Assignment on

Wavelets and MRA

By: Anant Milan Khalkho
Roll No.: PHD/EE/10002/18
What is Transform

First of all, why do we need a transform, or what is a transform anyway?

Mathematical transformations are applied to signals to obtain further information from that
signal that is not readily available in the raw signal. In the following tutorial I will assume a
time-domain signal as a raw signal, and a signal that has been "transformed" by any of the
available mathematical transformations as a processed signal.
There are a number of transformations that can be applied, among which the Fourier transforms
are probably by far the most popular.
Most of the signals in practice, are TIME-DOMAIN signals in their raw format. That is,
whatever that signal is measuring, is a function of time. In other words, when we plot the signal
one of the axes is time (independent variable), and the other (dependent variable) is usually the
amplitude. When we plot time-domain signals, we obtain a time-amplitude representation of
the signal. This representation is not always the best representation of the signal for most signal
processing related applications. In many cases, the most distinguished information is hidden in
the frequency content of the signal. The frequency SPECTRUM of a signal is basically the
frequency components (spectral components) of that signal. The frequency spectrum of a signal
shows what frequencies exist in the signal.

To obtain the frequency spectrum of the signal we perform Fourier analysis of the signal which
is done to obtain the magnitude of the frequency components in the signal. This can be
pictorially shown as in Fig. 1.

Fig.1 Splitting of white light into different fundamental components

It can be seen from the figure that as the prism is used to split the white light into different
component of colors the and the split white light can be used to get back the original signal
through the same prism similarly fourier transform is used to obtain various frequency
components and similarly inverse fourier transform can be used to obtain the original signal from
the split frequency component.

Difficulties faced with Fourier Transform

• Fourier series of continuous time periodic signal is given by:

1 𝑇
𝑐𝑘 = 𝑇 ∫0 𝑝 𝑥(𝑡)𝑒 −𝑗2𝜋𝐹0 𝑡 𝑑𝑡 (1)
𝑝

• By using Fourier we loose the time information about a signal as to when did a particular
event take place.

• It cannot locate drift, sudden changes, beginning and end of events etc.

• Use of complex number for calculation.

How do we obtain time information

We can obtain time information of the signal using the Kronecker delta function which is given
as.

1, 𝑘 = 𝑡
𝜓𝑘 = 𝛿𝑘 (𝑡) = { (2)
0, 𝑘 ≠ 𝑡

This function can show the time information of the signal but it cannot show the frequency
information of the signal.

Fig. 2 Kronecker delta function

Why do we need frequency information?
Often times, the information that cannot be readily seen in the time-domain can be seen in the
frequency domain.
Let's give an example from biological signals. Suppose we are looking at an ECG signal
(ElectroCardioGraphy, graphical recording of heart's electrical activity). The typical shape of a
healthy ECG signal is well known to cardiologists. Any significant deviation from that shape is
usually considered to be a symptom of a pathological condition.
This pathological condition, however, may not always be quite obvious in the original time-
domain signal. Cardiologists usually use the time-domain ECG signals which are recorded on
strip-charts to analyze ECG signals. Recently, the new computerized ECG recorders/analyzers
also utilize the frequency information to decide whether a pathological condition exists. A
pathological condition can sometimes be diagnosed more easily when the frequency content of
the signal is analyzed.

Stationary and Non-Staionary Signals

FT gives the frequency information of the signal, which means that it tells us how much of each
frequency exists in the signal, but it does not tell us when in time these frequency components
exist. This information is not required when the signal is so-called stationary.

Let's take a closer look at this stationarity concept more closely, since it is of paramount
importance in signal analysis. Signals whose frequency content do not change in time are called
stationary signals. In other words, the frequency content of stationary signals do not change in
time. In this case, one does not need to know at what times frequency components exist , since
all frequency components exist at all times.
Fig.3 Stationary Signal and its Fourier Transform

Note that the Fourier transform tells whether a certain frequency component exists or not. This
information is independent of where in time this component appears. It is therefore very
important to know whether a signal is stationary or not, prior to processing it with the FT.

