Lecture 13 Wavelet transformation II
Fourier Transform (FT) Forward FT: Inverse FT: Examples: Slide from Alexander Kolesnikov ’s lecture notes
Two test signals: What is difference? x(t)=cos( 1 t)+cos( 2 t)+cos( 3 )+cos( 4 t) x 1 (t)=cos( 1 t) x 2 (t)=cos( 2 t) x 3 (t)=cos( 3 t) x 4 (t)=cos( 4 t) x 1 (t) x 2 (t) x 3 (t) x 4 (t) a) b) 1 = 10 2 = 20 3 = 40 4 =100 Slide from Alexander Kolesnikov ’s lecture notes
Spectrums of the test signals a) b) Signals are different, spectrums are similar Signals are different, spectrums are similar Why? Slide from Alexander Kolesnikov ’s lecture notes
Short-Time Fourier Transform (STFT) Window h(t) Signal in the window Result is localized in space and frequency: Why? Input signal
STFT: Partition of the space-frequency plane
Problems with STFT Uncertainity Principle: Improved space resolution Degraded frequency resolution Improved frequency resolution Degraded space resolution Problem: the same and t throught the entire plane! STFT is redundant representation Not good for compression
Solution: Frequency Scaling Smaller frequency make the window more narrow Bigger frequency make the window wider More narrow time window for higher frequencies here s is scaling factor
New partition of the space-frequency plane Coordinate, t Frequency,
New partition of the plane Discrete wavelet transform Short-time Fourier transform Wavelet functions are localized in space and frequency Hierarchical set of of functions
Frequency vs Time
FT vs WT From one domain to another domain.
Scale and shift Scale Shift
Five steps to calculate WT 1.Take a wavelet and compare it to a section at the start of the original signal. 2.Calculate a number, C, that represents how closely correlated the wavelet is with this section of the signal. 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.
Scale and frequency
Example of Wavelet functions Haar Ingrid Dauhechies
Biorthogonal
Example of Wavelets Coiflets Symlets
Examples of Wavelet functions Morlet Mexican Hat Meyer
Decomposition: approximation and detail One-level decomposition Multi-level decomposition
Haar wavelets
Scaling function and Wavelets Wavelet function: Scaling function : The functions (t) and (t) are orthonormal The most important property of the wavelets: To obtain WT coefficients for level j we can process WT coefficients for level j+1. The most important property of the wavelets: To obtain WT coefficients for level j we can process WT coefficients for level j+1. where
Haar: Scaling function and Wavelets
Daubechies wavelets of order 2 Scaling function Wavelet function
Discrete wavelet transform Wavelets details Low-resolution approx. NB! k j j1j1
Haar wavelet transform
Haar wavelet transform: Example Input data : X={x 1,x 2,x 3,…, x 16 } Haar wavelet transform : (a,b) (s,d) where: 1) scaling function s=(a+b)/2 (smooth, LPF) 2) Haar wavelet d=(a-b) (details, HPF) X={10,13, 11,14, 12,15, 12,14, 12,13, 11,13, 10,11} [11.5,12.5, 13.5,13, 12.5,12, 10.5] {-3, -3, -3, -2, -1,-2,-1} [12, 13.25, 12.25, 10.5] {-2,0.5,-0.5} {-3, -3, -3, -2, -1,-2,-1} [12.625, ] {-1.25, 1.75} {-2,0.5,-0.5} {-3, -3, -3, -2, -1,-2,-1} [12]{1.25} {-1.25, 1.75} {-2,0.5,-0.5} {-3, -3, -3, -2, -1,-2,-1}
Inverse Haar wavelet transform: Example Inverse Haar wavelet transform : (s,d) (a,b) 1) a=s+d/2 2) b=s d/2 Y= [12]{1.25} {-1.25, 1.75} {-2,0.5,-0.5} {-3, -3, -3, -2, -1,-2,-1} [12.625,11.375] {-1.25, 1.75} {-2,0.5,-0.5} {-3, -3, -3, -2, -1,-2,-1} [12, 13.25, 12.25, 10.5] {-2,0.5,-0.5} {-3, -3, -3, -2, -1,-2,-1} [11.5,12.5, 13.5,13, 12.5,12, 10.5] {-3, -3, -3, -2, -1,-2,-1} {10,13, 11,14, 12,15, 12,14, 12,13, 11,13, 10,11} X={10,13, 11,14, 12,15, 12,14, 12,13, 11,13, 10,11} [11.5,12.5, 13.5,13, 12.5,12, 10.5] {-3, -3, -3, -2, -1,-2,-1} [12, 13.25, 12.25, 10.5] {-2,0.5,-0.5} {-3, -3, -3, -2, -1,-2,-1} [12.625, ] {-1.25, 1.75} {-2,0.5,-0.5} {-3, -3, -3, -2, -1,-2,-1} [12]{1.25} {-1.25, 1.75} {-2,0.5,-0.5} {-3, -3, -3, -2, -1,-2,-1}
Wavelet transform as Subband Transform To be continued...
Wavelet Transform and Filter Banks
h 0 (n) is scaling function, low pass filter (LPF) h 1 (n) is wavelet function, high pass filter (HPF) is subsampling (decimation)
Inverse wavelet transform Synthesis filters: g 0 (n)= (-1) n h 1 (n) g 1 (n)= (-1) n h 0 (n) is up-sampling (zeroes inserting)
Wavelet transform as Subband filtering