CHAPTER 5 WAVELET TRANSFORM

5.1 Introduction Wavelets are a set of functions that can be used effectively in a number of situations, to represent natural, highly transient phenomena that result from a dilation and shift of the original waveform. Wavelet Transform represents a powerful signal processing with a wide variety of applications that is particularly useful for the analysis of non- stationary signals. 5.2 Wavelet Frames Wavelet and frame analysis is found in such diverse fields as data compression, pseudodifferential operator theory and applied statistics. Frame bound gaps, smooth Parseval wavelet frames, generalized shearlets, Grassmannian fusion frames and p-adic wavlets. The phenomenon of a frame bound gap occurs when certain sequences of functions, converging in L2 to a Parseval frame wavelet, generate systems with frame bounds that are uniformly bounded away from 1. In the 90's, Bin Han proved the existence of Parseval wavelet frames which are smooth and compactly supported on the frequency domain and also approximate wavelet set wavelets. 5.3 Sub-Band Filtering Schemes For many signals, the low-frequency content is the most important part. It is what gives the signal its identity. The approximations are the high-scale, low-frequency components of the signal. The details are the low-scale, high-frequency components. The filtering process, at its most basic level, looks in Fig. 5.1.

Fig. 5.1 Basic Filtering Process

5.4 Biorthogonal System of Wavelets An optimal wavelet sampling approximation, vanishing moments should be equally distributed between the scaling function and the wavelet function, and the approximation error converges to zero with exponential decay. Thus vanishing moments of the wavelet function is necessary for the smoothness of the wavelet system; vanishing moments of the scaling function will improve the symmetry; and equal vanishing moments distribution will provide an optimal wavelet sampling approximation with exponential decay. Based on these observations, we introduce biorthogonal Coifman wavelet systems, a family of compactly supported biorthogonal wavelet systems with the vanishing of moments equally distributed between scaling functions and wavelet functions. Biorthogonal Coifman wavelet systems provide multiplication-free discrete wavelet transform, and converge to the noncompactly supported sine wavelet system. 5.5 Multi Resolution Analysis A multiresolution analysis (MRA) or multiscale approximation (MSA) is the design method of most of the practically relevant discrete wavelet transforms (DWT) and the first component to multiresolution analysis is vector spaces. For each vector space, there is another vector space of higher resolution until you get to the final signal. Also, each vector space contains all vector

spaces that are of lower resolution. The basis of each of these vector spaces is the scale function for the wavelet. 5.6 Wavelet and Scaling Functions Wavelets are functions in continuous times that have special properties; usually the letter is used for the wavelet function. The term wavelet means a small wave. The smallness refers to the condition that this (window) function is of finite length (compactly supported). The wave refers to the condition that this function is oscillatory. The term mother implies that the functions with different region of support that are used in the transformation process are derived from one main function, or the mother wavelet.

Fig. 5.2 Different Families of Wavelet

The wavelet function is determined by the highpass filter, which also produces the details of the wavelet decomposition. There is an additional function associated with some but not all wavelets. This is the so-called scaling function, . The scaling function is very similar to the wavelet function. It is determined by the lowpass quadrature mirror filters, and thus is associated with the approximations of the wavelet decomposition. In the same way that iteratively upsampling and convolving the highpass filter produces a shape are approximating the wavelet function, iteratively Upsampling and convolving the lowpass filter produces a shape approximating the scaling function. An example of mother
wavelet can be seen in Fig. 5.2.

Scaling, as a mathematical operation, either dilates or compresses a signal. Smaller scales correspond to dilated (or stretched out) signals and large scales correspond to compressed signals. The relation between scale and frequency is that low scales correspond to high frequencies and high scales to low frequencies. The different wavelet families make different trade-off between how compactly the basis functions are localized in space and their smoothness. The scaling function can be calculated using the dilation equation as follows. (5.1)

(t) = 2
Where, ho Low pass coefficients

h( k) (2t
0 k =0

k)

N Number of filter coefficients Scaling function As the dilation equation is recursive to itself, there is not always a solution for . The scaling function is a function in continuous time, but is not likely to be continuous; rather it may not be smooth and even contain jumps. Finally, the wavelet function y can be calculated from the scaling function with the wavelet equation.