Look at the Fig. 3, which shows the signal:

x(t)=cos(2*pi*5*t)+cos(2*pi*10*t)+cos(2*pi*20*t)+cos(2*pi*50*t) (3)

that is , it has four frequency components of 5, 10, 20, and 50 Hz., all occurring at all times. Its
FT is also Shown in the figure.
Fig. 3 Non-stationary signal and its Fourier Transform.

Now, look at Fig. 4: Here the signal is again the cosine signal, and it has the same four
frequencies. However, these components occur at different times. The FT of the signal is also
shown in the subsequent figure.

We see in the above figure, it is (almost) same with the previous FT figure. There are major four
peaks corresponding to 5, 10, 20, and 50 Hz, and all the peaks occur at different times, But this
time information is lost.

Short-time Fourier Transform

Using STFT we can obtain the time information of the signal. In STFT, the signal is divided into
small enough segments, where these segments (portions) of the signal can be assumed to be
stationary. For this purpose, a window function "w" is chosen. The width of this window must
be equal to the segment of the signal where the signal is stationary. Fig.4 illustrates how STFT is
performed and Fig.5 shows STFT using Gaussiuan window.

Fig.4 How window is used in STFT

Fig.5 STFT using Gaussian window

STFT is compromise between time frequency representation of the signal, here both time and
frequency are represented in limited precision. The precision is represented by the size of the
window, a particular size of the window is same for all the frequency.

The equation for STFT is given as:


STFTXw (t , f )   xt .wt  t exp  j 2ft dt


What is wrong with STFT

The problem with STFT is the fact whose roots go back to what is known as the Heisenberg
Uncertainty Principle. We cannot know what spectral components exist at what instances of
times. What one can know are the time intervals in which certain band of frequencies exist,
which is a resolution problem i.e narrow window has poor frequency resolution and wide
window has poor time resolution. Fig.6 illustrates this problem.

Fig. 6 STFT resolution problem

Multi-Resolution Analysis
• MRA is designed to give good time resolution and poor frequency resolution at high
frequencies and good frequency resolution and poor time resolution at low frequencies.

• This approach makes sense especially when the signal at hand has high frequency
components for short durations and low frequency components for long durations.

• The signals that are encountered in practical applications are often of this type.

Wavelet Analysis
A wavelet is a waveform of effectively limited duration that has an average value of zero. Fig.7
shows Daubechies 10 wavelet.

Fig.7 db10 wavelet

Wavelet Transform is An alternative approach to the short time Fourier transform to overcome
the resolution problem.

Properties of wavelet transform

 Short time localized waves with zero integral value.

 Possibility of time shifting.

 Flexibility.

The continous wavelet transform is given by the equation:

  t  
CWTx  , s    xt .
1
 dt
*
(4)
s 
 s 

Where ψ(t) is the mother wavelet.

The parameter scale in the wavelet analysis is similar to the scale used in maps. As in the case of
maps, high scales correspond to a non-detailed global view (of the signal), and low scales
correspond to a detailed view. Similarly, in terms of frequency, low frequencies (high scales)
correspond to a global information of a signal (that usually spans the entire signal), whereas high
frequencies (low scales) correspond to a detailed information of a hidden pattern in the signal
(that usually lasts a relatively short time).

Fig. 7 Wavelet analysis for s=1.

The wavelet is placed at the beginning of the signal at the point which corresponds to time=0.
The wavelet function at scale ``1'' is multiplied by the signal and then integrated over all times.
The result of the integration is then multiplied by the constant number 1/sqrt{s}. This
multiplication is for energy normalization purposes so that the transformed signal will have the
same energy at every scale. The final result is the value of the transformation, i.e., the value of
the continuous wavelet transform at time zero and scale s=1. In other words, it is the value that
corresponds to the point tau =0 , s=1 in the time-scale plane.
The wavelet at scale s=1 is then shifted towards the right by tau amount to the location t=tau ,
and the above equation is computed to get the transform value at t=tau , s=1 in the time-
frequency plane.

Let us consider scale s=1 Fig. 7 illustrates how the window is shifted and the wavelet analysis is
done. Similarly Fig.8 shows wavelet analysis for s=5.

