annotated-chap-3-gw.ppt

Lecture 2. Intensity Transformation
and Spatial Filtering

4/3/2023 2
Spatial Domain vs. Transform Domain
► Spatial domain
image plane itself, directly process the intensity values of
the image plane
► Transform domain
process the transform coefficients, not directly process the
intensity values of the image plane

4/3/2023 3
Spatial Domain Process
( , ) [ ( , )])
( , ) :input image
( , ) :output image
:an operator on defined over
a neighborhood of point ( , )
g x y T f x y
f x y
g x y
T f
x y


4/3/2023 4

4/3/2023 5
Intensity transformation function
( )
s T r


4/3/2023 6
Some Basic Intensity Transformation
Functions

4/3/2023 7
Image Negatives
Image negatives
1
s L r
  

4/3/2023 8
Example: Image Negatives
Small
lesion

4/3/2023 9
Log Transformations
Log Transformations
log(1 )
s c r
 

4/3/2023 10
Example: Log Transformations

4/3/2023 11
Power-Law (Gamma) Transformations
s cr


4/3/2023 12
Example: Gamma Transformations

4/3/2023 13
Cathode ray tube
(CRT) devices have an
intensity-to-voltage
response that is a
power function, with
exponents varying
from approximately
1.8 to 2.5
1/2.5
s r


4/3/2023 14

4/3/2023 15

4/3/2023 16
Piecewise-Linear Transformations
► Contrast Stretching
— Expands the range of intensity levels in an image so that it spans
the full intensity range of the recording medium or display device.
► Intensity-level Slicing
— Highlighting a specific range of intensities in an image often is of
interest.

4/3/2023 18
Highlight the major
blood vessels and
study the shape of the
flow of the contrast
medium (to detect
blockages, etc.)
Measuring the actual
flow of the contrast
medium as a function
of time in a series of
images

4/3/2023 22
Histogram Processing
► Histogram Equalization
► Histogram Matching
► Local Histogram Processing
► Using Histogram Statistics for Image Enhancement

4/3/2023 23
Histogram Processing
Histogram ( )
is the intensity value
is the number of pixels in the image with intensity
k k
th
k
k k
h r n
r k
n r

Normalized histogram ( )
: the number of pixels in the image of
size M N with intensity
k
k
k
k
n
p r
MN
n
r



4/3/2023 25
Histogram Equalization
The intensity levels in an image may be viewed as
random variables in the interval [0, L-1].
Let ( ) and ( ) denote the probability density
function (PDF) of random variables and .
r s
p r p s
r s

4/3/2023 26
( ) 0 1
s T r r L
   
. T(r) is a strictly monotonically increasing function
in the interval 0 -1;
. 0 ( ) -1 for 0 -1.
a
r L
b T r L r L
 
   

4/3/2023 27
. T(r) is a strictly monotonically increasing function
in the interval 0 -1;
. 0 ( ) -1 for 0 -1.
a
r L
b T r L r L
 
   
( ) 0 1
s T r r L
   
( ) is continuous and differentiable.
T r
( ) ( )
s r
p s ds p r dr


4/3/2023 28
0
( ) ( 1) ( )
r
r
s T r L p w dw
   
0
( )
( 1) ( )
r
r
ds dT r d
L p w dw
dr dr dr
 
  
 
 

( 1) ( )
r
L p r
 
 
( ) 1
( ) ( )
( )
( 1) ( ) 1
r r r
s
r
p r dr p r p r
p s
L p r
ds
ds L
dr
   

  
 
 

4/3/2023 29
Example
2
Suppose that the (continuous) intensity values
in an image have the PDF
2
, for 0 r L-1
( 1)
( )
0, otherwise
Find the transformation function for equalizing
the image histogra
r
r
L
p r

 


 


m.

4/3/2023 30
Example
0
( ) ( 1) ( )
r
r
s T r L p w dw
   
2
0
2
( 1)
( 1)
r w
L dw
L
 


2
1
r
L



4/3/2023 31
0
Continuous case:
( ) ( 1) ( )
r
r
s T r L p w dw
   
0
Discrete values:
( ) ( 1) ( )
k
k k r j
j
s T r L p r

   
0 0
1
( 1) k=0,1,..., L-1
k k
j
j
j j
n L
L n
MN MN
 

  
 

4/3/2023 32
Example: Histogram Equalization
Suppose that a 3-bit image (L=8) of size 64 × 64 pixels (MN = 4096)
has the intensity distribution shown in following table.
Get the histogram equalization transformation function and give the
ps(sk) for each sk.