(t) =2
Where, H1 High pass filter coefficients N Number of High pass filter coefficients Wavelet function

(5.2)

h (k)
1 k =0

(2t k)

It can be seen that once the scaling function is known, the mother wavelet can be calculated directly without recursion. 5.7 Iterated Filter Banks The original signal, S passes through two complementary filters and emerges as two signals. Unfortunately, if we actually perform this operation on a real digital signal, we wind up with twice as much data as we started with. Suppose, for instance, that the original signal S consists of 800 samples of data. Then the approximation and the detail will each have 800 samples, for a total of 1600.To correct this problem, we introduce the notion of downsampling. This simply means throwing away every second data point. While doing this introduces aliasing (a type of error) in the signal components, it turns out we can account for this later on in the process. The process on the right, which includes downsampling, produces DWT coefficients. To gain a better appreciation of this process, lets perform a one-stage discrete wavelet transform of a signal. Our signal will be a pure sinusoid with high-frequency noise added to it. Here, Fig. 5.3 shows schematic diagram with real signals inserted into it.

Fig. 5.3 Schematic Diagram of Stage Filtering

Notice that the detail coefficients cD consist mainly of the high-frequency noise, while the approximation coefficients cA contain much less noise than does the original signal. We may observe that the actual lengths of the detail and approximation coefficient vectors are slightly more than half the length of the original signal. This has to do with the filtering process, which is implemented by convolving the signal with a filter. The convolution smears the signal, introducing several extra samples into the result. 5.8 Continuous Wavelet Transform (CWT) The continuous wavelet transform was developed as an alternative approach to the short time Fourier transform to overcome the resolution problem. The wavelet analysis is done in a similar way to the STFT analysis, in the sense that the signal is multiplied with a function, similar to the window function in the STFT, and the transform is computed separately for different segments of the time-domain signal. However, there are two main differences between the STFT and the CWT: i. The Fourier transforms of the windowed signals are not taken, and therefore single peak will be seen corresponding to a sinusoid, i.e., negative frequencies are not computed.

ii. The width of the window is changed as the transform is computed for every single

spectral component, which is probably the most significant characteristic of the wavelet transform. The term wavelet means a small wave. The smallness refers to the condition that this (window) function is of finite length (compactly supported). The wave refers to the condition that this function is oscillatory. The term mother implies that the functions with different region of support that are used in the transformation process are derived from one main function, or the mother wavelet. In other words, the mother wavelet is a prototype for generating the other window functions. The term translation is used in the same sense as it was used in the STFT; it is related to the location of the window, as the window is shifted through the signal. This term, obviously, corresponds to time information in the transform domain. However, we do not have a frequency parameter, as we had before for the STFT. 5.9 Discrete Wavelet Transform (DWT) In wavelet analysis, the wavelet function is compared to a section of the signal under study, obtaining a set of coefficients that represent how closely the wavelet function correlates with the signal. Wavelet Transform (WT) 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. 5.9.1 Multilevel Decomposition The decomposition process can be iterated, with successive approximations being decomposed in turn, so that one signal is broken down into many lower-resolution components. A multiresolution analysis (MRA) or multiscale approximation (MSA) of Fig. 5.4 is the design method of most of the practically relevant discrete wavelet transforms (DWT) and the first component to multiresolution analysis is vector spaces.

Fig. 5.4 Wavelet Decomposition Tree

For each vector space, there is another vector space of higher resolution until you get to the final signal. Also, each vector space contains all vector spaces that are of lower resolution. The basis of each of these vector spaces is the scale function for the wavelet and represents the detailed version of the high-frequency components of the signal and the approximation version of the low-frequency components. This is called the wavelet decomposition tree and the output Frequency bands of Wavelet Decomposition shown in Fig. 5.5.