Fig. 8 Wavelet analysis for s=5.

TIME AND FREQUENCY RESOLUTIONS
Fig. 9 is used to explain how time and frequency resolutions should be interpreted. Every box in
Fig. 9 corresponds to a value of the wavelet transform in the time-frequency plane. The boxes
have a certain non-zero area, which implies that the value of a particular point in the time-
frequency plane cannot be known. All the points in the time-frequency plane that falls into a box
are represented by one value of the WT. Although the widths and heights of the boxes change,
the area is constant. That is each box represents an equal portion of the time-frequency plane, but
giving different proportions to time and frequency.

Fig. 9 Time frequency resolution.

Application example (Alzheimer’s Disease Diagnosis)

The signals shown below are drawn from a database signals that includes event related potentials
of normal people, and patients with Alzheimer's disease. Since these are not test signals like
simple sinusoids, it is not as easy to interpret them. They are shown here only to give an idea of
how real-life CWTs look like.
Fig. 10 Waveforms of Normal person and corresponding WT
Fig. 11 Waveforms of person suffering from Alzheimer’s Disease and corresponding WT

Fig.10 shows the waveform corresponding to normal person and its corresponding wavelet
transform and Fig.11 shows waveform corresponding to a person suffering from Alzheimer's
disease and its corresponding wavelet transform.
Discrete Wavelet Transform (DWT)
The continuous wavelet transform was computed by changing the scale of the analysis window,
shifting the window in time, multiplying by the signal, and integrating over all times.

In the discrete case, filters of different cutoff frequencies are used to analyze the signal at
different scales. The signal is passed through a series of highpass filters to analyze the high
frequencies, and it is passed through a series of lowpass filters to analyze the low frequencies.

 The resolution of the signal, which is a measure of the amount of detail information in the
signal, is changed by the filtering operations
 The scale is changed by upsampling and downsampling operations.

Fig.12 DWT Tree

Fig.12 shows DWT tree, the process halves time resolution, but doubles frequency resolution.
Here g[n] and h[n] are quadrature mirror filter.
References
[1] https://www.eecis.udel.edu/~amer/CISC651/IEEEwavelet.pdf

[2] P. M. Bentley and J. T. E. McDonnell, "Wavelet transforms: an introduction," in Electronics

& Communication Engineering Journal, vol. 6, no. 4, pp. 175-186, Aug. 1994.

[3] S. Mallat, Academic press - A Wavelet Tour of Signal Processing 2/E. San Diego, Ca:
Academic Press, 1999, pp. 2-121

[4] http://web.iitd.ac.in/~sumeet/WaveletTutorial.pdf

[5] https://www.math.aau.dk/digitalAssets/120/120646_r-2003-24.pdf

[6] http://www.wavelet.org

[7] Robi Polikar, Multiresolution Wavelet Analysis of Event Related Potentials for the Detection
of Alzheimer's Disease, Iowa State University, 06/06/1995

[8] Amara Graps, An Introduction to Wavelets, IEEE Computational Sciences and Engineering,
Vol. 2, No 2, Summer 1995, pp 50-61.

[9]Resonance Publications, Inc. Wavelets. http://www.resonancepub.com/wavelets.htm

[10] R. Crandall, Projects in Scientific Computation, Springer-Verlag, New York, 1994, pp. 197-
198, 211-212.

[11] Y. Meyer, Wavelets: Algorithms and Applications, Society for Industrial and Applied
Mathematics, Philadelphia, 1993, pp. 13-31, 101-105.

[12] G. Kaiser, A Friendly Guide to Wavelets, Birkhauser, Boston, 1994, pp. 44-45.

[13] W. Press et al., Numerical Recipes in Fortran, Cambridge University Press, New York,
1992, pp. 498-499, 584-602.

[14] M. Vetterli and C. Herley, "Wavelets and Filter Banks: Theory and Design," IEEE
Transactions on Signal Processing, Vol. 40, 1992, pp. 2207-2232.

[15] I. Daubechies, "Orthonormal Bases of Compactly Supported Wavelets," Comm. Pure Appl.
Math., Vol 41, 1988, pp. 906-966.