4/3/2023 33
0
0 0
0
( ) 7 ( ) 7 0.19 1.33
r j
j
s T r p r

    
 1

1
1 1
0
( ) 7 ( ) 7 (0.19 0.25) 3.08
r j
j
s T r p r

     
 3

2 3
4 5
6 7
4.55 5 5.67 6
6.23 6 6.65 7
6.86 7 7.00 7
s s
s s
s s
   
   
   

4/3/2023 34

Figure 3.22
(a) Image from Phoenix Lander. (b) Result of histogram
equalization. (c) Histogram of image (a). (d) Histogram of image
(b). (Original image courtesy of NASA.)

4/3/2023 38
Question
Is histogram equalization always good?
No

4/3/2023 39
Histogram Matching
Histogram matching (histogram specification)
— generate a processed image that has a specified histogram
Let ( ) and ( ) denote the continous probability
density functions of the variables and . ( ) is the
specified probability density function.
Let be the random variable with the prob
r z
z
p r p z
r z p z
s
0
0
ability
( ) ( 1) ( )
Define a random variable with the probability
( ) ( 1) ( )
r
r
z
z
s T r L p w dw
z
G z L p t dt s
  
  



4/3/2023 40
Histogram Matching
0
0
( ) ( 1) ( )
( ) ( 1) ( )
r
r
z
z
s T r L p w dw
G z L p t dt s
  
  


 
1 1
( ) ( )
z G s G T r
 
 

4/3/2023 41
Histogram Matching: Procedure
► Obtain pr(r) from the input image and then obtain the values of s
► Use the specified PDF and obtain the transformation function G(z)
► Mapping from s to z
0
( 1) ( )
r
r
s L p w dw
  
0
( ) ( 1) ( )
z
z
G z L p t dt s
  

1
( )
z G s



4/3/2023 42
Histogram Matching: Example
Assuming continuous intensity values, suppose that an image has
the intensity PDF
Find the transformation function that will produce an image
whose intensity PDF is
2
2
, for 0 -1
( 1)
( )
0 , otherwise
r
r
r L
L
p r

 


 


2
3
3
, for 0 ( -1)
( ) ( 1)
0, otherwise
z
z
z L
p z L

 

 




4/3/2023 43
Histogram Matching: Example
Find the histogram equalization transformation for the input image
Find the histogram equalization transformation for the specified histogram
The transformation function
2
0 0
2
( ) ( 1) ( ) ( 1)
( 1)
r r
r
w
s T r L p w dw L dw
L
    

 
2 3
3 2
0 0
3
( ) ( 1) ( ) ( 1)
( 1) ( 1)
z z
z
t z
G z L p t dt L dt s
L L
     
 
 
2
1
r
L


1/3
2
1/3 1/3
2 2 2
( 1) ( 1) ( 1)
1
r
z L s L L r
L
 
   
     
 
   

 

4/3/2023 44
Histogram Matching: Discrete Cases
► Obtain pr(rj) from the input image and then obtain the values of
sk, round the value to the integer range [0, L-1].
► Use the specified PDF and obtain the transformation function
G(zq), round the value to the integer range [0, L-1].
► Mapping from sk to zq
0 0
( 1)
( ) ( 1) ( )
k k
k k r j j
j j
L
s T r L p r n
MN
 

   
 
0
( ) ( 1) ( )
q
q z i k
i
G z L p z s

  

1
( )
q k
z G s



4/3/2023 45
Example: Histogram Matching
Suppose that a 3-bit image (L=8) of size 64 × 64 pixels (MN = 4096)
has the intensity distribution shown in the following table (on the
left). Get the histogram transformation function and make the output
image with the specified histogram, listed in the table on the right.

4/3/2023 46
Obtain the scaled histogram-equalized values,
Compute all the values of the transformation function G,
0 1 2 3 4
5 6 7
1, 3, 5, 6, 7,
7, 7, 7.
s s s s s
s s s
    
  
0
0
0
( ) 7 ( ) 0.00
z j
j
G z p z

 

1 2
3 4
5 6
7
( ) 0.00 ( ) 0.00
( ) 1.05 ( ) 2.45
( ) 4.55 ( ) 5.95
( ) 7.00
G z G z
G z G z
G z G z
G z
 
 
 

0

0
 0

1
 2

6

5

7


4/3/2023 47

4/3/2023 48
Obtain the scaled histogram-equalized values,
Compute all the values of the transformation function G,
0 1 2 3 4
5 6 7
1, 3, 5, 6, 7,
7, 7, 7.
s s s s s
s s s
    
  
0
0
0
( ) 7 ( ) 0.00
z j
j
G z p z

 

1 2
3 4
5 6
7
( ) 0.00 ( ) 0.00
( ) 1.05 ( ) 2.45
( ) 4.55 ( ) 5.95
( ) 7.00
G z G z
G z G z
G z G z
G z
 
 
 

0

0
 0

1
 2

6

5

7

s0
s2 s3
s5 s6 s7
s1
s4

4/3/2023 49
0 1 2 3 4
5 6 7
1, 3, 5, 6, 7,
7, 7, 7.
s s s s s
s s s
    
  
0
1
2
3
4
5
6
7
k
r

4/3/2023 50
0 3
1 4
2 5
3 6
4 7
5 7
6 7
7 7
k q
r z










4/3/2023 51

4/3/2023 52

4/3/2023 53

4/3/2023 54

Figure 3.24
(a) An image, and (b) its histogram.