Fig. 5.5 Output Frequency bands of Wavelet Decomposition

The decomposition procedure starts with passing signal (sequence) through a half band digital lowpass filter with impulse response. Filtering a signal corresponds to the mathematical operation of convolution of the signal with the impulse response of the filter. A half band lowpass filter removes all frequencies that are above half of the highest frequency in the signal. The highest frequency component that exists in a signal is p radians, if the signal is sampled at Nyquists rate (which is twice the maximum frequency that exists in the signal). That is, the Nyquists rate corresponds to p rad/s in the discrete frequency domain. After passing the signal through a half band lowpass filter, half of the samples can be eliminated according to the Nyquists rule, since the signal now has a highest frequency of p/2 radians instead of p radians. Simply discarding every other sample will subsample the signal by two, and the signal will then have half the number of points and the scale of the signal is now doubled.

5.9.2 Wavelet Reconstruction The other half is how those decomposed components can be assembled back into the original signal with no loss of information. This process is called reconstruction, or synthesis. The mathematical manipulations that effects synthesis is called the inverse discrete wavelet transform (IDWT).To synthesize a signal; we reconstruct it from the wavelet coefficients as shown in Fig. 5.6.

Fig. 5.6 Wavelet Reconstruction

The filtering part of the reconstruction process is the choice of filters that is crucial in achieving perfect reconstruction of the original signal. Recall that the downsampling of the signal components performed during the decomposition phase introduces a distortion called aliasing. It turns out that by carefully choosing filters for the decomposition and reconstruction phases that are closely related (but not identical), we can cancel out the effects of aliasing. Where wavelet analysis involves filtering and downsampling, the wavelet reconstruction process consists of upsampling and filtering. Extending this technique to the components of a multi-level analysis, we find that similar relationships hold for all the reconstructed signal constituents. That is, a reconstructed signal component of Fig. 5.7 given below is the way to reassemble the original signal.

Fig. 5.7 Reconstructed Signal components

Note that the coefficient vectors A and D because they were produced by downsampling, contain aliasing distortion, and are only half the length of the original signal cannot directly be combined to reproduce the signal. It is necessary to reconstruct the approximations and details before combining them. 5.10 Construction of compactly supported Wavelets Wavelet packet decomposition (WPD) (sometimes known as just wavelet packets) is a wavelet transform where the signal is passed through more filters than the DWT. Although the time and frequency resolution problems are results of a physical phenomenon (the Heisenbergs uncertainty principle) and exist regardless of the transform used, it is possible to analyze any signal by using an alternative approach called Wavelet Packet Transform (WPT). WT analyses the signal at different frequencies with different resolutions. 5.10.1 Wavelet Packets Wavelet packets are the particular linear combination of wavelets. They form bases which retain many of the orthogonality, smoothness, and localization properties of their parent wavelets. The coefficients in the linear combinations are computed by a recursive algorithm

making each newly computed wavelet packet coefficient sequence the root of its own analysis tree. 5.10.2 Wavelet Packet Tree In wavelet analysis, a signal is split into an approximation and a detail. The approximation is then itself split into a second-level approximation and detail, and the process is repeated. For n-level decomposition, there are n+1 possible ways to decompose or encode the signal. In wavelet packet analysis, the details as well as the approximations can be split as shown in Fig. 5.8 and the output frequency bands of Wavelet Packet Decomposition shown in Fig. 5.9. This yields 2n different ways to encode the signal. This is called wavelet packet decomposition tree.

Fig. 5.8 Wavelet Packet Decomposition Tree

Fig. 5.9 Output Frequency bands of Wavelet Packet Decomposition

For instance, wavelet packet analysis allows the signal S to be represented as A1 + AAD3 + DAD3 + DD2. This is an example of a representation that is not possible with ordinary wavelet analysis. 5.11 Summary Wavelets based harmonics elimination is efficient since it estimates harmonics in nonstationary conditions, (i.e.), harmonics varying with respect to time. The conventional method DFT which is not applicable for non-stationary signals so wavelets can be introduced to suppress the harmonics which present in non-stationary signals by analyzing harmonics in certain decomposed frequency bands.