Você está na página 1de 24

ee.sharif.

edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
1
Digital Image Processing
Wavelets and Multi Resolution Processing
1
If you painted a picture with a sky,
clouds, trees, and flowers, you would use
a different size brush depending on the
size of the features. Wavelets are like those
brushes.
Ingrid Daubechies
ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
2
Digital Image Processing
Wavelets and Multi Resolution Processing
2
Image: A non-stationary Phenomenon
ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
3
Digital Image Processing
Wavelets and Multi Resolution Processing
3
Image Pyramid
ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
4
Digital Image Processing
Wavelets and Multi Resolution Processing
4
Gaussian (up) and Laplacian (down) Pyramid
ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
5
Digital Image Processing
Wavelets and Multi Resolution Processing
5
Subband Coding (1D)
ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
6
Digital Image Processing
Wavelets and Multi Resolution Processing
6
Subband Coding (2D)
ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
7
Digital Image Processing
Wavelets and Multi Resolution Processing
7
Four-band Split:
A(LL): Approximation
H(HL): Horizontal
V(LH): Vertical
D (HH): Diagonal
A H
V D
ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
8
Digital Image Processing
Wavelets and Multi Resolution Processing
8
Multi-Level Decomposition:
Haar Basis function
ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
9
Digital Image Processing
Wavelets and Multi Resolution Processing
9
Two Stage FWT
*
Analysis:
*: Fast Wavelet Transform
ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
10
Digital Image Processing
Wavelets and Multi Resolution Processing
10
Two Stage FWT Synthesis:
ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
11
Digital Image Processing
Wavelets and Multi Resolution Processing
11
2D FWT:
Analysis
Decomposition
Synthesis
ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
12
Digital Image Processing
Wavelets and Multi Resolution Processing
12
Three Scale FWT:
Approximation
Horizontal Edge
Vertical Edge
Details
ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
13
Digital Image Processing
Wavelets and Multi Resolution Processing
13
Modifying DWT
*
:
Edge Detection
*: Discrete Wavelet Transform
ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
14
Digital Image Processing
Wavelets and Multi Resolution Processing
14
Modifying DWT
*
:
Noise Reduction, Denoising:
Compute DWT of Noisy Image
Thresholding the details!
Compute IDWT of alerted coefficients
We will discuss more, later.

*: Discrete Wavelet Transform
ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
15
Digital Image Processing
Wavelets and Multi Resolution Processing
15
Wavelet Packet Analysis:
3 scale full analysis three
ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
16
Digital Image Processing
Wavelets and Multi Resolution Processing
16
Example (1):
Full wavelet packet decomposition
ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
17
Digital Image Processing
Wavelets and Multi Resolution Processing
17
Image/Signal Denoising:
Noisy image/signal model:

X(t): Corrupted Signal
S(t): Uncorrupted Signal
N(t): Additive Noise
Wavelet Denoising Scheme



W, W
-1:
Forward and Inverse wavelet transform
D (.,): Thresholding operator ( being the threshold)

( ) ( ) ( )
X t S t N t = +
( )
( )
( )
1
,

Y W X
Z D Y
S W Z

=
=
=
ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
18
Digital Image Processing
Wavelets and Multi Resolution Processing
18
Motivation for Thresholding:
Small coefficients: Dominated by noise.
Large coefficients: Dominated by signal.
Then replacing small coefficients with zero!
Some Assumption:
Wavelet de-correlating property generate a sparse signal.
Noise spreads out equally along all coefficients.
The noise level is NOT too high.

ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
19
Digital Image Processing
Wavelets and Multi Resolution Processing
19
Hard and Soft Thresholding:
Hard:



Soft:

( ) ,
0
u u
D u
u

>

=

s

( ) , 0
u u
D u u
u u



>

= s s

+ <

ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
20
Digital Image Processing
Wavelets and Multi Resolution Processing
20
Threshold Selection:
The most important question.
Very Low threshold: Noisy-Like result
Very High Threshold: Too smooth result.
Several methods proposed:
VisuShrink
SureShrink

ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
21
Digital Image Processing
Wavelets and Multi Resolution Processing
21
VisuShrink (Universal Thresholding):

N: Sample (Signal/Image) size (# of pixels in image)
: Noise variance
Thresholding sub-band:
All
Details (HL,LH, HH)
Noise Estimation:


( )
2
N
Ln N o =
2
N
o
( )
2
2
1
,
0.6745
ij
N ij
median Y
Y HH o
(
(
= e
(

ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
22
Digital Image Processing
Wavelets and Multi Resolution Processing
22
SureShrink (Adaptive Thresholding):
Sub-band adaptive thresholding (each detail sub-band)
Based on Steins Unbiased Estimator for Risk (SURE), a
method for estimating the loss in an unbiased fashion.
:

Wavelet coefficients in the j
th
sub-band
For the soft threshold estimator:

We have:

Optimal threshold:



{ }
1
d
i
i
Y
=
( )
,
i i
Y D Y =
{ } ( )
2
1
( ; ) 2# : min ,
d
i i
i
SURE Y d i Y Y
=
= s +

*
arg min ( ; ) SURE Y

=
ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
23
Digital Image Processing
Wavelets and Multi Resolution Processing
23
NormalShrink:
For maximum J scale and for scale k:

( )
2
2
2
1
2

,
0.6745
: Standard deviation of the considered subband
: Length of subband k
N
k
Y
k
ij
N ij
Y
L
Ln
J
median Y
Y HH
L
k
o
|
o
|
o
o
=
| |
=
|
\ .
(
(
= e
(

ee.sharif.edu/~dip
E. Fatemizadeh, Sharif University of Technology, 2011
24
Digital Image Processing
Wavelets and Multi Resolution Processing
24
Challenges:
Wavelet base.
Threshold Selection.
Threshold function.

Você também pode gostar