Figure 3.25
(a) Histogram equalization transformation obtained using the histogram
in Fig. 3.24(b). (b) Histogram equalized image. (c) Histogram of
equalized image.

Figure 3.26
Histogram specification. (a) Specified histogram. (b) Transformation  
q
G z ,
labeled (1),
and
 
1
k
G s ,

labeled (2). (c) Result of histogram specification. (d)
Histogram of image (c).

4/3/2023 58
Local Histogram Processing
Define a neighborhood and move its center from pixel to
pixel
At each location, the histogram of the points in the
neighborhood is computed. Either histogram equalization or
histogram specification transformation function is obtained
Map the intensity of the pixel centered in the neighborhood
Move to the next location and repeat the procedure

4/3/2023 59
Local Histogram Processing: Example

Figure 3.33
(a) Original image. (b) Result of local enhancement based
on local histogram statistics. Compare (b) with Fig. 3.32(c).

4/3/2023 61
Using Histogram Statistics for Image
Enhancement
1
0
( )
L
i i
i
m r p r


 
1
0
( ) ( ) ( )
L
n
n i i
i
u r r m p r


 

1
2 2
2
0
( ) ( ) ( )
L
i i
i
u r r m p r



  

1 1
0 0
1
( , )
M N
x y
f x y
MN
 
 
 
 
1 1
2
0 0
1
( , )
M N
x y
f x y m
MN
 
 
 

Average Intensity
Variance

4/3/2023 62
Enhancement
1
0
Local average intensity
( )
denotes a neighborhood
xy xy
L
s i s i
i
xy
m r p r
s


 
1
2 2
0
Local variance
( ) ( )
xy xy xy
L
s i s s i
i
r m p r



 


4/3/2023 63
Enhancement: Example
0 1 2
0 1 2
( , ), if and
( , )
( , ), otherwise
:global mean; :global standard deviation
0.4; 0.02; 0.4; 4
xy xy
s G G s G
G G
E f x y m k m k k
g x y
f x y
m
k k k E
  

  


 


   

4/3/2023 64
Spatial Filtering
A spatial filter consists of (a) a neighborhood, and (b) a
predefined operation
Linear spatial filtering of an image of size MxN with a filter
of size mxn is given by the expression
( , ) ( , ) ( , )
a b
s a t b
g x y w s t f x s y t
 
  
 

4/3/2023 66
Spatial Correlation
The correlation of a filter ( , ) of size
with an image ( , ), denoted as ( , ) ( , )
w x y m n
f x y w x y f x y

( , ) ( , ) ( , ) ( , )
a b
s a t b
w x y f x y w s t f x s y t
 
  
 

4/3/2023 67
Spatial Convolution
The convolution of a filter ( , ) of size
with an image ( , ), denoted as ( , ) ( , )
w x y m n
f x y w x y f x y

( , ) ( , ) ( , ) ( , )
a b
s a t b
w x y f x y w s t f x s y t
 
  
 

Figure 3.35
Illustration of 1-D correlation and convolution of a kernel, w, with a function f consisting
of a discrete unit impulse. Note that correlation and convolution are functions of the
variable x, which acts to displace one function with respect to the other. For the
extended correlation and convolution results, the starting configuration places the
rightmost element of the kernel to be coincident with the origin of f. Additional padding
must be used.

Figure 3.58
Transfer functions of ideal 1-D filters in the frequency
domain (u denotes frequency). (a) Lowpass filter. (b)
Highpass filter. (c) Bandreject filter. (d) Bandpass filter. (As
before, we show only positive frequencies for simplicity.)

Table 3.7
Summary of the four principal spatial filter types expressed
in terms of lowpass filters. The centers of the unit impulse
and the filter kernels coincide.

4/3/2023 72
Smoothing Spatial Filters
Smoothing filters are used for blurring and for noise
reduction
Blurring is used in removal of small details and bridging of
small gaps in lines or curves
Smoothing spatial filters include linear filters and nonlinear
filters.

