet simply means stretching (or
compressing) it.
f(t) = sin(t)
scale factor1

08/10/22 7
 Scaling-- value of “stretch”
► Scaling a wavelet simply means stretching (or
compressing) it.

f(t) = sin(2t)
scale factor 2

08/10/22 8
 Scaling-- value of “stretch”
► Scaling a wavelet simply means stretching (or
compressing) it.
f(t) = sin(3t)
scale factor 3

08/10/22 9
 More on scaling
► It lets you either narrow down the frequency band of interest, or
determine the frequency content in a narrower time interval

► Scaling = frequency band

► Good for non-stationary data

► Low scalea Compressed wavelet Rapidly changing detailsHigh

frequency

► High scale a Stretched wavelet  Slowly changing, coarse features

 Low frequency

08/10/22 10
 Scale is (sort of) like frequency

Small scale
-Rapidly changing details,
-Like high frequency

Large scale
-Slowly changing
details
-Like low frequency

08/10/22 11
 Scale is (sort of) like frequency

The scale factor works exactly the same with wavelets.

The smaller the scale factor, the more "compressed"
the wavelet.
08/10/22 12
 Shifting

Shifting a wavelet simply means delaying (or hastening) its

onset. Mathematically, delaying a function  f(t)   by k is
represented by f(t-k)

08/10/22 13
 Shifting

C = 0.0004

C = 0.0034

08/10/22 14
Five Easy Steps to a Continuous Wavelet
Transform

1. Take a wavelet and compare it to a section at the start of

the original signal.
2. Calculate a correlation coefficient c

08/10/22 15
Five Easy Steps to a Continuous Wavelet
Transform
3. Shift the wavelet to the right and repeat steps 1 and 2 until you've
covered the whole signal.

4. Scale (stretch) the wavelet and repeat steps 1 through 3.

5. Repeat steps 1 through 4 for all scales.

08/10/22 16
 Coefficient Plots

08/10/22 17
 Discrete Wavelet Transform

► “Subset” of scale and position based on power of two

 rather than every “possible” set of scale and position in
continuous wavelet transform

► Behaves like a filter bank: signal in, coefficients out

► Down-sampling necessary
(twice as much data as original signal)

08/10/22 18
 Discrete Wavelet transform

signal

lowpass highpass

filters

Approximation Details
(a) (d)

08/10/22 19
 Results of wavelet transform
— approximation and details
► Low frequency:
 approximation (a)

► High frequency
 details (d)

► “Decomposition”
can be performed
iteratively

08/10/22 20
 Wavelet synthesis

•Re-creates signal from coefficients

•Up-sampling required
08/10/22 21
 Multi-level Wavelet Analysis

Multi-level wavelet
decomposition tree Reassembling original signal

08/10/22 22
 Subband Coding

08/10/22 23
 2-D 4-band filter bank

Approximation

Vertical detail

Horizontal detail

Diagonal details

08/10/22 24
 An Example of One-level Decomposition

08/10/22 25
 An Example of Multi-level Decomposition

08/10/22 26
 Wavelet Series Expansions

2
Wavelet series expansion of function f ( x)  L ( )
relative to wavelet  ( x) and scaling function  ( x)

f ( x)   c j0 (k ) j0 ,k ( x)    d j (k ) j ,k ( x)
k j  j0 k

where ,
c j0 (k ) : approximation and/or scaling coefficients
d j (k ) : detail and/or wavelet coefficients

08/10/22 27
 Wavelet Series Expansions

c j0 (k )  f ( x),  j0 ,k ( x)   f ( x) j0 ,k ( x )dx

and
d j (k )  f ( x), j ,k ( x)   f ( x) j ,k ( x)dx

08/10/22 28
 Wavelet Transforms in Two Dimensions
 ( x, y )   ( x) ( y )
 H ( x, y )   ( x) ( y )  j ,m ,n ( x, y )  2 j /2  (2 j x  m, 2 j y  n)
 V ( x, y )   ( x) ( y )  i j ,m ,n ( x, y )  2 j /2 i (2 j x  m, 2 j y  n)
 D ( x, y )   ( x) ( y ) i  {H ,V , D}

1 M 1 N 1
W ( j0 , m, n) 
MN
  f ( x, y)
x 0 y 0
j0 , m , n ( x, y )

1 M 1 N 1
i
W ( j , m, n) 

MN

x 0 y 0
f ( x, y ) i j ,m ,n ( x, y )

i  H ,V , D
08/10/22 29
 Inverse Wavelet Transforms in Two
Dimensions

1
f ( x, y ) 
MN m n
 W ( j0 , m, n) j0 , m,n ( x, y )

1 
   
MN i  H ,V , D j  j0 m n
i i
W  ( j, m, n) j ,m,n ( x, y )

08/10/22 30