Presentation is loading. Please wait.

Presentation is loading. Please wait.

Introduction To Wavelets

Similar presentations


Presentation on theme: "Introduction To Wavelets"— Presentation transcript:

1 Introduction To Wavelets
Transforming Problem Representation

2 Why Representation Is Key?
The curse of dimensionality Dealing with distractors Signal enhancement goals such as increasing low contrast Feature identification Image 1 Gene A. Tagliarini 12/2/2018

3 Sobel Filtering of Raw Image Data
Sobel filtered raw data image Note speckling Median filters can reduce speckle (this image is not median filtered) Image 2 Gene A. Tagliarini 12/2/2018

4 Another Spatial Representation
One level of a two-dimensional wavelet transform Image resized for comparison Contrast is enhanced Image 3 Gene A. Tagliarini 12/2/2018

5 Sobel Filtering Of Transformed Image
Source (Image 3) is ¼ the size of Images 1 and 2 Blocks arise from rescaling Image 4 Gene A. Tagliarini 12/2/2018

6 Sobel Filtering Of Raw Image Data
Sobel filtered raw data image Note speckling Median filters can reduce speckle (this image is not median filtered) Image 2 Gene A. Tagliarini 12/2/2018

7 One-dimensional Signal
A clean signal Short duration (circa 200 mSec) Typical of sonar transients Gene A. Tagliarini 12/2/2018

8 Noisy One-dimensional Signal
The transient occurs slightly after the center time displayed Detection-determining that a transient occurred Classification-determining which transient occurred Gene A. Tagliarini 12/2/2018

9 Common Wavelet Applications
Requiring only decomposition Signal processing Image filtering Feature extraction Requiring decomposition and reconstruction Data transmission using compression Matrix multiplication Gene A. Tagliarini 12/2/2018

10 Wavelet Transform Properties
Analogous to Fourier Transforms—but different Constant ratio of scale versus constant difference of frequencies Open choice of basis functions versus fixed choice of basis functions (sines and cosines) Compact support versus non-compact support Provides a decomposition for square integrable Localize in time and scale versus time and frequency Readily computable Gene A. Tagliarini 12/2/2018

11 The Basic Dilation Equation
Two simple solutions: c0 = 2, implies f(x) = d(x) c0=c1=1, implies f(x) = f(2x) + f(2x-1) and f(x) = c([0,1]) Gene A. Tagliarini 12/2/2018

12 A Graphical Example of Dilation
f(x) = c([0,1)) = f(2x) + f(2x-1) 1 0.5 Gene A. Tagliarini 12/2/2018

13 The Basic Wavelet Equation
Uses differences and the scaling function The Haar wavelet (based on the box function) is given by W(x)= f(2x) - f(2x-1) where f(x) = c([0,1)) Gene A. Tagliarini 12/2/2018

14 The Box Function And The Basic Haar Wavelet
f(2x) + f(2x-1) f(2x) - f(2x-1) Haar Wavelet Box Function Gene A. Tagliarini 12/2/2018

15 A Normalized Dilation Equation
Gene A. Tagliarini 12/2/2018

16 The Goal Of The Transformation Process
Write one function (the signal S) as a linear combination of the scaling and wavelet functions where the wj,k are the wavelet scalars and Wj,k(x) are translated and dilated wavelets Gene A. Tagliarini 12/2/2018

17 Matrix Representation Of The Haar Wavelet Transform
Gene A. Tagliarini 12/2/2018

18 Inverting The Effects Of The Haar Wavelet Transform
Gene A. Tagliarini 12/2/2018

19 The Inverse Matrix Is A Scaled Version Of The Transpose
Gene A. Tagliarini 12/2/2018

20 A Sample Signal And Its Decomposition (One Level)
Gene A. Tagliarini 12/2/2018

21 WT Computation: Low-pass Filter Output And Down-sampling
lp0 c0 c1 c2 c3 lp1 c0 c1 c2 c3 lp2 c0 c1 c2 c3 lp3 s0 s1 s2 s3 s4 s5 s6 s7 s8 s9 Gene A. Tagliarini 12/2/2018

22 WT Computation: High-pass Filter Output And Down-sampling
s0 s1 s2 s3 s4 s5 s6 s7 s8 s9 Gene A. Tagliarini 12/2/2018

23 For Two-Dimensional Signals (Using Separable Wavelets)
LL LH LP HP Original HL HH Process each row, storing LP results on the left and HP results on the right Process each column of the previous, storing LP results at the top and HP at the bottom One level of a 2-D transform Gene A. Tagliarini 12/2/2018

24 A Two-level Two-dimensional Example
Images from the TRIM-2 data set acquired from the Night Vision and Electronic Sensors Directorate, Night Vision Laboratories, Ft. Belvoir Images are in ARF format using 480x640 pixels, each byte representing one of 256 possible gray-scale levels Terrain board images simulate scenes viewed in infrared Gene A. Tagliarini 12/2/2018

25 So, Where Is The Compression?
Suppose the signal is represented with N samples Low-pass filtering produces n/2 values High-pass filtering produces n/2 values Compression arises from quantization and encoding of HP filter output Gene A. Tagliarini 12/2/2018

26 How Can One Generate Wavelet Filter Coefficients?
Rigorous mathematical analysis Daubechies coefficients Exploit parameterizations TeKolste Pollen C0 = [(1 + cos a + sin a)(1 – cos b – sin b) + 2 sin b cos a]/4 C1 = [(1 - cos a + sin a)(1 + cos b – sin b) - 2 sin b cos a]/4 C2 = [1 + cos(a – b) + sin(a – b)]/2 C3 = [1 + cos(a – b) - sin(a – b)]/2 C4 = 1 – c2 – c0 C5 = 1 – c3 – c1 Gene A. Tagliarini 12/2/2018

27 What Might One Ask Using The Pollen Parameterization?
What wavelet provides the best basis for compression? Can one minimize variability or magnitudes in the high-pass filter output? What wavelet might emerge using a small set of samples from a piece-wise linear function? 32 samples Corresponds to image regions having smooth gradients or homogeneous contents Gene A. Tagliarini 12/2/2018

28 Resulting Coefficients
Proc. SPIE, Vol. 2762, p. 89 Gene A. Tagliarini 12/2/2018

29 Concluding Comments Wavelets provide an approach to transforming problems from the time domain to a scale domain The transform can be tailored to meet processing objectives Optimization techniques can lead to results that correspond to (possibly difficult to obtain) analytical results There is much more to say about things like: Orthogonality, bi-orthogonality, and orthonormality Separability and nonseparability Super-wavelets Gene A. Tagliarini 12/2/2018


Download ppt "Introduction To Wavelets"

Similar presentations


Ads by Google