4/3/2023 73
Spatial Smoothing Linear Filters
The general implementation for filtering an M N image
with a weighted averaging filter of size m n is given
( , ) ( , )
( , )
( , )
where 2 1
a b
s a t b
a b
s a t b
w s t f x s y t
g x y
w s t
m a
 
 


 

 
 
 
, 2 1.
n b
 

4/3/2023 74
Two Smoothing Averaging Filter Masks

4/3/2023 76
Example: Gross Representation of Objects

4/3/2023 77
Order-statistic (Nonlinear) Filters
— Nonlinear
— Based on ordering (ranking) the pixels contained in the
filter mask
— Replacing the value of the center pixel with the value
determined by the ranking result
E.g., median filter, max filter, min filter

4/3/2023 78
Example: Use of Median Filtering for Noise Reduction

4/3/2023 79
Sharpening Spatial Filters
► Foundation
► Laplacian Operator
► Unsharp Masking and Highboost Filtering
► Using First-Order Derivatives for Nonlinear Image
Sharpening — The Gradient

4/3/2023 80
Sharpening Spatial Filters: Foundation
► The first-order derivative of a one-dimensional function f(x)
is the difference
► The second-order derivative of f(x) as the difference
( 1) ( )
f
f x f x
x

  

2
2
( 1) ( 1) 2 ( )
f
f x f x f x
x

    


4/3/2023 82
Sharpening Spatial Filters: Laplace Operator
The second-order isotropic derivative operator is the
Laplacian for a function (image) f(x,y)
2 2
2
2 2
f f
f
x y
 
  
 
2
2
( 1, ) ( 1, ) 2 ( , )
f
f x y f x y f x y
x

    

2
2
( , 1) ( , 1) 2 ( , )
f
f x y f x y f x y
y

    

2
( 1, ) ( 1, ) ( , 1) ( , 1)
-4 ( , )
f f x y f x y f x y f x y
f x y
        

4/3/2023 83

4/3/2023 84
Image sharpening in the way of using the Laplacian:
2
2
( , ) ( , ) ( , )
where,
( , ) is input image,
( , ) is sharpenend images,
-1 if ( , ) corresponding to Fig. 3.37(a) or (b)
and 1 if either of the other two filters is us
g x y f x y c f x y
f x y
g x y
c f x y
c
 
  
 
 
 ed.

4/3/2023 86
Unsharp Masking and Highboost Filtering
► Unsharp masking
Sharpen images consists of subtracting an unsharp (smoothed)
version of an image from the original image
e.g., printing and publishing industry
► Steps
1. Blur the original image
2. Subtract the blurred image from the original
3. Add the mask to the original

4/3/2023 87
Unsharp Masking and Highboost Filtering
Let ( , ) denote the blurred image, unsharp masking is
( , ) ( , ) ( , )
Then add a weighted portion of the mask back to the original
( , ) ( , ) * ( , )
mask
mask
f x y
g x y f x y f x y
g x y f x y k g x y
 
  0
k 
when 1, the process is referred to as highboost filtering.
k 

4/3/2023 88
Unsharp Masking: Demo

Figure 3.55
(a) Unretouched “soft-tone” digital image of size
469×600pixels
(b) Image blurred using
a
31×31 Gaussian lowpass filter with σ =
5.
(c) Mask. (d) Result of unsharp masking using Eq. (3-65) with k = 1.
(e) and (f) Results of highboost filtering with k = 2 and k = 3,
respectively.

4/3/2023 90
Unsharp Masking and Highboost Filtering: Example

4/3/2023 91
Image Sharpening based on First-Order Derivatives
For function ( , ), the gradient of at coordinates ( , )
is defined as
grad( )
x
y
f x y f x y
f
g x
f f
f
g
y

 
 
  
 
   
  
 
 
 

 
2 2
The of vector , denoted as ( , )
( , ) mag( ) x y
magnitude f M x y
M x y f g g

   
Gradient Image

4/3/2023 92
2 2
The of vector , denoted as ( , )
( , ) mag( ) x y
magnitude f M x y
M x y f g g

   
( , ) | | | |
x y
M x y g g
 
z1 z2 z3
z4 z5 z6
z7 z8 z9
8 5 6 5
( , ) | | | |
M x y z z z z
   

4/3/2023 93
z1 z2 z3
z4 z5 z6
z7 z8 z9
9 5 8 6
Roberts Cross-gradient Operators
( , ) | | | |
M x y z z z z
   
7 8 9 1 2 3
3 6 9 1 4 7
Sobel Operators
( , ) | ( 2 ) ( 2 ) |
| ( 2 ) ( 2 ) |
M x y z z z z z z
z z z z z z
     
     

4/3/2023 94

4/3/2023 96
Example:
Combining
Spatial
Enhancement
Methods
Goal:
Enhance the
image by
sharpening it
and by bringing
out more of the
skeletal detail

4/3/2023 97
Example:
Combining
Spatial
Enhancement
Methods
Goal:
Enhance the
image by
sharpening it
and by bringing
out more of the
skeletal detail

annotated-chap-3-gw.ppt

More Related Content

Similar to annotated-chap-3-gw.ppt (20)

Recently uploaded (20)

annotated-chap-3-gw.ppt