Download Experimenting with Multi-dimensional Wavelet Transformations

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
Experimenting with Multidimensional Wavelet
Transformations
Tarık Arıcı and Buğra Gedik
Outline of Project Goals
 Writing discrete wavelet transformation and inverse
transformation wrappers (in Matlab) to handle multidimensional data; possible uses include:

2D Images, 3D turbulence data or multi-attribute
sensor readings
 Using wavelets in some example applications
 Lossy compression, De-noising for images, Selfsimilarity analysis
 Studying the phases of the wavelet filters (that delays
the wavelet smoothes) and approximately computing
the delay amount using DSP methods
 Using this on Mammogram reconstruction

Possible uses of Bayesian? (not done)
DWTR / IDWTR wrappers
 Assume D dimensions
 Perform D sweeps, one across
each dimension, making recursive smoothes
calls for each D-1 dimensional slice
 Top level recursive calls go D-1
levels deep before calling the 1
dimensional wavelet transformation
functions
7 detail groups
 As a result 2^D-1 detail groups and
a single smooth group is
constructed for each level of
transformation
smoothes
3 detail groups
Example Applications: Lossy
Compression
Example Applications: De-noising
Example Applications: Self-similarity
Analysis
 Calculate the means of the detail squares for each
level and plot their log as a function of level
 If the line is linear, then there is self-similarity
 Brownian motion is self-similar, Random data (of
course) is not
Mammogram Reconstruction
Original Image
after wavelet
interpolation
after fixing
delay problem
 Assume all details are zero
 Perform inverse wavelet transformation
 Possible use of Bayesian Methods:
 Model missing details using a Bayesian approach
DSP Perspective: Problems Related
with Non-zero Phase Filtering
 Filtering in time domain is multiplication in frequency
domain
X[n]
h[n]
y[n]
 Phase(Y(f)) = Phase(H(f))+Phase(X(f))
Non-zero Phase Filtering
 cos(2pf0t+f) = cos(2pf0(t+f/(2pf0))
= f/(2pf0)
td is constant if f is a linear function of
frequency
 td

 Therefore, wavelet filters should be
(approximately) linear phase filters

Symmetric filters have linear phase

Ex: {1, 1} (Haar), {1, 2, 1}
Least Asymmetric (LA) Wavelet Filters
 Choose filter coefficients:
s.t. min |f(f) – 2pfv|
v=
-L/2+1, if L =8,12,16,20
-L/2, if L =10, 18
-L/2+2, if L =14
 LA(8) and LA(12) works best.
The End!
 Thanks!!!
Related documents